Improving an old Hubbard harpsichord kit
by Claudio Di Veroli, Copyright © 2016
Eight years after producing the webpage The saga of the Hubbard-Di Veroli harpsichord, for the benefit of those who may be about to assemble a Hubbard two-manual French harpsichord kit, or else for those who already have completed it, we have decided to put together in this webpage a list of suggested improvements that we believe are worthwhile. I have carried out most of them successfully in the Hubbard-Di Veroli instrument.
HDK72 - Hubbard's two-manual French harpsichord kit in the version produced in 1972, with case made of birch-faced plywood.
HDV - The Hubbard-Di Veroli harpsichord, based on the HDK72.
HKI - Hubbard's Harpsichord Kit Instructions manual, dated 1970 (supplied with the HDK72).
1. Whom is this webpage useful to
For many years, in spite of a few differences with respect to the original by Pascal Taskin (Paris 1769), Hubbard's two-manual French harpsichord Kit was a regular winner of international harpsichord-kit-making competitions. Even if the heyday of harpsichord kits is long gone, to this day amateurs continue to assemble Hubbard kits, including some that were delivered by Hubbard Harpsichords quite a few years ago. A well known organ maker purchased a Hubbard two-manual kit in the mid 1970's, then stored the crate away hoping one day to have time to devote to it. The time only came when he retired, more than 30 years later. The author inspected the kit shortly after it was completed a few years ago: it was an excellent harpsichord indeed.
If you never had a basic traning in either harpsichord making or woodworking techniques (possibly both), assembling a kit may still be a fascinating adventure, and you may learn a lot from the project. However, you are unlikely to eventually get a good instrument, let alone one that can be played upon, and this webpage will not bring you any significant help.
Photo: Hubert Bédard and the author in Bédard's atélier near the Bois de Boulogne, Paris 1975 =>
Assuming instead that you have the basic skills outlined above, which are needed to peruse to full success the HKI, this webpage will help you in getting a better final instrument by introducing improvements—some of them significant—into the instrument.
If your Hubbard is relatively recent, some of the improvements listed below may be already incorporated into its design and parts. If your kit is as old as mine (1972) chances are that it will benefit from all the improvements described below. It will be obvious, from their descriptions, that most of these changes are better carried out during assembly, but anyway all of them (though obviously with more effort) can be produced after the instrument is finished. The changes are described in the order they fit in a kit assembly project. Only two of the changes below were not carried out in my HDV: no. 2 (too complicated after the construction!), no. 12 (too expensive and risky) and no. 13 (to be carried on in the future).
2. Belly rail treble 4' recess
In a harpsichord the soundboard is installed over a piece of wood glued to the inside of the case: the “liner”. The exception is the front of the soundboard, where it meets the back row of jacks: here the soundboard is glued over the “belly rail”. In order to suitably convert the strings' vibrations into audible sounds, the soundboard bridges have to be free to vibrate, but in the treble end the 4' the strings are so short that, inevitably, the rightmost few inches of the 4' bridge end up glued over the belly rail, which in HDK72 is 3/4in (19mm) thick: the sound of these strings is therefore very significantly dimmed. The solution (not described in HKI) was known in ancient times, and is simple but only practical if carried out before installing the soundboard.
You should just produce a cutout into the upper part of the treble end of the belly rail, reducing there its thickness to only 1/8in. (3mm). In the picture below we have an inner view of the treble end of the upper belly rail before the soundboard is installed: it has only a small recess meant to hold the treble end of the 4' hitchpin rail. The additional recess we suggest is shown by the red line.
The figure below shows the recess in a different view: from above, with soundboard and strings installed. The lines in black, red and blue show what is found below the soundboard:
- The black line shows the uppermost part of the upper belly rail, before any recess is cut.
- The blue line shows the 4' hitchpin rail, glued to the underside of the soundboard.
- The red line shows the recess to be performed in the upper end of the belly rail.
The left part of the recess allows the 4' bridge and its strings to vibrate freely, while the right part (the only one described in the HKI, detail #10) holds the treble end of the 4' hitchpin rail.
3. Music desk
For the music desk the HKI provided two alternative designs: the traditional large frame meant to be stored outside the instrument, and a model that fitted inside but needed to be moved forwards to allow folding it back before closing the lid. In a letter Frank Hubbard kindly praised my design, a third alternative that simply folds forwards towards the player, as seen in the pictures below.
