I finished making my fourth guitar just a couple of days before Christmas. It is, in many respects, very much like the previous two. It’s based on the same Barbero plan, has Cypress back and sides, and this time a European spruce top and walnut for the bridge, head veneer and bindings. It’s 650mm scale rather than 660mm. I built it for a friend, Andy, who heard guitar #3 and wanted one like it. The only instructions I received from Andy were to use mechanical pegs (I opted for Wittner pegs) and that he liked the square head shape of my previous two guitars.
Overall, I am pleased with the outcome as it is very consistent with guitar #3. The air resonance of #4 is 90.3Hz (F#-42 cents) compared to 90.2Hz (F#-44 cents) of #3, and the weight is 1247g compared to 1189g of #3. The mechanical tuners weigh an additional 34g over the ebony turners used on #3, so taking that into account, the overall total weight difference is just 24g.
I used hide glue for the first time which was difficult at first but became easier as I got used to it. It gels very quickly and needs to stay quite runny (i.e. warm) to achieve coverage over large areas. I didn’t feel confident enough using it to joint the top and back, nor to glue the bridge with it after having french polished the guitar, but I’ll definitely continue to use it for certain tasks such as making the rosette and gluing the linings, purfling and binding, when the quick grab time is useful.
My workmanship is slowly improving. Throughout the build process, I’m often reminded of David Pye’s concept of the workmanship of risk vs. workmanship of certainty, whereby the use of jigs and machine tools can increase the the certainty of the outcome, compared to workmanship that relies more on hand tools, individual judgement and the maker’s skill and is therefore more risky. I work in a very free manner but a few more jigs and templates would be helpful and improve the accuracy of my work. When I interviewed luthiers in their workshops during the course of my research, it was very common to see a variety of jigs and a small number of machine tools. Machine tools are mainly used to reduce the labour required, whereas jigs improve the accuracy and consistency of the work. One of the reasons apprenticeships in lutherie are so rare is that machines have replaced the labour that apprentices used to be employed for. The labour time that I put into an instrument is of little consequence because I’m not trying to make money and I enjoy the physical and leisurely pace of work, but I do want to make some more jigs before I make the next guitar (which will also be a flamenco guitar for another Andy).
There was a point while making this guitar that I felt like I was achieving more autonomy in my work. I wasn’t constantly referring to the DIY books or previous notes quite so much and I am beginning to intuit what comes next in the process; not entirely – not like the luthiers I spoke to who have all the measurements in their head – but I’m experiencing a growing sense of taking the lead, rather than being led. As someone who only makes a couple of instruments a year, I think it will be a while before I fully embody the process of making.
My third guitar is a 660mm scale flamenco blanca and like the last one, it’s based on Courtnall’s Barbero flamenco plan, with an eye on Courtnall’s 1933 Santos Hernandez plan and Brune’s 1951 Barbero plan, too. Like the previous two, it took me about six months to complete during evenings, weekends and holidays. I was asked to make it for Rob Johns, my guitar teacher, who liked the previous instrument that I had made for myself. As it happened, I loaned him #2 for much of the time I was working on #3, so it was nice to deliver his guitar and have my own back.
The back and sides are cypress, the top is engelmann spruce, the neck is Spanish cedar with a rocklite fingerboard, and the bridge is Madagascan rosewood. It differs slightly from #2 in that the back braces and end block are spruce, rather than Spanish cedar and the previous fingerboard was rosewood. I tried to advance my decorative skills a little more with #3 by making and inlaying a red/green/white purfling and back inlay that carries onto the heel. The rosette is a copy of a Barbero rosette. Overall, the workmanship is an improvement on the last guitar although not at the level I am satisfied with.
The sound of the instrument will change as it is played in, but first impressions are very satisfying. My hope is to make flamenco guitars that have similar characteristics to the sound of this wonderful Siguirilla, played by Paco del Gastor. Turn up your speakers and listen to the first minute (or more!)1
My guitar has a very punchy ‘cut’, decent bass and clear trebles, and just a little sustain – more sustain than the previous guitar. Notably, the instrument is very loud. I’ll try to get a recording of it after it’s had a few weeks of being played.
The overall weight of the instrument is 1189g compared to 1119g for #2. I deliberately chose a pale, lightweight piece of cedar (with graphite rod and beech peg bushings) for the neck (448kg/m3) and adjusted the thickness of the back, sides and top according to the % difference in density compared to #2, but it still ended up being 70g heavier. The rocklite fretboard (801kg/m3) is around the same density of some rosewoods. The back and sides (626kg/m3) are 1.8mm and 1.6mm respectively, and the top (424kg/m3) is 2.2mm around the lower bout gradually moving to 2.5mm in the upper bout and 3mm around the soundhole. The very low profile rosewood bridge (980kg/m3) weighed 16g with the bone on the tieblock attached. The bridge measures 6.1mm in front of the saddle with 1.7mm wings.
