Old Oud - New Project

Thanks patheslip. The angle of the tear drop - for this preliminary design - has been determined 'by eye' but referencing the angle to another part of the oud - tail end of bowl, bridge position, neck joint etc. is a good suggestion worth investigating, just to see how it turns out. This would also apply if the upper surface of the tear drop is made curved rather than flat. All a bit of guesswork at this stage.

Damp, humid weather but bowl levelling can go ahead regardless. I use a block plane to bring the edge of the bowl quickly to the required level - frequently placing the bowl, inverted, on my flat MDF work surface to check for high/low spots. To prevent the bowl slipping during this procedure, I use a cheap, rubber drawer liner, placed on my lap. For final levelling, I use a flat board - with fine grit sandpaper glued to one end - to ensure that the angle around the edge of the bowl is perfectly correct.

Due to the slight rotation of the end block, previously reported, the bowl will be levelled to about 3 mm higher than design at the tail end of the bowl.

According to Robert Lundberg "Historical Lute Construction" - some surviving lute bowls of the 16th/17th C were not levelled perfectly flat but seem to have been slightly 'scooped out' below the sound hole position (providing tension to the soundboard in this area perhaps?). However, there is some question as to whether this observed 'dip' in the sound board might have been due to later repairs and consequent bowl distortions over the centuries.
Interestingly, Richard Hankey (DR Oud) in his book "The Oud - Construction and Repair" also observes this feature in the surviving Syrian, Nahat ouds of the early 20th C.
At this point, I am not sure if I will leave the bowl edge perfectly flat or slightly 'scooped'.

While still speculating about early oud pegbox geometry, I looked at the pegbox geometry of an old (early 20th C?) Egyptian oud, having the now conventional, modern 'S' shaped profile - just for comparison.

The attached image is a full size tracing of the peg box profile of the Egyptian oud. It may not be typical of oud pegbox design of the period but demonstrates a potential weakness in the design geometry.
For convenience, I have numbered the pegs in sequence from nut to tail.
It can be seen that with this configuration, all strings lower than peg #4 'ride' either on this peg or peg #5. This is a situation that may affect the accurate tuning of strings attached to pegs #5 to #12.

However, by adjustment to the upper and lower curvature of radius of the pegbox it is possible to arrive at a geometry where the strings do not (quite) come into contact with adjacent pegs and where all strings are contained within the peg box profile. Perhaps this might explain why the modern style peg boxes are 'S' shaped?

So, perhaps, all early oud 'sickle' profile peg boxes must have looked like example 'C' previously posted (or more extreme even with greater radius of curvature implied by the early Persian miniatures?). If so, this would have made access to the peg box (for string changes) more of a challenge (but not impossible).
More to think about!

Hi JD

My Tasos oud has a bowl whose edge is 'scooped' out, ie the curve in the sound board is not simply due to the strings pulling the bridge forward but extends right to the edges of the sound board.

Whether and how this effects the sound I don't know (logic would tell me it would as it would effect the direction of the many waves and ripples that a fixed bridge soundboard creates when the strings are activated) ... but it does sound pretty amazing !!

Great thread



Leon

The bowl is now levelled and the neck is planed down to match. Some adjustment to this 'level playing field' - 'scooping' the bowl edge and neck 'set back' etc. - will be made later from this datum.
Neck centre line alignment has been determined using a 'centre-finder' ruler taped across the maximum width of the bowl and a straight edge.
The next stage will be to trim and fit the sound board. This will - in turn - determine the final bridge position and neck length.
A preliminary check of dimensions indicates that the string length will be just over 56 cm.

Although hot , humid conditions prevent any component assembly, there is plenty of other work that can continue.

The sound board braces have been shaped to a preliminary 'wedge' shaped section - replicating the brace geometry on my old Egyptian oud. All braces are of equal height (15 mm), thickness (4 mm) and taper, at the ends, over a distance of about 60 mm to a depth of about 9 mm at the edge of the bowl.

To shape the brace sections, I used the very nice little rosewood plane available from Lee Valley ( Cat#07P15.01) for under $20 Can. (prices have recently been increased so am not sure of the current cost. I paid $16.95 Can.). This inexpensive, well made, tool is perfect for the job - the sole is just the right width for planing braces of 15 mm depth - the wooden body does not scar the surface of the sound board (as a metal plane would) and the plane may be used by either 'pushing' or 'pulling' - dependent upon grain direction.
When planing over the sound holes, the bracing of the rosettes was protected from potential damage with a strip of
1 mm thick wood. A strip of thick card would do just as well.

To taper the ends of the braces, a metal template was made of the required profile and traced onto the rib with a pencil. The little block plane was then used to trim each brace end to the pencil line. The plane cuts quickly and smoothly without risk of grain 'tear out'.

With preliminary planing and shaping of the braces completed, the ends of the braces must be trimmed in preparation for fitting the sound board to bowl. At this point the sound board is about
12 mm oversize.
The bowl was inverted over the braces, aligned and the brace positions marked in pencil on the side of the bowl. The angle of each brace end was determined (approximately) using a small bevel gauge. This angle was transferred to a piece of stiff card and the angle marked on the end of the brace in pencil using the card as a template.
The brace ends will be cut a little oversize to allow for final fitting.

The brace ends are then trimmed using a fine toothed razor saw (smooth cutting across the grain) and the waste then cut and sliced away with sharp paring chisel.
Final fit is done by shaving and shaping the ends with a sharp knife and file. A close fit to the bowl is necessary to achieve optimum strength of the sound board to bowl assembly.

The saw used here is another good quality low cost tool from Lee Valley (also available from luthier supply companies). The Lee Valley saw has 53 teeth per inch with 0.010 inch kerf - stiff backed and cuts on the draw, cat# 60F03.10, cost $6.50 Can.

Final fitting of soundboard to bowl.

Two support blocks, 10 mm thick, have been temporarily attached to both ends of the sound board with double-sided adhesive tape. The blocks raise the edge of the bowl slightly above the braces so that the ends of the braces can be observed and marked to the required length and correct angle (the angle of the outside ribs - and hence the brace ends - vary along their length - just to complicate things!)

The sound board is placed upon a flat working surface, and, with everything in correct alignment, the bowl is temporarily attached to the support blocks (using 'masking' tape) and the brace ends marked in pencil ready for trimming.

The final fitting of the sound board to bowl is an important and exacting process requiring careful, step by step trimming/fitting of the end of each brace - a process of trial and error - not to be rushed.

The tools used to trim the brace ends must be razor sharp and carefully controlled to avoid damage to the sound board surface. A good, low cost, tool for the job is a commercial, single edged razor blade (cost about $10 for a hundred!). The sound board surface may be protected against accidental cuts with a piece of veneer but it is better to tape the blade so that only a small section does the cutting (safer to handle as well). The blade is held in both hands - for precise control - and used with a slicing cut - removing small amounts at a time.

The amount of material to be removed from each brace is a matter of judgment - sight, sound and feel all playing a part - until the bowl fits the face of the sound board - with everything in perfect alignment and without any forcing - and all bar ends are felt to be in firm contact with the side of the bowl.

At this stage, the surplus material around the edge of the sound board can be trimmed away - leaving about 3 mm surplus for finishing later. I used a fret saw and fine jewellers saw blade to trim the waste.

Now that the sound board has been fitted, it is possible to determine the exact position of the bridge and hence the length of the neck (nut to neck joint = 1/3 string length, neck joint to bridge = 2/3 string length, according to the original, speculative design geometry). From this geometry, the string length is 56.1 mm - close enough!

At this stage - due to current high relative humidity levels (70%) - gluing the bridge to sound board will be postponed, but work can now continue in final trimming and shaping of the neck.

With the sound board fitted, the exact bridge location determined and the neck centre line established - a 'string template' has been made to verify adequate width of the neck at the neck joint. This template will also be used later to set the bridge alignment prior to gluing.

The template represents the outside area covered by five double courses of strings - measured to the outside of the strings at the nut and bridge - with a string length of 56.1 cm. The string spacing at the nut is the spacing that I use for my lutes (33 mm overall). The spacing at the bridge is also one that I use for my lutes which happens to coincide exactly to the string spacing of my old Egyptian oud (65 mm).
The template has been made from thin tinplate (because I have the facilities to accurately cut this material) but I could just as easily have made the template from thick cardboard cut with a sharp knife and metal straight edge.

Checking the neck layout with the template confirms that there will be about a 3.5 mm space - measured from the outside edges of the treble and bass strings, to the edge of the fingerboard - at the neck joint. This should be satisfactory.

The next step will be to trim the neck to the required profile and cross section.

The neck blank has been planed back to allow adjustment to string 'action' at a later date. This has eliminated the extra wedge of material added to the top surface of the the neck blank (as a precaution). The fingerboard will, therefore, taper from about 1.5 mm at the neck joint to about 3 mm at the nut.

Having established the neck centre line and profile, the neck blank has been cut to the plan and elevation profile on a band saw.
As the neck blank is now tapered in plan and elevation, it must be temporarily mounted (with wood screws) to a pine block so that it may be securely held in a vice in order to shape the required cross section - which will be semicircular like the bowl section.

Rough shaping is done with a paring chisel and fine shaping with a block plane. Final finishing work will be done with a file.

The neck is made from Sitka Spruce , so will be veneered with harder Ash wood (but also to be consistent with the wood of the bowl - a cosmetic consideration).The section of the neck must, therefore, be shaped about 1.5 mm undersized. As a guide, metal templates of the required section at the neck joint and nut were fabricated in order to trace, in pencil, the final neck section.