4. Removing leads from the keys
The keyboards included in HDK72 were made in Germany by Kluge. The keys were balanced (unhistorically) using lead weights (two per key), whereby the total mass of a lower natural key was 80 gr. An excess of mass adds inertia, making the action heavier especially in chords and fast passages. In the lower manual I removed one lead from each key and rebalanced it by undercutting the front, as found in ancient instruments, reducing the mass down to 66 gr. per key. An even better (but expensive) solution is to discard the Kluge keyboards altogether and install better ones instead, such as those supplied with the Hubbard two-manual kit in later years. (Have not checked whether Hubbard Harpsichords provides separate keyboards: they are not listed in their extensive spare parts catalogue).
5. Replacing the coupling pads
In the HDK72, when the manuals are coupled, “coupling dogs” glued to the lower keys push wooden buttons screwed to the upper keys. In order to cushion the action, the buttons have 3mm-thick cylindrical cloth pads glued to them. This damping is effective in preventing any action noise: problem is, it is excessive. The upper and lower keys have different rotating centres and therefore the dogs not only push up against the upper buttons, but also slide towards the front. This means that part of the player's energy imparted to a lower key is wasted producing a temporary deformation to the thick felt: the result is a very heavy lower manual coupled action. The solution is to return, as much as the HDK72 design allows, to the traditional arrangement, based on thin leather rather than thick cloth. I followed these steps in February 2015 on the HDV:
- I removed the upper keyboard and removed from it each key.
- I removed the original thick green cloth pads with a razor blade.
- I removed from the buttons any remains of the pads and glue using sandpaper.
- In order to compensate for the thinner leather, one could unscrew the button two full turns, but the screws provided with the HDK72 are too short for this, so I had instead 2mm-thick wooden cylinders made by a top harpsichord technician, Martin Spaink of Amsterdam. He also supplied me the necessary leather.
- I glued these wooden cylinders into the buttons.
- I made 63 leather punches of the same diameter.
- I glued them to the wooden cylinders.
- I lubricated the leather punches with graphite powder. The picture below shows on the left the original thick cloth pad and on the right the new cylinder with leather and graphite.
- Finally I re-regulated the coupling action for proper “staggering” of the three string choirs. With precision weights I measured the reduction in 8'+8' action weight at about 20gr per note.
5a. Further coupling action improvement
Even the improved cylinder buttons shown in the previous section are not an optimal solution: as you press the lower key, its coupler dog not only pushes the upper key, but also slides back and forth over the cylinder's leather. This is due to the geometry of the lower and upper keys: their extremities have different rotation centres and different diameters. Is this problem inevitable? How significant it is? Can it be improved? To resolve these questions I produced with cardboard an exact mock-up replica of the keys' coupling action, with a coupling dog that can be adjusted to different sizes.
With the mock-up I carefully reproduced and measured this back and forth displacement. In my instrument it was 1.3mm. Perhaps this is not a lot, but is certainly not ideal. Is it possible to reduce it? In the mock up I "removed" the button and "simulated" a longer dog pushing directly on the upper key: the displacement is reduced to 0.9mm! This can be further improved, because in HDK72 there is space available to move the dog about 10 mm towards the belly rail, so that a larger dog pushes agains the upper-key recess (below the front 8' jack), and bypasses the upper key altogether when the upper manual is uncoupled: there is ample space for the dog to move up between the front 8' jack and the 4' jack. Now I measured the displacement and it was down to negligible 0.5mm. I am pretty sure the reduced friction from 1.3mm to 0.5mm can be significant. Baroque makers knew what they were doing!
The only question is whether it is worth not having regulating buttons and all the trouble of a new coupling action for each note. In the upper key you have to remove the button and cover with leather the recess below the jack. In the lower key you have to remove the dog, procure/make a much larger dog, plug it further back in the key and file the dog's top for proper coupling adjustment. This for each one of 63 pairs of keys.
6. Piano Stop
Since the upper manual plays only one string choir, there is no need to turn its jacks OFF. This is why in most two-manual harpsichords the upper 8' register is fixed. However, this is by no means necessary: by providing it with a stop lever with a very high lever ratio (see the shortest stop lever in the picture below), we can regulate the action so that, in its forte position, the loudness is identical or better almost imperceptibly softer than the lower 8', while in the piano position the difference is clearly audible. Even in the piano position though, the upper 8' can be used as a solo stop, and it also adds a significant “punch” to the lower 8' when the manuals are coupled.