My obsession with the weight of the instrument is partly because of the success of the previous guitar in terms of how it sounds and the weight of that instrument was influenced by an observation in Gore & Gilet’s books on the characteristics of classic sounding flamenco guitars, which they observed were around 1100g. My obsession with weight is also because without the cumulative intuitive knowledge, based on the experience of making dozens of instruments, I’m relying heavily on the basic material properties of the wood that I can easily record and adjust. For my next guitar, I intend to measure the stiffness of the wood as well as the density.
The body resonance of the previous guitar is 92.189Hz (F#2 -6 cents) and a top resonance of 191.78 (G3 – 38 cents). The new guitar has a resonance of 90.17Hz (F#2 -44 cents) and the top is 191.78Hz (G3 -38 cents). Almost identical. The body resonance on #3 guitar is in the middle of two fretted notes (F and F#) which is the ideal place for it to be so as to avoid ‘wolf’ notes.
Action and saddle height
In addition to the weight, another technical consideration was to improve on the action for this guitar because I ended up re-fretting guitar #2 to eventually achieve what I wanted. First, I looked for what data I could find and compiled a spreadsheet of 55 flamenco guitars, recording the saddle height and action at the 12th fret. The data came from the Solera Flamenca website. Here’s a summary of the data:
Mean action and saddle height: 2.66mm and 7.78mm
Median action and saddle height: 2.6mm and 7.8mm
Mode action and saddle height: 3mm and 7.6mm
I realise that the data on each guitar is not necessarily how it left the workshop, but reflects the set up by the seller. I’m also assuming the data is accurate because people are spending large sums of money to buy instruments online and unseen, so when they receive them, the set up needs to be as described. I believe the action on that website is recorded with the bass E string.
Data on the weight of those guitars is not often recorded but looking at the weight of 16 cypress guitars from famous makers, the numbers suggest that Gore and Gilet’s sample of six guitars of around 1100g were slightly lighter than average:
Mean = 1200g
Median = 1210g
Mode = 1290g
The scale length across the 55 instruments is consistent:
Mean = 655mm
Median = 655mm
Mode = 655mm
The nut width is pretty consistent, too:
Mean = 52.9mm
Median = 53mm
Mode = 52mm
I was interested to see whether there was anything characteristic about the set up of older instruments.
For the seven pre-WWII guitars, the average action is 2.5 and saddle height is 7.6mm.
For 28 guitars up to 1969, the average action is 2.6 and saddle height is 7.8mm.
For 13 guitars made after 1980, the average action is 2.8 and saddle height is 8.1mm.
Again, the action and saddle height may have been adjusted and it’s quite possible that the nut and saddle have been replaced on some of the older instruments, but there does seem to be a trend towards a slightly higher saddle and action.
In the journal, American Lutherie, Richard Brune discusses flamenco setup and writes:
“Originally, all Andalusian Torres models had a low action, about 1/8″ or considerably less between the 12th fret and the string. In the 20th century beginning around the 1930s and continuing after WWII, due to the influence of Segovia and other classical players who were playing larger halls, the standard ‘classical’ action began to increase, culminating in the impossibly high actions seen on Ramirez instruments from the 1960s to the present. However, action is always a matter of individual preference. Many modern flamenco players use a very high action, as did Ramon Montoya for precisely the same reasons, to avoid buzzing. More important is the consideration of neck angle and setup with negative, neutral, or positive relief (back bow, straight, or up bow) which greatly affect the ‘percussive’ quality of the traditional flamenco guitar. Corollary to this is the height of the strings above the soundboard at the bridge, which again was always very low for all Spanish instruments until the advent of nylon strings in the 1950s. This relates to torque on the top, which directly relates to how thin the top can be made, and how lightly it can be braced. Again, traditionally the Spanish guitars were all very lightly constructed, but since the 1930s, beginning with Hauser, classical guitars have gotten thicker and more substantial, with higher actions, taller bridges, and heavier bracing.”
Brune, R. (2000) American Lutherie #61.
So, with all of this in mind, I wanted to more accurately design the action and saddle height into the build of guitar #3, aiming for a saddle height of 7mm and action at the 12th fret of 2.5mm, which I know Rob would appreciate. Just as importantly, I wanted the string height at the 1st fret to be as low as possible, too, without the open strings buzzing. This involved setting a neck angle of 1.5mm into the solera, tapering the underside of the fingerboard slightly between the 12-19th frets to fit the angle, and gluing the slotted fingerboard onto the neck. I then made and temporarily fitted the finished bridge, nut and saddle and planed the top of the fingerboard until a straight edge was 3.5mm above the surface, which would result in action of 2.5mm above 1mm fretwire. I also had to take into account the 2mm dome of the soundboard (which collapses to under 1mm without string tension using my chosen bracing pattern) and anticipate the amount the strings would pull the soundboard upwards. This put me in the region of where I wanted to be so I had enough room to make adjustments to the saddle and nut when the instrument was finally strung up. The finished instrument has a saddle of 7mm and action of 2.5mm, as I had planned.