Now that the peg box geometry has been defined, a template of the sides has been made in thin metal.
The sides will be a laminate of core material with a veneer of Ash wood.
The Ash wood veneer is just for appearances - to match the the rest of the oud. A piece of Ash wood with curved grain - approximately matching the curve of the pegbox sides - has been cut into veneer about 2-3 mm thick for this application.

For the inside core of the peg box, two alternatives are under consideration. Beech and Elm wood.
Recent research by ALAMI has indicated that Hackberry - a wood very close to Elm in its cell structure - may have been a candidate for a wood used in early ouds (See 'Wood Fit for a King' on the Oud Maintenance forum) - although Elm wood (according to Farmer's translation) - may now be a less certain possibility.
On the other hand, there seems to be no disagreement about Beech wood being a wood that was used for early ouds. Beech is the wood used on my old Egyptian oud for the inside core.
I have lots of air dried Elm wood to choose from (about 20 years air dried) and some European Beech (only just sufficient for the job - reclaimed from old furniture, about 50 years old).
Both woods would be suitable for the job. Elm is very tough and difficult to split, similar to Ash in outward appearance. Beech is a darker wood but, perhaps, more stable than Elm. Elm is lighter in weight (less dense) than Ash or Beech. However, I do not know if there is any evidence for Elm being used in surviving ouds or lutes.
Decisions decisions! So I may make up laminates of both woods to judge appearance etc. before making a final choice.

Thanks ALAMI - a good question that I have not given thought to before - except for double peg box, extended neck lutes, where balance of the instrument may be noticeably affected if heavy woods are used for the neck.
I try to build lutes as lightly as possible consistent with strength.

Curious about where the centre of gravity of this oud may be, I ran some tests this morning to find out. The oud is, of course, not complete but could be assembled sufficiently to assess the balance point - the pegbox with pegs having an estimated weight of 67 grams and fingerboard an estimated weight of 26 grams. The other unfinished components were weighed on a digital scale (plus or minus 1 gram accuracy). The bowl, soundboard and bridge assembly weigh 434 grams, neck 88 grams - total weight at this stage of 615 grams.
The temporarily assembled oud (with weights representing the pegbox and fingerboard taped in their appropriate locations) was balanced upon a cardboard tube clamped to my bench (to prevent it rolling).
The balance point (centre of gravity or C.G.) was found to be a few millimeters above bar #5. Interestingly, this is almost at the mid point of the overall length of the oud (Point B) measured from the top of the nut to the bottom of the bowl (315:325 mm). Furthermore, using the same procedure to find the balance point of just the bowl with sound board and bridge in place, the C.G. (Point A) was found to be exactly at the mid point of the length of the bowl measured from the neck joint to the bottom of the bowl. This rather surprising result (I would have expected the balance point to be located lower down towards the bridge) was verified by calculating moments of the individual component weights about the measured C.G. of the assembled oud.

I then used the same procedure to find the C.G. of two of my lutes - both fully complete with strings and tied on frets. As the lutes have proportionally longer necks than the oud, the C.G. was found to be located closer towards the neck block. In each case, the C.G. position was mid way between the brace (located immediately above the upper edge of the sound hole), and the inside edge of the neck block. Surprisingly, again the location of the C.G. was found to be exactly at the mid point of the overall length (top of nut to bottom of bowl) of the smaller lute (34.5:35.5 mm) and very close to mid length for the larger lute (39.5:40.5 mm).
The small lute is based upon a late 16th C lute by Giovanni Hieber and weighs 647 grams and the large lute - a reconstruction of an early 16th C lute by Laux Maler - weighs 726 grams.
So, for these three instruments with quite different dimensions, the balance point happens to be very close to the mid point of overall length. These results may have some significance or may be just coincidental.

I checked out my old Egyptian oud that has a relatively large, thick and heavy bowl, weighing 919 grams assembled (without strings). It has a softwood neck core. The balance point was found to be at 39:33 mm of the overall length - i.e. positioned more towards the bridge than on the above instruments.

I understand that Nahat ouds - although they have smoothly rounded bowls externally are then (unlike my Egyptian oud) scraped down inside the bowl to a uniform bowl thickness of about 1.5 mm. The balance point on these lighter bowl instruments might, therefore, again be at the mid point of overall length?

It would be interesting to know where the balance point of other surviving old ouds might lie. Any offers?

The neck core has been worked down to about 1.5 mm undersized with a block plane and then finished smooth, and straight with 120 grit garnet paper. I generally avoid use of sand paper where possible (dust as well as grit on work surfaces that can damage tool cutting edges) but this is a good application - the wide surface of the paper removing all slight dips and irregularities in the surface left by the plane. I check for surface irregularities by 'feel' - fingertips can be sensitive enough to detect the slightest irregularity otherwise not visible.

The neck will be veneered with Ash wood cut (with a band saw) from the same rib stock as the bowl. Veneering may be done with a single piece or multiple pieces. A single piece veneer will not be used as it will be difficult to find a piece with straight and uniform grain from the stock in hand. I could veneer the core with eleven strips - as a continuation of the bowl ribs but have decided instead to use a two piece 'book matched' veneer which may be more of a challenge.
Having made a card template of the veneer panel (measured from the finished core) the chosen veneer stock has been planed to just under 2mm thick and cut oversize to the template. The veneer stock chosen has a grain direction matching the taper of the template as well as some 'pin' knots that should add some visual interest to the otherwise plain character of the Ash wood. All cosmetic - mainly for appearance.
The curvature of the neck section is quite small (about 40 mm in diameter at the nut end) so the relatively thick veneer will be 'marinated' prior to bending to soften the wood and allow the veneer to readily conform to the tight curvature of the neck. This is something of an experiment. If it fails, then I will veneer the neck with strips as an alternative.

The thinner (1.9 mm) veneers have been marinating in a 50% Ammonia Solution and 50% Methanol (wood alcohol) solution for 7 days in preparation for pre-bending.

The veneer will be bent on the neck which has been protected with thin card taped in place - centre lines marked as a reference for positioning the veneer.
The first veneer - after removal from the marinade - was heated on both sides using a hot air gun until all residual fluid was seen to evaporate from the pores of the wood grain on the surface (about 3 minutes or so). The veneer was then centered on the neck, tied in place with wide 'Hockey Boot' laces, and left to dry overnight. The wide laces provide a fairly uniform pressure on the veneer to minimise any compression defects on the surface of the softened wood - particularly at the edges.

On removal from the neck the following day, the veneer had obtained a uniform 'set'. To ensure maintenance of this profile, the veneer was tied with masking tape and left to dry for a longer period.
The second veneer was subject to the same operation this morning.

The thicker veneer test pieces will be left to marinate for another 7 days - just to see what happens. The detailed results (macro photos of cell structure etc.) will be posted in a new thread (Marinated Wood Part 2) for information.

With no in depth experience of musical instrument acoustics it will be necessary to begin at the beginning in this experimental investigation.
The attached image is an engraving of a 'Helmholtz Resonator' - a device used by the great 19th C scientist Professor Hermann Helmholtz to investigate the tonal components of sound. The device is a glass sphere with a short neck open to the atmosphere (a) at one end and a small hole (b) at the other placed in the ear to better hear the resonant frequencies. The resonators are tuned to respond strongly to a specific tone frequency by altering the volume of the sphere and dimensions of the neck. Helmholtz employed a number of resonators in his experiments - all tuned to different tone frequencies.
The resonator works when the mass of air in the neck is forced (by the pressure of a sound wave) against the trapped air within the sphere which acts like a spring causing the air in the neck to oscillate at the tuned frequency.

In Appendix II of his book "On the Sensations of Tone", Helmholtz provides some useful information giving the dimensions of his resonator devices. He explains that spherical resonators are the most efficient (compared to those of cylindrical, conical or other geometrical shape) - partly because their other proper tones are very distant from the primary (resonant) tone and so are but little amplified and partly because the spherical shape gives the most powerful resonance. He goes on to say that "the walls of the sphere must be firm and smooth, to oppose the necessary resistance to the powerful vibrations of air which take place within them, and to impede the motion of air as little as possible by friction".
He lists the dimensions of 10 of his resonators the ratio of diameter of the neck to diameter of sphere ranging from 0.35 for the smallest sphere to 0.23 for the largest - recommending an optimum ratio of between 0.2 and 0.25 for best results.

The design of a spherical resonator involves gas thermodynamics, use of differential calculus and all of that good stuff. However, others have done this work reducing the determination of the resonant frequency to the simple equality given in the attached image where:

f = resonant frequency in Hz or cycles per second
c = the speed of sound in air (343 metres/sec. at 20 C).
A = the cross sectional area of the neck.
V = the volume of the sphere.
L = the equivalent length of the neck

A copy of "On the sensations of Tone" may be downloaded free of charge from

http://www.archive.org

Search "acoustics helmholtz" under "Texts"

I know less than nothing about acoustics, but I'm enjoying your postings nevertheless. I especially like your tying together of theory and experiment; Friar Bacon would be proud of you.

I wonder what the effects of a rose are on the sounds, both in terms of resonant frequency and of filtering out some notes. I though these could include dampening down of high frequency squeaks and plectrum sound, but the size of the spacing between the parts of the rose is measured in millimetres. This would give frequencies of well above hearing: bat level.. So perhaps I'm talking nonsense.