Needless to say, this “piano stop” is never used as an expressive device during performance, which would be both ridiculous and unhistorical: it is only used to “customise” the instrument in order to fit different repertoires. I use the upper 8' forte for German music, piano for all the others.
The piano stop also serves a different purpose. When weather is extremely dry or wet, the case of any harpsichord undergoes deformation, perhaps minimal but enough to put the action slightly out of regulation. Since a lower 8' jack plucks the string on its left but an upper 8' jack plucks the string on its right, a case deformation makes one 8' stop slightly louder and the other slightly softer. The “piano stop” yields a good mitigation for this situation.
7. Stringing: improving the material
The HDK72 included yellow brass strings for the bass and steel strings for the midrange and treble. These materials were the best available in the early 1970's, but more recently much better strings are available based on historical alloys. A much better yellow brass is now available. Red brass is best to improve fundamentals in the extreme extreme bass range. Parts of the midrange are best strung with low-carbon iron for a better transition between brass and steel. (Note that the HDK72 has a scaling too long for low-carbon iron throughout: the strings would break.) As for how to determine the optimal schedule using the new alloys, see the following.
8. Stringing: improving the schedule
The details are explained in another webpage: Restringing the Hubbard-Di Veroli harpsichord, especially sections 4, 7, 8 and 10. This project was carried out in 2011.
For optimal sound quality and evenness, a schedule should
(a) reproduce the historical string tensions and
(b) smooth-out the transitions between different string materials and sizes (diameters). The HKI suggested to string the kit using a schedule from Hubbard's well-known Harpsichord Regulating and Repairing booklet of 1963. One problem is that this schedule departed significantly from historical schedules: in recent years Hubbard Harpsichords has produced and published improved schedules. However, the real problem is that no ready-made schedule is really optimal, because inevitably bridges and pins are never exactly where they were planned to be: the optimal schedule has to be calculated for every individual instrument using its vibrating string lengths measured between nut pins and bridge pins. See the already-mentioned “Restringing” webpage, from which the graph on the right hand side has been extracted.
The details about formulae and calculations are shown in another webpage, Taskin Harpsichord Scalings and Stringings Revisited, especially sections 4 and 8.
9. The double and singing buff
Most Baroque French harpsichords had no buff batten at all. This feature originally appeared in 17th century Flemish instruments, and was only adopted in France towards the middle of the 18th century. It was usually acting on the upper 8', although a few instruments had the buff batten fitted with pads acting on either 8' choir. Accordingly, back in 1974 we glued to the HDV's buff batten two sets of pads, yielding a quite “pizzicato” buffed upper 8', while the lower 8' has smaller pads and sounds remarkably like a guitar: this effect has been mentioned in forums in 2009 and is justified by the historical variety of buff pads. I have now set up a separate webpage with guidelines for amateurs about how to fit a “singing buff”, from which I have extracted the following photograph showing leather buff pads in the extreme bass and extreme treble of the HDV.
10. “Cathedral effect” and diagonal dampers
The HDK72 design assumes “flag dampers”: when a key is not played, its jacks do not rest on the key but hang from the damper. This prevents the string from vibrating when not plucked. The same happened in historical virginals, spinets and Italian harpsichords: most of them had two 8' choirs and no means to disengage any of them. However, other harpsichord-making traditions used different arrangements. The historical French two-manual instrument has “diagonal dampers”: if the stop is disengaged, the strings are free to vibrate sympathetically when other stops are played. This “cathedral effect” is clearly audible, especially in small rooms.
If one has built a Hubbard two-manual kit with flag dampers, the temptation is to convert each damper to a diagonal one. In the upper 8' stop does not make any sense: it is always on, and therefore it can not provide any cathedral effect (unless one installs a special stop knob specifically for this purpose of course: note that this requires a significantly different lever ratio than a “piano stop” lever). As for the lower 8' stop, the conversion to diagonal dampers is not that useful either, because this stop is seldom disengaged: the upper 8' + 4' combination is very rarely used, and the solo 4' with its tiny-mass strings has already an audible cathedral effect from all those 8' strings even if damped. However, converting to diagonal the 4' dampers is relatively simple and very effective, and the effect is clearly audible when playing solo any of the 8' stops. This 4' damper conversion hardly affects the regulation at all, provided that (in agreement with Hubbard's directions), with all the stops and the coupler ON, the 4' is the first choir that speaks in the “staggering sequence”. A full discussion about dampers and the cathedral effect is included in the book Playing the Baroque Harpsichord, section 6.2 and Appendix J.