Although a seemingly trivial thing, the golpeador or tap plate is something that I’ve spent a long time trying to figure out a method for fitting. There are basically three methods: using self-adhesive plastic sheets from luthier suppliers; using blank plastic sheet (0.15mm) from craft suppliers, cutting it to size and shape and gluing with white PVA glue; or using epoxy rather than PVA. I’ve tried the PVA and epoxy methods before on mine and other people’s guitars and prefer the epoxy method that Aaron Green outlined. The PVA never dried thoroughly for me, whereas the West System epoxy bonds slow enough to get all the air bubbles out and position the plastic perfectly, and then dries hard to produce a nice glassy tap. I like this approach, also, because it feels like the fitting of the golpeador is part of the build process – something I make – rather than an accessory that I fit. Incidentally, gluing the golpeador didn’t change the overall air resonance of the guitar, whereas gluing on the bridge lowered it by about 70 cents.
The next guitar
My next guitar will be another flamenco blanca but this time using non-tropical and local wood. Flamenco guitars are traditionally made from cypress because, as Richard Brune argues, it is the only suitable tonewood native to Spain and was therefore cheaper to produce instruments from. In the mid-to-late 19th century, when the Spanish guitar design was established by Antonio de Torres, most players were Gypsy flamenco players who required cheap, loud, percussive guitars to play in the cafes. The flamenco guitar is therefore basically a folk instrument and I want to see what I can produce that is faithful to my own context. I also want to see what a Western red cedar top will sound like using the same Barbero design. The back and sides will be aromatic Cedar of Lebanon, which is light and strong and looks similar to cypress (the set I have is 503kg/m3 and comes from a tree cut in the UK). The neck will be lime from the UK (545kg/m3), the fretboard will be English bog oak (922kg/m3), and the bridge (630kg/m3), bindings and head veneer will be English walnut. The WRC top I have is 347kg/m3. I’ll report back in 6 months!
I recently finished making my second guitar, a ‘flamenco blanca’, or a traditional, lightweight, nylon-stringed guitar made from Cypress, Spruce and Cedar. It largely follows Courtnall’s 1950 Barbero plan, but with an eye on Courtnall’s 1933 Santos Hernandez plan and Brune’s 1951 Barbero plan, too. Studying the three plans offered a great deal of information about what makes a traditional flamenco guitar sound the way it does and the different design choices available to the maker.
I also studied Gore and Gilet’s books for any references to flamenco guitar design and found their analysis of six vintage flamenco guitars to be extremely useful. It gave me a couple of key design goals: to build an instrument that weighs around 1100g and with air and top resonances within the ranges they identify.
Finally, I wanted the sound of the instrument to be very dry and percussive with a loud, short, ‘cut’ when appropriately played. I also like to hear the strings buzz under the control of the player. This video of luthier, Richard Brune, playing a 1970 Manuel Reyes guitar has been a constant source of inspiration:
I built this guitar for myself, wanting to learn to play flamenco but not having an instrument designed for that style of playing or sound. The first guitar I made, a contemporary classical design, is completely unsuited to flamenco. As Gore and Gilet write in their book:
“The flamenco ‘cut’ is provided by the high mobility of the top, which contributes to high excursions of the soundboard and the necessarily low sustain relies on air damping and energy diffusing into the low mass sides and back via an impedance matched top and sides junction; exactly the opposite of our objective for all other types of guitar.” (p.1-91)
Based on their analysis of six vintage instruments, Gore and Gilet suggest that a good traditional sounding flamenco guitar will have an air resonance below 100Hz and a top resonance of around 180Hz (between F and F#) or around 190Hz (between F# and G). They suggest a bridge of no more than 15g, back and sides that are no thicker than 2mm and an overall weight of around 1100g.
Throughout my time making this guitar, I kept an eye on the weight and density of every piece of wood, as well as the overall weight as it came together. The soundboard is made from Engelmann Spruce, thinned to 2.5mm in the lower bout area. The bridge is Madagascan Rosewood, weighing 15g, the Cypress back and sides are 2mm and 1.5mm respectively. The beautifully figured fretboard, head veneer, binding and pegs are all rosewood. The finished instrument with strings and golpeador weighs 1119g, has an air resonance of 88.15Hz (F2 +16.62 cents) and a top resonance of 189.76 (F#3 +44 cents). 1 Here’s a graph of the tap tone data:
And here’s the finished instrument.
I’m pleased with the outcome and the instrument sounds as I had hoped it would. Of course, the workmanship could be improved and I made a number of small mistakes along the way that needed rectifying, especially early on. The main thing I wish I had rectified in the build process is the saddle height. I had drawn out the guitar beforehand to determine the string height and neck angle and opted for a flat (no angle) neck, a scale length of 660mm and no fretboard tapering. I calculated that I could get a saddle height of 9mm and action at the 12th fret of 2.25mm.2 However, the finished setup is a saddle height of 10mm and an action of 2.25mm.3
The whole experience made me realise that although it was my second guitar, it was in fact the first guitar I was making on my own without direction and supervision from a teacher. In craft work like this, the experience of having to think for myself, face my own errors of judgement or execution and learn how to correct or live with my mistakes, was part of the process of learning, as well as a useful source of humility. Even with all of the detailed plans and scientific information at hand, I still had to rely on my own intuition and relatively inexperienced feel for the wood and tools. Fortunately, I’ve been asked to make the same guitar for someone else and so have the chance to improve on the same design.