Turning to the oud, the first step is to determine the volume of the bowl. As the bowl is complete and the sound board yet to be glued in place, a direct volumetric measurement was possible using granular solids. I used cracked corn but rice or wheat grains or fine sawdust would do just as well. The volume of grain was carefully measured using a kitchen 1 litre measure, the bowl being filled to the brim. Using this method the bowl volume was measured as exactly 12,000 cubic centimeters.

A second method used was to divide the bowl template into
1 cm wide segments, measure the mean radius of each segment and calculate the volume of each one which for this instrument is semicircular in cross section (see attached image). Adding the volume of all segments gave a total calculated volume of 12,360 cubic centimeters.
This method over estimates the volume because the bowl section is not a smooth semicircle but is made from wide ribs and so facetted like a lute. Also the calculation takes no account of the thickness of the interior rib reinforcement strips.
The difference of 360 c.c. is equivalent to an evenly spread thickness of about 1.5 mm on the interior surface area of the bowl. (The interior surface area was calculated by measuring the area of the rib pattern and multiplying by 11). This seems reasonable.
Therefore a volume of 12,000 c.c. for the bowl will be taken as close enough.

The first question is - does an instrument with a complex rose pattern, like a traditional oud or lute, act as a Helmholtz resonator in the same way as, for example, a classical acoustic guitar with an open sound hole?
I currently have two functional lutes - a copy of a seven course late 16th C lute by Giovanni Hieber and a reconstruction of an early six course 16th c lute by Laux Maler. These were both tested - like the classical guitar in this project - by sliding a card over the rose while sounding each course. Both instruments responded most loudly when the sound hole was fully uncovered (as in the guitar tests) when the 5th course was sounded. The Hieber lute is tuned in 'g' so the Helmholtz resonance in this case is 'c' and the Maler lute is tuned in 'd' so the resonance is 'g' - i.e. at concert pitch A440.
The lute makers of the 16th C to 17th C seemed to have understood the relationship of sound hole diameter to bowl volume as it affects bass response as the evidence suggests that they adjusted the sound hole diameters accordingly.
Lute rosette patterns were standardised so the luthiers - if they required a smaller or larger sound hole area would either cut only part of a pattern or would add an additional ring of holes to increase the area - dependent (presumably) on the relationship of sound hole diameter to bowl volume giving maximum bass response.
For example the first image shows a standard 16th C rose pattern in its basic form, expanded with an additional ring of holes and reduced in area by cutting only part of the pattern.
The second image shows another pattern (the 'Gerle' rosette subject of this project). Some larger lutes used this pattern, increased in diameter with an additional ring of holes, (e.g. on a great octave bass lute by Michelle Harton, 1602) or smaller in diameter by using only part of the pattern (e.g. Martin Hoffmann, 1692).

The rosette pattern chosen for this project (often used by many luthiers of the 16th and 17th C) is called here the 'Gerle' pattern.
The first step in the analysis is to break down the rosette into its basic design elements and to measure the open area of each element.
As all of my historical rosettes were originally traced on transparent paper, the 'Gerle' pattern was placed on a mm squared paper background, scanned and then enlarged.
The area of each element in the design was then measured by counting the squares.

I am using my classical guitar as a test bed - to learn more about the resonance phenomena - with an open sound hole compared to resonance with a rosette or reduced sound hole diameter. Although a guitar is being used, I would expect ouds or lutes to provide similar results.

In an attempt to quantify the results for comparative purposes, a reasonable quality digital recorder (Zoom H2) - positioned directly over the sound hole - has been used to record the sound for each test and the recordings analysed using Audacity - the free audio editor.

Test 1
A 3 mm thick card was used to seal the sound hole closed and - after sounding the 5th (A) string - was quickly slid to fully open the sound hole. At the point where the soundhole was completely uncovered, the sound volume suddenly increased with a definite pressure pulse. Repeating the procedure for the 6th (E) string, there was a similar sound pulse at the point where the sound hole was uncovered by about a third.

The first image (stereo channels) shows the 'loudness response' with the A string sounded - X being the point when the soundhole is fully uncovered.
The second image shows the same response with the E string sounded - X being the point when the soundhole is partially uncovered.
The last image shows the frequency response analysis for the A and E strings sounded.
The peak loudness points for the A string graph are A = 111Hz A2, B = 217Hz A3 and C = 435Hz A4.
For the E string graph, A = 81Hz E2, B = 161Hz E3 and C = 328Hz E4.
A couple of sound clips are also attached to illustrate the sound pulse that can be distinctly heard when the (Helmholtz) resonant frequency of the enclosed volume of air in the guitar body occurs.

All a bit' rough and ready' at the present time as I learn how to use the hardware and software and interpret the results - to try to understand what it all means!



[file]11480[/file]

[file]11482[/file]

[file]11483[/file]

The attached images show the experimental set up to test the Helmholtz resonances with the A and then the E string sounding.
All a bit rough and ready - just to get a feel for the possibilities. The guitar back is resting on a carpeted floor. The Zoom H2 digital recorder is positioned directly over the sound hole mounted on a camera tripod. The recorder has four microphones - two facing forward and two facing rearward. So that the digital display could be read during the test, the two rear facing microphones were used to record sound - aligned along the centre-line of the guitar sound board (for convenience).
The first image shows the start position of the card - completely covering the sound hole.
The second image shows the position of the card that gave maximum sound amplitude with the E string sounding (about a third uncovered) and the last image shows the position of the card that gave maximum sound amplitude with the A string sounding (fully uncovered).
The waveforms and frequency analysis charts previously posted confirm that the sound hole to body volume geometry accentuates the sound volume of the two bass strings using the Helmholtz resonance effect.

With the recorder placed in this position, the right microphone (facing the finger board) records a lower sound volume than the left microphone (facing the bridge). The recorder will, therefore, be re-oriented for repeat testing to minimise this effect.
I am also surprised to find that the sustain of the A string - given by the waveform - is only about 5 seconds, possibly because I may have exerted some pressure on the card in the fully withdrawn position thus restricting the free vibration of the sound board (?).
All to be re-checked with further tests.

The second trial will investigate what happens when the guitar sound hole is covered with a rosette with an open hole area, relative to the open sound hole area of the guitar, of 0.29.
To save time, the test rosette has been made from 1mm thick card accurately pierced with holes of precise diameter using a tinsmiths hole punch. A preliminary test with the rosette taped in place gave similar results to those of Test 1 except that the sound volume at resonance was slightly reduced and the sustain of the A string was much greater. However, as the card has some flexibility, the test will be repeated with the rosette reinforced with small cross braces and with the rosette taped in position all around its circumference.

The third test will be to reduce the guitar sound hole diameter so that its area - like that of the rosette - is 0.29 of the full sound hole area.
Preliminary tests indicate that the Helmholtz response is minimised (almost eliminated) with the A string sounding and maximised with the E string sounding - which is what might be expected. However, this test will be run again with the inside edge of the card sound hole reinforced to reduce any flexibility and the card fully taped in position like the rosette.

The small sound hole and rosette test cards have been reinforced and coated with a hard 'spar' varnish ready for a trial re-run.
The digital recorder was positioned as before but tilted to try to reduce the difference in recorded sound amplitude between the left and right microphones.
Ideally, a special rig should be contrived to ensure a consistent plucking force on the strings. However, with a little practice, reasonable consistency was achieved without need for special equipment. The strings were plucked with fingertips not finger nails.
Using "Audacity" freeware audio editing software, the waveforms were produced from the recorded sound tracks for comparison - the horizontal axis measured in seconds and the vertical axis in decibels (a logarithmic measure of sound level or sound intensity). The waveforms were 'photographed' to image file using screen capturing software (freeware MWSnap)

Re-Test 1. The response with guitar sound hole uncovered and then when sliding a card from fully covered to open was recorded with first the A and then the E string sounding.
The sliding card action produced a sound pulse when the sound hole was fully uncovered with the A string sounding and when the sound hole was about a third uncovered with the E string sounding (points X on the waveform graphs) - confirming the results previously posted. The pulses are the Helmholz resonances for frequency A and E.

The recorded maximum sound duration of the waveforms is only about 5 seconds yet the sound of the string vibration could be heard for about 15 seconds.

Comparative test results for the rosette and small sound hole are to follow.

I am also at present trying to understand what is going on shayrgob - so you are in good company. Now retired from the workforce, my training as a professional engineer did not include instrument acoustics.
Hopefully the data collected here will eventually be of some use at a later date.

A little surprised by the brief sustain of the recorded waveform (perhaps because the sound samples were recorded at lowest sound levels?), I ran a test to check the sustain of the guitar using my own hearing as judge. The sustain in each case, measured with a stop watch, was about 16 seconds for the open sound hole, rosette and small sound hole - no difference between them. With the guitar sound hole sealed with 3mm thick card taped in place, the measured sustain was about 14 seconds. I conclude from this that the guitar might work quite well acoustically if it had no sound hole at all!

Test 2. Rosette taped in place and sound recordings made first with the rosette open and then from closed to open (as in Trial 1) sounding first the A and then the E strings.
The attached images show the waveforms.
It is interesting to note that, compared to the guitar open sound hole tested in Test 1, the duration of the waveform for the A string is almost doubled with a much smoother attenuation (i.e. reduction in amplitude with time of the waveform) from the maximum sound levels.
On the other hand, the E string waveforms in Test 1 and 2 are comparable.

Test 3 with the small sound hole to follow.

Test 3. The small sound hole is 29% of the area of the guitar sound hole (and has the same open area as the rosette).
As anticipated, the smaller sound hole area moves the Helmholtz effect pressure pulse to tone E - almost, but not quite, as a small pulse can still be seen in the waveform of the A string.
Note also that the attenuation of waveform is about 6.5 seconds compared to about 5 seconds for the open guitar sound hole.
The difference between the rosette and the small sound hole waveforms with the A string sounding - both of equal open area - is distinctive.

Note that when comparing the waveforms, the scale of the horizontal time line can vary. The sound amplitude in Decibels is consistent between files.

Out of curiosity, an additional test 4 with a sound hole area of only 16 % of the guitar sound hole will be tried next - to see if the Helmholtz resonant effect will be moved further away from both A and E tones.

I should have noted that in test 3 (with a 49 mm diameter sound hole) when the Helmholtz resonance effect shifted to E as expected, the A string waveform shows a pressure pulse but this was inaudible.

Test 4 - the sound hole diameter is reduced to 35 mm from the full sized guitar sound hole diameter of 87 mm.
As expected, this shifted the Helmholz resonance effect to a lower tone than A or E although a small (but inaudible) pressure pulse can still be seen in each of the waveforms.
The overall sustain of both strings increased to about 20 seconds.

It might concluded from these tests that there may be some latitude in sound hole diameters that will work to produce a specified resonance tone. This may be due to the fact that a guitar (or lute or oud for that matter) is a much less than perfect Helmholtz resonator so that the resonant pitch is less critically focussed than it would be if using a rigid walled spherical resonator.

The waveforms for the rosette - with only 29% of open sound hole area of the guitar - show the same sound pressure pulses at almost the same sound amplitude as that of the guitar sound hole. Interestingly, In the case of the A string waveform, the sustain is almost doubled with a smoother attenuation.
The rosette is essentially a multitude of small sound holes - each (possibly?) with a much longer end correction than the open sound hole of the guitar - so the cumulative effect may add up to be equal to that of the guitar sound hole.
The sound pulse in the case of the guitar may be quickly dispersed (attenuated) whereas it would take a longer period of time for this to happen with a rosette? Just guessing at this point in time, however.

Now that I have become more familiar with using the Zoom H2 digital recorder and the Audacity audio editing software, my previous concerns about the sensitivity of the equipment for recording sound were unfounded.
In fact, the audio fidelity of the recorder is very good so that the full sustain of the guitar over time could be clearly heard and measured by listening to the sound tracks played back through the stereo headphones of the recorder.
The Audacity software also does a good job in accurately measuring and indicating the sustain duration in the sound volume bar for each recorded channel - although, of course, the full sustain could not be heard through my low quality computer speakers. The sound amplitude, presented in the recorded waveforms, is somewhat limited so that after the first burst of sound lasting a few seconds the continuing sound wave cannot be easily distinguished without considerably magnifying the scale of the waveform.
A neat feature of the software is the frequency analysis facility which allows a small section of the waveform to be highlighted and the peak frequencies against volume in decibels in the measured sound displayed as a graph. Moving a cursor along the graph displays the peak frequencies and equivalent tones (C2, G5 etc.) as well as the frequency at the cursor position.

This concludes the sound hole/"Helmholtz resonance" tests for this unplanned interlude diversion in the thread but further tests will be continued once the project oud is completed.
To this end, I plan to build a second soundboard with the same bracing as the first but with four sound holes - designed to sealed closed or left open, fitted with reduced diameter sound holes, with or without rosettes etc. Various sound hole combinations may then be subject to testing - all in the interests of science!

I just though that I ought to complicate matters, as the acoustics of ouds, lutes and guitars have proved to be so simple.
To add to the rosettes of oud and lute, tornovoz or tornavoz have been used in guitars, as well as the parchment roses from the baroque. There's a whole world out there to investigate.
There does seem to be some consensus that rosettes increase the sustain, especially lower down the register (vis http://thamesclassicalguitars.com). This supports your experimental results.

I wonder whether the complexities of shape in oud, lute or guitar are such that Helmholtz has little to offer here.
Keep up the good work.

The engraving, upon which the design of the project oud is based, is supposed to appear in an early 14th C copy of the Kitab al- Adwar by Safi al-Din al-Urmawi - although this is yet to be positively confirmed.

A detailed analysis of the scale system developed by Safi al-Din can be found in the paper "Safi al-Din al-Urmawi and the Theory of Music" by Dr Fazil Arsian published by the Foundation for Science Technology and Civilisation. Free for download and recommended reading for those interested at

http://www.muslimheritage.com

Dr H.G. Farmer in his paper "The Lute Scale of Avicenna" describes the historical development of the various scale systems for the oud from the earliest times up to the theoretical works on music by Ibn Sina (otherwise known as Avicenna) in the 11th C.
He ends his article by stating that although the scale system developed by Avicenna was certainly an improvement on Al-Farabi's scale and tuning, it was not until the time of Safi al-Din that an absolutely perfect scale was evolved.

The strings of a four course oud from bass to treble were named "Bamm', 'Mathlath', 'Mathna' and 'Zir'- tuned a fourth apart.
Although Ziryab was first to add a fifth string to the oud
(9th C Moorish 'Spain'), it was Al Kindi who first suggested adoption of this innovation throughout the East. Al Kindi called the fifth string 'zir thani' but the name given to it by later writers was 'Hadd'. The 'Hadd' string was a treble string placed above the 'Zir' string (or below - depending upon one's perspective!) and tuned in interval of a fourth higher.
The fifth string allowed the expansion of the tonal range of the oud to two octaves.
The various proposed theoretical placement of the frets, founded upon Pythagorean intervals, is complex (no wonder frets were eventually abandoned by the oudists!) and, as I am currently still on a learning curve on this subject, am in no position to go into any detailed discussion about this or that theoretical fret placement. In any case, oud players of the time did not use all of the fret positions dictated by theory but selected those that suited the musical customs of a particular region, their personal preferences, best for a particular mode etc.

Dr Fazil Arsian describes, in schematic, the oud tablature used by Safi al-Din. As the number of frets (7) on the oud engraving matches the tablature schematic this may be partial confirmation that the engraving is indeed part of the Kitab al-Adwar.
Dr Fazil Arsian's paper is published in English so he accordingly transcribes the original Arabic text into Roman (while referencing the original Arabic). The attached image shows his transcribed tablature format overlaid on a tracing of the oud engraving.
For information and comparison an example of early German lute tablature is attached. This is from the book 'Utilis and Compendiara' by Hans Judenkunig, dated 1523.
While everything is in reverse and upside down - note that the German tablature - although for a 6 course lute - was originally designed for a 5 course lute (open strings from bass to treble numbered 1 to 5). The additional 6th course in the bass requiring a new set of characters A, B, C .... (This modification can also be found in the earliest examples of 16th 'French' tablature originally devised for a five course lute).

German tablature - the earliest example for the lute - was supposed to have been invented by blind organist Conrad Paumann - but was it? Or was this just an adoption of 13th C oud tablature?

As an aside, I have always been intrigued by the name of the author of 'Utilis and Compendiara' which translates to 'King of the Jews'. Does this have some, as yet, unexplained significance - a secret title, 'nom de plume' perhaps? But why?
More to follow.

In his paper "The Lute Scale of Avicenna", Dr H.G. Farmer observes a variation of the theoretical scale system proposed by Ibn Sina. This improvement on the Al-Farabi scale was achieved tuning by tuning the 'Zir' string a major third from from the 'Mathna' string because "it embraced the elusive Zalzalian notes in the second octave"

This is an interesting theoretical scale variation, as it relating (as it seems) to tuning of the Spanish Vihuela and early Guitar.
The early 4 course Guitar (in my opinion) is a 'cut down, low cost' version of the 6 course vihuela - eliminating the need for a (very expensive?) treble string(s) and costly basses. Likewise for the 5 course vihuela.
The attached image is a comparative schematic of the relative tunings for information.

Next, a preliminary calculation and evaluation of string diameters (gauges) based upon historical evidence for oud strings made from silk and gut.

It is clear from the early Arabian and Persian texts about the oud (see "The Structure of the Arabian and Persian Lute in the Middle Ages" by Dr H.G. Farmer) that the strings were made from silk as well as sheep's gut - sometimes all silk strings were used, sometimes all gut and sometimes a combination of silk and gut.
The string gauges for gut and silk were the same or nearly so.
The texts provide some information on the structure and manufacture of the strings. In the case of gut strings the information is not very helpful (gut being a coarser material than silk filament).
However, for the silk strings, sufficient information is provided to allow calculation of the string gauges for each string.
The 14th C Persian work "Kanz al-Tuhaf' is particularly informative giving the number of silk threads for each kind of string - 'Bamm' 64 threads, 'Mathlath' 48 threads, 'Mathna' 32 threads, 'Zir' 24 threads and 'Hadd' 16 threads. The number of threads per string agrees closely with data given in the earlier
10th C work "Ikhwan al-Safa".
The silk threads were twisted into a uniform cylinder and bound together with gum paste which also added additional mass to a string.

The first step is to try to determine the cross sectional area of the silk thread used to make these early strings.
The string length of the project oud is 56 cm which determines the maximum pitch of the 'Hadd' string (to avoid frequent breakage). For a simply twisted gut string made from modern gut this would be g' (at A440 pitch) so it might be assumed that this upper limit will apply to a silk string as well - at least for this preliminary assessment. Assuming a string diameter of 0.44 mm for the 'Hadd' string, it is then a simple matter to calculate the diameters of the other strings. The attached notes show the procedure for information.
The results are 'Bamm' 0.88 mm diameter, 'Mathlath' 0.76 mm diameter, 'Mathna' 0.62 mm diameter, 'Zir' 0.54 mm diameter and 'Hadd' 0.44 mm diameter - all these dimensions seeming reasonable at first glance.
However, there is a potential problem.

Returning to the traditional oud tuning of Al-Kindi et al. as previously posted. The string length of an oud (or lute or vihuela etc.) strung in gut or silk dictates the maximum pitch of the treble string and the design pitch of the instrument. 16th C lute and vihuela practice was to tension the top string as high in pitch as it would go without frequently breaking and to tune the remainder of the strings accordingly. Why?

Taking the project oud as an example and tuned in fourths would give a tuning of B e a d' g' - g' tone being the upper limit for the top string (in gut or silk) at International concert pitch standard A440. (Note that the modern pitch standard was established in 1939 - prior to that there was no commonly agreed standard, certainly not in the 14th C - as far as we know).
Assuming the instrument to be strung in 'Pyramid' plain gut and using the 'Pyramid' string slide rule (a low cost, handy tool) for convenience we find that string tension for the 'Hadd' string, 0.44 mm diameter, g' tone is 3.7 Kg (or 37 Newtons), for the 'Zir' string, 0.54 mm diameter, d' tone - 3.3 Kg, for the 'Mathna' string,
0.62 mm diameter, a tone - 2.3 Kg, for the 'Mathlath' string, e tone - 2.1 Kg and for the 'Bamm' String, 0.88 mm diameter,
B tone - 1.6 Kg.
So here the problem is that the string tension of the 'Bamm' string is too low for the string to sound well.
Tuning according to the Ibn Sina alternative tuning of fourth, fourth, major third, fourth would give a tuning of c f b d' g' would increase the respective string tensions - bass to treble - to 1.8 Kg. 2.3 Kg, 3.1 Kg, 3.4 kg and 3,7 Kg - the tension of the 'Bamm' being just about at the limit for bowed stringed instruments never mind plucked stringed instruments.

The strong revival of interest in the European lute began during the 1960's. Lutenists attempting to string their lutes with modern gut strings (that worked reasonably well for bowed instruments) were to be greatly disappointed with the results - the trebles sounded well but the bass strings were totally hopeless.
Marin Mersenne in the early 17th wrote that " the sound of the largest diameter lute strings of the lute can be heard to last for a sixth or third of a minute - that is to say for the time that the pulse of a healthy man at rest beats 10 or 20 times". The lute of Mersenne's time had 10 courses. The performance of these thick gut strings would appear - astonishingly - to match or exceed the performance of modern nylon, metal overspun, bass strings. To ensure that there would be no mistake about what he meant, Mersenne gives two reference datum points.
(Also Parisian males of the early 17th C would seem to be a pretty healthy bunch of guys!)

This has been the dilemma facing modern researchers and makers of 'historical' instrument strings - how to construct the bass (gut or silk) strings of a plucked string instrument like an oud, lute, vihuela, guitar etc. so that they sound well.
Time out! - more to follow.

The attached image is a scan of the original Mersenne text - hopefully my translation renders the meaning correctly.

The gut lute strings of Mersenne's time were sold bound up in 'knots' - so were very flexible (like bootlaces). The attached image, an engraving of a lute string knot from Mersenne's 'Harmonie Universelle', is an indication of the flexibility of the string. Modern (non historical) gut strings could not be wrapped up in this fashion without being damaged so are, supplied wound up in coils.

Recent research by historical string makers like Eph. Segerman (Northern Renaissance Instruments) and Mimmo Peruffo (Aquila) has gone a long way to developing an understanding early string technology as it relates to gut strings.
For silk strings much promising research and development work has been undertaken by Alexander Rakov of New York State on a non commercial basis.

For those interested in exploring the subject matter in detail, informative research papers are available for free download from N.R.I. ('A Primer on the History and Technology of Strings') at

http://www.nrinstruments.demon.co.uk/StrPrim.html

and from Aquila (see 'Our works') at

http://www.aquilacorde.com/indicemusicaclassicaing.htm

Briefly, the tonal range of the thicker gut strings can be extended by increasing the amount of twist of the gut strands during manufacture or by constructing the strings like a yarn or a rope for an even greater extension of the tonal range.
Mimmo Peruffo has also produced strings 'loaded' with materials to increase the linear mass of a string and reduce the diameter accordingly.
Whether or not these methods were the same as those used by the early string makers is impossible to determine as no original strings survive and the written historical evidence is scanty and (frustratingly) incomplete.

Silk strings might have been made by just simply twisting the filaments together (like a plain gut string), the whole structure then bound together with 'gum' (It is assumed by some that 'gum' means 'Gum Arabic'. However, Alexander Rakov has made successful strings using the natural gum (sericin) attached to raw silk filaments).
Alternatively they might have been made like a yarn or rope. Early Chinese strings of great antiquity were made like ropes, some with a smooth outer wrapping of silk and a roped silk core. Although traditional silk strings have generally been replaced by those made from nylon (or even metal) for instruments like the Chinese zither (Ch'in) there has, in recent years. been a revival to try to manufacture silk strings the 'old way'.
It is likely that the silk strings first used on early ouds came from China. Later, Moorish 'Spain' may have been a centre for instrument silk and gut string manufacture - sericulture and wool being important, flourishing agricultural industries prior to their demise brought about by government punitive restrictions in the latter half of the 16th C.

For the project oud with five courses, it is anticipated that if gut strings for the 'Bamm' and 'Mathlath' are to be used a 'high twist' gut string should be sufficient to give reasonable acoustical results - but some experimentation, in an effort to obtain optimum results with different string combinations - silk or gut - will be necessary

For information, the second image attached shows a medium twist gut string compared to a modern reconstruction of a wrapped silk string made for a Chinese Ch'in - the latter illustrating the flexible roped construction of the string core.

Metal overspun instrument bass strings - made by winding metal wire over a flexible core - so widely used today, were not invented until the late 17th C and likely did not come into general use until later in the 18th C. Instruments such as ouds and guitars then commonly used gut trebles with silk cored metal overspun basses until about 1960 when nylon strings became generally available. A set of oud strings that I purchased in Cairo in the early 60's were all gut trebles with silk cored metal overspun basses.

Clearly, use of this type of string is not an option for an instrument that is to represent a 14th C oud or lute. Of the two authentic alternatives, high twist gut strings from one of the historical string makers are readily available today and should provide a satisfactory performance.
To my knowledge, historically accurate silk oud strings are not commercially available - so part of the work of this project will be for me to make my own silk strings (hopefully!)

Sericulture and use of silk filament to make fabric is an important industry going back thousands of years in China and so the technology is well documented - much of it freely available. Less so the methods for making silk strings. Surviving early Chinese texts dating to the 12 C are more helpful than the brief description in 14th C Persian Kanz al-Tuhaf - but every scrap of information could be useful.
Silk is a good string making material being strong and elastic and producing uniform strings.

Silk filament is secreted by a variety of moth caterpillars the most prominent being the 'Bombyx mori' species, originally domesticated by the Chinese solely for silk production. The caterpillar pupates to an adult moth inside a silk cocoon - the cocoon silk being produced by the caterpillar in a continuous filament measuring up to 1500 metres long. After (sadly) killing the pupa inside a cocoon, the silk may then be reeled off as a continuous filament.
The raw silk filament from the cocoon (Bave) is actually a double strand (two Brins) of silk protein (fibroin) cemented together with gum (Sericin). Silk filament may be used in this condition (Raw silk) but usually the Sericin is removed by boiling in water (De-gummed silk).

Silk filament is quite variable in its physical properties dependant upon species of moth, conditions during raising and feeding the caterpillars etc. The cocoon Bave is tapered along its length ranging in diameter from about 30 microns (0.03 mm), for the first 200 - 300 metres reeled from a cocoon, down to about 12 microns
(0.012 mm).
In order to make a silk thread of sufficient diameter for practical use, the filaments from several cocoons are simultaneously reeled off and twisted together to make a thread.

We already know the number of threads that make up 14th C oud strings but how many cocoon Bave filaments were used to make a single thread?
If a string maker was to select only the first 200 metres or so of the Bave reeled from a cocoon, the cross sectional area of the Bave would be about 0.00071 sq. mm.
We already know that a 'Hadd' string of 0.44 mm diameter was made from 16 threads of cross sectional area each measuring 0.0095 sq mm. From this we can conclude that there are 13 cocoon Bave filaments to a single thread.

This is interesting as it compares with the 12 Bave filament thread used to make the ancient Chinese Ch'in strings (more modern Ch'in strings had 9 Bave filament threads).

Taking it a bit further, if the oud strings were made with 12 Bave filaments, the diameter of a 16 thread 'Hadd' string would then become 0.42 mm. This then may be a better starting point for the project oud silk strings rather than using the arbitrary string diameter of 0.44 mm. This will have the disadvantage of proportionally reducing the diameter and linear mass of the other strings with a possible reduction in acoustic performance of the bass strings. However, the elastic, twisted construction of the silk strings may be more than enough to compensate. Time will tell.

The attached image are my notes on silk filament for the information of those interested.

Next is to review how oud strings of the 14th C were made.

Re- commencing the construction part of this project - after an enforced hiatus of 3 months - the thread resumes where last left with veneering of the neck.

To recap, the two piece neck veneer - of 'Black Ash' wood, matching the ribs of the bowl - was 'marinated' in a 50/50 solution of 10% Ammonia solution and wood alcohol (Methyl Hydrate). The veneer blanks were then heated with a hot air gun, tied tightly to the neck (used as a mold) and allowed to dry.
Interestingly, on removal from the neck, the veneer then shrank to a tighter curvature than the neck radius. This will have the advantage of ensuring a close fit of veneer to neck when the veneer is glued in place.

For information, the attached macro image of the veneer cell structure - before and after (approximately) - shows the slight deformation (stretching) of the outer layers (or compression of the inner layers - or both) of the early wood cells. Otherwise, the cell structure appears to remain unaffected by the 'marinating' process.

The neck veneer is in two" book matched" pieces 1.8 mm thick (which should finish at around 1.5 mm). The joint edges were dressed straight with a plane and finished on a sanding block.

To facilitate handling, the neck has been temporarily screwed to the wooden block previously used to hold the neck during the shaping operation.

The first piece of veneer, held in position on the neck centre line with masking tape is then tightly clamped in place - after application of the glue - using a rubber bands.

Each pre-bent marinated veneer has a degree of longitudinal curvature as well as that of the cross section (due to balancing of transverse and longitudinal stresses in the wood). This makes the precise fitting of the veneer joint a little more complicated - requiring a certain amount of trial and error fitting.

With the joint fitting perfectly, the second veneer piece was coated with glue and the joint faces pulled tightly together with masking tape. The veneer was then clamped to the neck with rubber bands.

The veneered neck is now ready for final trimming and fitting to the bowl.

The pegbox sides will be made from Beech wood veneered with Ash. Beech has been chosen because it is one of the woods recommended as best for oud making in the early Arabic and Persian texts. Also Beech was often used by lute makers of the 16th and 17th C for pegboxes and my old Egptian oud has a pegbox of beech veneered with Mahogany.

I only have a small piece of old Beech wood (50 or 60 years old) to work with - just about sufficient to make the peg box sides - so there is no room for error. The Ash veneer has been selected with a curved grain pattern to match the curve of the pegbox sides. The Beech wood blanks have been sawn to about 8 mm thick and the Ash veneer to about 2mm. After the veneer has been glued to the sides, the Beech will be planed down to about 6 mm thickness and the veneer to about 1mm.

In order to hold the small pieces of Beech securely for the planing operation, I used a strip of double sided carpet tape to temporarily secure the blanks/ veneer to a flat piece of wood held in a vice.
This is a technique adapted from Dr. Oud's (Richard Hankey) recently published (and recommended) e-book 'Oud Repair' - for sanding thin patches. (For those interested, the book may be purchased on line at
http://www.droud.com/repair.htm)

The brand of carpet tape used here is fabric based and the 'tacky' glue easily holds everything in place during planing. To release the blanks/veneer from the tape, I used a thin spatula to carefully (!) pry things apart. The adhesive can also be softened with heat if necessary.

The shiny metal thing in the images is a metal layout template of the peg box sides.

The veneered peg box sides have been sawn and planed close to final profile and thickness.
The 'inside' curve of the sides have been shaped using spokeshaves. My favourite spokeshave is an 'old fashioned', wooden boxwood, low angle blade tool that belonged to my father. This type of spokeshave is subject to wear through use so I repaired this one by inlaying a brass strip in front of the cutting edge.
I prefer the low angle blade as it is easier to control than the more modern high angle bladed spoke shave. However, this little 'high angle' bronze spoke shave also works nicely for instrument and fine detail work. It is one of a three part set purchased from Lee Valley some years ago at less than half the current catalogue price of $27.50 Can. but still good value (Cat# 61P10.10).

The neck has been fitted to the neck block so the position and size of the peg box rebate can now be determined. A big advantage of the dovetail style of neck joint is that it facilitates precise set up without having to glue the neck in place.

The peg box rebate has been accurately sawn on a band saw by first centering and temporarily screwing the neck to a mounting block employed as a jig against the bandsaw fence and square guide for the horizontal and vertical cuts. The sawn surfaces will, of course, have to be 'cleaned up' with a chisel when the peg box is fitted.

The peg box block (at the nut end) will be made from spruce in order to simulate the unusual (and more complicated!) peg box construction found on my old Egyptian oud where the 'block' has been cut integral with the neck. The block at the 'tail' end of the pegbox will be made from a piece of beechwood and the back from Ash veneer.

The peg box will be tapered along its length - using the same taper as the peg box on my old Egyptian oud. To facilitate handling and gluing of the end blocks a simple tapered jig has been made from scrap pine. The pegbox sides - aligned, levelled and set at the correct overall width - are clamped to the jig with self levelling spring clamps.
The end blocks, tapered to an exact fit, have been glued in place. The blocks have been made much longer than their finished size so that the peg box may be securely held in a vice for the next step of the operation.

The top surface of the peg box must now be smoothed and levelled in preparation for the ebony/boxwood tile decoration that will match that on the bridge and the sound board and fingerboard banding (yet to come).

Levelling is done by 'draw filing' i.e. drawing a fine file sideways along the top surface. This removes any slight irregularities left by the spoke shave and ensures that both sides of the peg box are flat and even. The levelling is judged by feel (the slightest irregularity can be easily felt under the file) and by sight (light reflected from the surfaces reveals any small hollows remaining).
Final finishing of the longitudinally curved surfaces is done using flexible sanding sticks (actually nail manicure emery boards - cheap and readily available). This should only be done once work with the file is complete to avoid dulling the file cutting edges with any fine grit left in the wood.

Quote: Originally posted by jdowning

That's correct Ronny. You will find a discussion about the sound board wood on page 4.

There is sufficient material left over from the bridge tiles to cover the top surface of the peg box. Preparation of the tile blank is discussed on page 7.
Five strips must be cut from the tile blank. Ebony and boxwood are hard and brittle woods so the strips are cut using a sharp knife and fine tooth razor saw to avoid chipping of the edges of the wood.

Each strip is first marked using a straight edge and knife to scribe a line (several very light cuts). This line then acts as a guide for the saw. The work must be firmly clamped to a flat board during the sawing operation as any movement of the work will surely result in breakage of the tiles.

The knife used here is a readily available, good quality 'Olfa' brand craft knife. The blades on this knife can be snapped off in segments to reveal fresh, sharp cutting edges. A fresh cutting edge was used for this work.

The ultra-thin razor saw, 52 teeth per inch, is from Lee Valley (cat# 60F03.10 - $6.95 Can). The same saw is available from many luthier supply houses. This saw has a stiff back and cuts on the draw stroke so is easy to control for more delicate cutting operations. Nice, inexpensive tool.

The sequence of operations now is first to glue on the tiles, trim the peg box to final dimensions, drill the peg holes, finish the interior surfaces of the peg box before finally, fitting and gluing on the peg box back plate. Then, the peg box terminal may be fitted and shaped and the peg holes may be reamed and pegs fitted.

Quote: Originally posted by Ronny Andersson

Quote: Originally posted by jdowning   That's correct Ronny. You will find a discussion about the sound board wood on page 4. I am ashamed to have missed this. If I had known I would have helped you in getting hold of Lebanon cedar from old stock.

The tile strips have been glued to the top surface of the peg box. The strips were flexible enough to conform to the peg box curve without difficulty.
The width of the strips is currently about 8 mm. This will be reduced to about 4 mm wide (the same width as the sound board banding) when the inside face of the peg box is tapered in section to provide string clearance. This will be done after the peg holes have been drilled - which is the next step.

To support the peg box at the correct angle for drilling the peg pilot holes, a simple jig has been made from wood offcuts. - to be used on a drill press. This jig will be used one time only and then scrapped so does not need to be too fancy but does need to be accurate.
The support block on the jig is planed to the inside taper of the peg box. This provides backing to the inside face of the sides to eliminated any splintering due to the drill breaking though. The block is reversible so that the peg holes may be drilled from either side.
I prefer to drill the holes from each side (two operations) rather than from one side and drilled right through in a single operation. This requires accurate set up of the jig (square and level) and use of a metal layout pattern to ensure matching hole centres on each side of the pegbox.

The holes are drilled to 7 mm - just a bit less than the minimum diameter of my peg reamer - using a utility grade brad-point drill (designed for drilling wood). I do not like this type of bit as much as the (much more expensive) lipped brad-point kind as they do not cut as cleanly or as accurately but, as the holes will eventually be opened up with a peg reamer, this does not really matter at this stage. The main concern is to achieve as close an alignment of the holes as possible (although any slight misalignment may be corrected when the holes are reamed out when fitting the pegs). Hole alignment can be verified by pushing the drill bit though the holes and checking for squareness against the peg box sides.

Having tapered the upper inside surfaces of the pegbox, the bottom of the peg box has been trimmed to size. This was done on a band saw - holding the peg box square and at the correct angle on the drilling jig. The band saw cuts through the base of the jig in this operation but - no problem - the jig is disposable - for "once off use" - and will eventually end up on the wood stove.

Having trimmed the base of the peg box to size, the base of the drilling jig has been removed allowing the jig to be held in a vice. This is now used to hold the peg box for final smoothing and finishing of the curved bottom edges prior to fitting the peg box back plate. The taper of the jig mounting block holds the peg box securely during this final operation.

The question about how to interpret the enigmatic peg box finial represented in the engraving is still not fully resolved so it has been decided to proceed with the design previously posted (see page 9).
In order to better visualise the contours and geometry of the finial - as well as gaining an appreciation of the best way to handle the carving work and assembly - a 'mock up' of the finial has been made using soft pine and cardboard.

The material of the body of the finial has yet to be decided upon - ash, ebony, boxwood (or even pear wood stained black perhaps) - but minimising the weight is an objective.

A curved plate of ebony with Persian boxwood inlay (all about 3mm thick overall) will be mounted on the finial block. The outline of the plate will be pear shaped - identical to the geometry of the sound board. In order to arrive at the best proportion for the plate, several models of different size have been cut from thin card and tried out as part of the mock up assembly.
"If it looks right it is right" (to my eye) is the final criterion!

The attached image shows the trial and error shaping of the finial in order to arrive at the 'best' geometry. Here it can be seen that quite a bit more material can usefully be removed from the finial block to reduce weight and slim down the proportions.

One idea for material is to use a piece of Ash selected so that the wood grain follows the profile of the finial block. Just a thought at this stage, however.

The peg box back-plate has been fitted and the peg box trimmed and finished to size. The peg box is now ready to be be fitted to the neck and to receive the finial.

The back-plate has been made from a piece of the Ash rib stock - 2 mm thick, hot bent to the peg box curve and glued in place.
The peg box ends have been trimmed to size using a razor saw to avoid any danger of chipping the veneer and banding (the band-saw blade is too coarse for this work). After trimming, the end grain cut faces have been smoothed flat on a fine sanding block.

The body of the peg box finial will be made from box wood with a 3 mm thick central spine of ebony and the 'tear' shaped plate from ebony with a central inlay of box wood.

The first step in making the finial is to prepare the blank for the finial body. This is a 'sandwich' of Persian boxwood with a central core of African ebony. The blank has been made well oversize - wasteful but necessary for ease of handling during the preparation and carving stages of the finial body.

The small radius inner curve of the finial body has been cut with a spur bit woodworking drill to provide a clean cut result. A 'sacrificial' block of hardwood was clamped to the blank to provide support for the drill - the cut being made with everything clamped securely and square in a drill press running at lowest possible speed and with slow feed.

The rough profile of the finial body has been marked out ready for preliminary carving.

The upper part of the finial body has been shaped close to finished dimensions using sanding drums in a drill press. For final finishing to size the finial body will first be registered - correctly aligned - to the peg box end with a small piece of hardwood dowel inserted into pre-drilled holes. This will also facilitate accurately gluing the awkward shaped completed finial to the peg box.

The finial cap - 'tear drop' shaped to the same geometry as the body of the oud - will be made from ebony with a matching 'tear drop' inlay of boxwood. The cap will be made in two parts glued together - a base in solid ebony and the upper ebony/boxwood inlay.
The first step is to cut the hole for the inlay starting with a stick of ebony (for ease of handling) about 6 mm thick. After laying out the profile of the inlay, a 3/4 inch (19 mm) diameter hole was first clean cut through the ebony. This was done using a Forstner drill bit in a drill press at low speed/ low feed - with the work securely clamped to the drill table (essential for safety). The remaining waste was then cut out using a fret saw with fine toothed blade and finished to size with a fine double cut half round file.

Current extreme cold weather conditions are restricting workshop activities at present but - on the positive side - humidity in our heated kitchen is now around 50% RH - so time to start planning the assembly of sound board to bowl.

The base of the 3 piece, 'tear shaped' finial cap has been temporarily screwed in position and the upper inlay glued to the base. The screw will be removed after the cap has been glued to the finial body - the screw hole then being covered with the boxwood inlay.

At present the cap is oversize so has to be brought down to the required profile and reduced in thickness by shaping the upper surface into a curve.

The peg box finial has now been fully assembled and sculpted to its finished dimensions.
Final sanding and polishing will not be done until the finial is glued to the peg box - which will be after the pegs are fitted and the peg box glued to the neck.

As this is the first time I have tried using a dovetail neck joint I was a bit uncertain as to how it would work out. The dovetail fit is accurate (a sliding push fit) and the vertical joint surfaces close fitting - a big advantage for trial set up.
Uncertain of my ability to work quickly enough to successfully assemble the joint with traditional hot hide glue - and not wanting to use epoxy - I opted to use PVA glue (Lee Valley 2002 GF).

The problem with water based glues is that they almost immediately cause slight swelling of the wood so some additional clearance in the joint must be provided to avoid assembly problems. As it turned out the dovetail was too close fitting so that the joint 'seized up' before it could be fully pushed into place. Hammering the joint with a block of wood and a mallet forced everything (apparently) into place with glue squeeze out in the vertical joint. However, once the glue had dried a gap in the lower part of the vertical joint had opened up. The gap - about 0.25 mm wide and about 5 mm deep - was free of glue so was a weakness. Attempting to dis-assemble this tightly fitting assembly, under the circumstances, was not considered to be an option

To correct this situation and provide a fully glued vertical joint, the joint was widened with a saw so that black dyed wood veneer could be glued in place. The saw used for this operation was a flush cutting saw (Lee Valley Cat. 05K34.01 cost $19.95). This saw cuts only on one face - on the pull stroke - and so is easy to control for this work. The saw cuts an accurate kerf (slot) just a bit wider than the veneer thickness (about 0.65 mm).
After gluing, the veneer was trimmed back with a knife and 'safe edged' file (Nicholson Auger File).

Had I used epoxy glue this problem likely would not have occurred! Nevertheless, next time around I will still use a dovetail neck joint but with a slightly looser fit to allow use of hot hide glue. We live and learn!!

With the neck attached and repaired, the bridge could be aligned ready for gluing.

After taping the previously fitted sound board in place on the bowl with centres aligned and the brace ends in the correct position relative to the side ribs, a 'string template' was used to determine the correct bridge position. The full size drawing of the oud was used to double check the dimensions.
The 'string template' - reported in an earlier post - is made to the geometry of the outside strings at the nut and bridge positions relative to the neck joint position. (I made this from tinplate as I have the material and equipment to accurately cut the template - otherwise it would have been made from thin plywood).

The front edge of the bridge was then located with a strip of masking tape on which the string positions were marked (from the template) for reference and the bridge glued in place.

With the bridge now glued to the sound board, the overall alignment is 'double checked' by again temporarily taping the sound board to the bowl using the string template to confirm precise bridge and nut position relative to the neck joint and instrument centre-line.

Note that about 2 mm. extra length has been left at the nut. This will be trimmed to the exact length later - after the fingerboard has been glued in place.
Set back of the neck at the nut end is about 2 mm to allow for any adjustments to the action (by planing the fingerboard) should this be necessary. (So the fingerboard will initially be about 2mm thicker at the nut end than at the neck joint)

With everything as it should be, the alignment is fixed - in preparation for the sound board to bowl gluing operation - by drilling holes for two wooden pegs (i.e. toothpicks - just under 2 mm in diameter) positioned at the bottom of the bowl and on the neck. The peg hole at the bottom of the bowl will eventually be covered by the edge banding and that on the neck by the fingerboard. This is a method sometimes used by the 16th C lute makers.

Just so that I don't forget, a makers label has been printed (using non historical Microsoft Word and laser printer!) and glued in a position where it will be visible through one of the two sound holes.
Additionally, I have marked the end plate of the bowl with my brand - another practice of some of the 16th C lute makers.

I know of one professional guitar maker who goes so far as to ink stamp all of the sound board braces with his mark - helpful, no doubt, in determining any future unauthorised modifications to his work.

Before gluing the sound board to bowl, both sound board and bowl were 'tap tested' to see if any useful data might be obtained for future reference.
I was unable to determine anything from tap testing the sound board by following the instructions given by Robert Lundberg in "Historical Lute Construction". Lundberg's guidance applies to a lute with typical late 16th C/ early 17th C bracing (treble and bass bar below a bridge designed for more than seven courses) so this may be the reason?
However, Lundberg does admit that his method offered is not definitive and does little to detail the actual procedures that he uses.

In an effort to obtain some measurable data for future reference and comparison, the sound board and bowl were each 'tap tested' and the sound recorded with a Zoom H2 digital recorder. The sound files were then analysed (using 'Audacity software).
The sound board was tapped at the centre of the bridge and the bowl, on the side, at its widest point.
The attached images show the spectrum analysis results. What do they all mean with respect to a completed instrument? I have no idea at this point in time!

A quantity of hide glue will be allowed soak overnight in preparation for gluing the sound board to bowl tomorrow.

In the meantime, I am considering using for this project - in order to save time and progress the work - a set of lute pegs that I turned from Brazilian Rosewood some years ago. The design of the pegs is quite plain - lacking the usual decorative groove(s) or, alternatively collar just below the peg head. So, I thought that it would be interesting to investigate the feasibility of adding a collar to these pegs made from a contrasting material - in this case wire made from the metal tin.
Tin looks like the precious metal silver but does not quickly tarnish like silver. For this reason, it was often used by early civilisations as a substitute for silver for decorative metal coatings and inlays.

Tin wire is readily available from most hardware stores in the form of 'Lead Free' solder - now used in place of lead based solders for applications such as domestic water systems. This material is pure tin except for the addition of a small amount of silver necessary to make the solder flow more easily. Solid wire solder of this kind is usually 2.5 mm or greater in diameter but rosin cored lead free solder is available in smaller diameter down to 1.5 mm. The latter material is the solder used for these trials - the larger diameter solid wire being suitable for 'dot' inlays in the end of a peg head.

Tin wire solder is quite soft and easily cut with a sharp knife and formed by hand. To make a peg collar, the wire is first wrapped around a wooden dowel of suitable diameter - like a spring - and then cut with a fine razor saw to make the rings (an old jewelers trick). The diameter of the individual rings may be further reduced by cutting a little material from the gap and then reshaping the ring on a tapered mandrel (I use the handle of an artists paint brush)

The peg is prepared for accepting the collar (ring) by filing a groove just below the peg head with a small triangular file.
During this operation, the peg is held, for convenience, in a peg shaper (with the blade removed).
Once the ring diameter has been adjusted to the correct size, the collar may be permanently glued in place with epoxy cement.

The method works but my wife - as an independent critic - is not impressed with the result. So, whether or not it will be used for this project remains to be seen!

Out of interest, the assembled oud components have been individually weighed and total weight of bowl, soundboard, peg box and pegs is about 600 grams. The fingerboard, inlays and varnish are not expected add more than an additional 50 grams in total - so this is within target weight expectations.

CORRECTION! the 1.5 mm lead free wire solder used here is not flux cored (although the packaging says that it is) but is a solid wire alloy of 96% pure tin and 4% silver. This product is from the Bernzomatic 'Speciality Solder Kit SSWS 100'.

The sound board has been glued to the bowl and is being left overnight for the glue to fully dry.

Using hot hide glue, the edges of the bowl and ends of the braces were first coated with glue. Next the neck block face was coated with glue and the sound board (correctly centred with the alignment pins) was clamped to the neck block with tape and a cork lined caul held in place with elastic bands to provide pressure to the centre area of the joint. Working around the edge of the sound board - bit by bit - hot glue was worked into the joint surfaces with a thin spatula, the gelled glue liquified with a hot iron and the joint taped in place. At each bar end positions the glue joint was heated with the hot iron applied to the side rib and side pressure applied with a small block of wood taped in place.

With the sound board completely glued and taped, the caul at the neck block was removed and the edges of the joint in that location were reheated and re-taped to ensure a tight joint.

Tomorrow the tape will be removed and the sound board trimmed and finished to size around the edge of the bowl.

Yes, I shall be using the half-binding tool Samir (not the Lee Valley mini rebate plane adaption) but only for the sound board edging as I shall cut the rebate for the banding on straight sides of the fingerboard on a router for convenience. This means that the fingerboard must be prepared prior to gluing to the neck.

I have decided to use boxwood for the fingerboard as this is very hard and close grained (unlike the more open grained Ebony) and - 'off the plane' has a mirror finish. It is a quite a resonant wood too - so might be a good (but costly) choice as keys for an xylophone.
I have two seasoned logs of 30 year old Persian boxwood as well as a quantity of very old boxwood like material in small billets. I was told that this was 'South American boxwood' by the seller, when I bought it over 30 years ago, but he likely did not know exactly what the species was. However, it is very similar in colour, texture and hardness to the Persian boxwood as well as being more consistent in straightness of grain - as can be seen from the attached image (Persian boxwood sample above the 'S.A. boxwood' sample). Note that the core of the Persian boxwood is significantly offset to one side of the log so the wood is potentially unstable Reaction (or Compression - not sure which) wood. This means that this particular piece may be less suitable for fingerboard material - although, given the thinness of the fingerboard - this likely would not be too significant once glued to the more stable Spruce neck.

The fingerboard will not extend from the nut to the neck joint - as appears to be common practice on modern ouds. Following standard procedure for lutes of the 16th/17th C (which may also have reflected earlier oud practice - judging from the old Persian miniature paintings) the sound board will overlap the neck joint providing additional strength and reinforcement to the joint.

I have used 'standard' masking tape for clamping the sound board during the gluing operation - strong enough without the risk of grain 'pull out' when the tape is removed. To minimised this risk, the tape is pulled back on itself at an angle to the wood grain. I should mention that the overhanging edges of the sound board should be rounded off slightly - prior to the gluing operation - to avoid tearing the tape as tension is applied.

The sound board overlap has been trimmed using a sharp chisel followed by a file and curved scraper. Removal of the surplus surface glue enables inspection of the sound board to bowl joint. As a result, a small section (about 5cm long) just below the neck block was found to be unglued so this will be corrected by inserting the spatula into the joint with hot glue, melting the glue in the joint with the hot iron and then re-taping. I am not completely satisfied with the perfection of joint at some of the brace locations but will leave well alone as attempting to reheat the glue with a hot iron to 'improve' matters may weaken the glue (as it is not possible to re-moisten the bar ends).

The sound board has a dip in the centre measuring about
2 mm. The sound board was braced when relative humidity was about 55% so the curvature (from the original flat) is likely due to a combination of shrinkage of the hide glue and the sound board wood across the grain at the current RH of around 45%.
Given this situation, no additional 'dip' was deemed necessary by profiling the sides of the bowl. Lets see how it works out in practice!

The small section of 'dry' joint on the sound board/bowl glue line has been repaired by introducing water into the joint with a thin spatula dipped in hot water. Hot hide glue was then worked in with the spatula, heated with a hot iron, the sound board edge taped down and left for the glue to fully harden.

Having decided to use Rosewood lute pegs for this project, the pegs have been fitted to the peg box. This is more conveniently done before gluing the peg box to neck.
I use three tools for this job - the wood drill used to make the pilot holes, a 1:25 taper reamer (made from a standard Morse taper reamer reground to specification by a local tool maker) and a standard 1:30 taper violin peg reamer.

The pilot holes were drilled in the peg box sides before assembling the peg box. so the alignment of each peg hole was first verified using the pilot drill. Three holes were found to be slightly out of alignment - the remainder being pretty well 'spot on'. The misalignment was corrected by applying side pressure to each reamer as the hole was being cut.

All the pegs are numbered in sequence so that they may be correctly located for final assembly.
As the pegs have been recovered from an earlier lute project, the string holes are already drilled.

The completed pegbox assembly has been 'dry fitted' to the oud to check proportions. The peg box is more compact than that of a modern oud (overall length being about equal to bowl depth) - hence the need for (lute like) closer spaced pegs with relatively small peg heads.

Next to make the fingerboard.

The section of sound board overlapping the neck joint has been inlaid with ebony banding to the full sound board thickness at this location (just under 2 mm). This is for convenience and appearance in blending the sound board edge banding and the fingerboard banding.

The boxwood fingerboard has been planed oversize to about 4 mm maximum thickness so that when finished will taper in thickness from the nut down to 2 mm at the neck joint end. The blank has also been cut to the correct taper of the neck but has been made longer than necessary to allow for final fitting and adjustment to length.

A 4 mm wide rebate for the banding has been cut (for convenience) on a router table. As the boxwood is very hard and brittle, very light depth cuts were necessary to prevent chipping the sides of the blank. About 0.8 mm of material has been left underneath the rebate so that the fingerboard can be fully assembled, trimmed and adjusted in thickness and taper - for correct action - before gluing it to the neck.

The Persian boxwood/Ebony tiles for the edge banding will be made in the same way as the banding for the bridge and peg box (see page 7) so there is no need to repeat the procedure here. The only difference is that the banding for the fingerboard will be made thicker (about 3.5 mm thick).

The peg box has been completed by gluing the finial in place and filing it to blend in with the peg box contours.

To glue the peg box to the neck I have used a hot hide glue with a higher 'bloom strength' than the hide glue used for building and assembling the bowl and sound board. This stronger glue is faster setting so can be used for uncomplicated joints.
Several 'dry' test runs were made to determine the quickest method of clamping the peg box to the neck. In this case I have employed an improvised combination of a spring clamp together with elastic bands using a couple of screws providing temporary anchor points in the neck. (the screw holes will, of course, eventually be hidden by the fingerboard.

The material under the oud is a 'non-slip' rubber carpet underlay to facilitate handling of the oud on the slick kitchen counter top next to the stove and glue pot.

This week end might be an appropriate time to order unbleached bone for the nut and gut for the frets - I need replacement fret-gut for my lutes anyway as the stuff does not last forever.
The gut frets will be double tied (rather than being a single gut strand often used by many modern lutenists, myself included - a cost consideration as fret gut is quite expensive) this being generally the way early lute frets were tied and the oud engraving seems also to represent double strand frets. Greater cost unfortunately but a double strand fret should last longer (i.e.wear better).
In order to keep the string 'action' above the frets as low as possible I shall likely use graduated frets that reduce in diameter from nut to soundboard.

With the peg box now glued in place, it is possible to make a preliminary estimate of the string action in order to finalise the finger board taper.
With a wooden dowel in the position of the nut and a nylon string connected to the bridge, measurements of string clearance at the neck joint and nut can be determined. The string clearance at the neck joint on the treble side measures between 2.5 and
3 mm. However, about 1 mm of the original neck 'set back' at the nut seems to have been 'lost' since assembly of the sound board to the bowl.
Before finally establishing the measured finger board taper from nut to neck joint, the oud has been moved away from the heated kitchen (at 45 to 50% Relative Humidity) to a cooler part of the house to stabilise for a few days at a higher - more normal year round - relative humidity level (60 to 65% RH). This may result in a recovery of the original neck set back as the wood of the oud expands slightly.
In the meantime work on assembling the fingerboard and banding can proceed on the assumption that a maximum thickness of 4 mm for the finger board will be adequate to provide the required taper.