11. Replacing the lower key Philips screws
Historically the length of a wooden jack was regulated by gluing paper pads to either the jack's bottom or the key top. The HDK72 used more convenient screws: the upper 8' jacks were provided with a hole in the bottom in order to fit there the jack-length screws.
Unfortunately, for whichever reason, the arrangement of lower-manual jacks in HDK72 was completely different (please note that Hubbard Harpsichords changed the jack model after 1974). Philips regulating screws were meant to be inserted at the end of the lower keys, exactly below the jacks. This was hardly satisfactory because a very long thin Philips screwdriver (included in the HDK72) was needed to regulate the screws. Even worse, felt pads were meant to be glued to the screw head to support the jacks, but the pads had central holes for the screwdriver: this made the action inaccurate.
The solution was easy: I did not use the Philips key screws at all. Instead, I drilled each one of the 126 lower manual jacks (easier said than done as shown in the picture below) and installed into them jack-length screws (similarly to the upper 8' jacks arrangement), which I got from Hubbard Harpsichords. The picture below shows me drilling the jacks (the headphones are meant to help with the very strident noise produced by the operation!).
12. Fitting traditional wooden jacks
The HDK72 had jacks made of a material that has been named “black Delrin”, and a new white Delrin model was provided after 1974. [Hubbard jacks have been criticised for having regulating screws, but this in itself is not necessarily detrimental to an effective action and certainly not to sound]. Compared with the almost perfectly rigid traditional jack made of pearwood or similar fruit-tree wood, the Delrin jack is as effective and accurate. The problem is that, as the plectrum bends against the string, the jack body also bends a tiny fraction of a millimetre towards the string. The total effect on action weight is hardly significant, but there is no discussion that, for the discerning player, wooden jacks have a slighly better “feeling” and are therefore recommended. Unfortunately, making and fitting wooden jacks is beyond the capabilities of an amateur or even the average harpsichord tuner. This change therefore requires a professional harpsichord maker and is quite expensive. Another problem is that the HDK72 gap is too narrow to fit registers and jacks with historical width. The HDK72 jack is 9.5mm wide (from front to back) and I have a 10mm-wide test wooden jack: it works fine, but it is visibly too narrow and fragile, and it is unlikely to withstand decades without significant deformation. With the above considerations in mind, I have discarded any plans to replace the original jacks in my HDV: to all practical purposes they are working as well today as more than 40 years ago when the instrument was premiered.
13. Quilling with feather and bristle springs
Many players nowadays prefer plectra made of the traditional quill (historically raven was preferred, but Canada goose and condor provide the same effect and are longer-lasting). Unfortunately, the Hubbard tongues were bored for flat Delrin, and require re-boring for quill. This is not an easy operation, and can easily break the tongue: ensure you have enough spares before starting the project!
Given the known issues with the self-springing Hubbard tongues, which sometimes require maintenance, a better solution is to replace the tongues as well. This is however complicated because:
- Traditional tongues are stopped in the bottom while Hubbard tongues are stopped by the upper screw: you need especially-made tongues and possibly some reworking of the slots in the jacks.
- It is not easy to bore into the Delrin jack the small holes needed to insert the bristle spring.
14. Alternative damper fitting
Regardless of the damper shape discussed above in no. 10, the HDK72 fits the dampers by gluing small blocks of cloth into a damper support: this is a small piece of Delrin with a slot, where a screw mounts the thing onto the jack. This damper+slot+screw arrangement allows later, when the damper becomes worn, to lower it as needed (needless to say, the same can be achieved in the traditional slot-in-the-jack of historical harpsichords!). Problem is, glued-in dampers depend on the durability of the glue. As recommended by Hubbard we used contact cement, which it worked to a charm ... for 30 years. Then I started to notice poor fast repetition in the 8' jacks (much less so in the 4' jacks where the dampers are significantly smaller), in spite of tongues and plectra being fine. Dampers also “looked” fine, but they were not: they were no longer glued in their lowest end, and therefore moved in-and-out horizontally as they came into contact with the string. This lateral movement was a barely visible fraction of a millimetre, but enough to interfere with quick repetition. Mitigation: a drop of glue with a toothpick. Solution: replace with a new damper. Effective for a further 30 years perhaps ...
For the 8' jacks I suggest using the same material in an alternative arrangement: the drawing shows the jacks viewed from above. Instead of gluing the damper cloth into the support, the support should first be screwed-in to the jack but with the flat part reversed (i.e. against the back of the jack). The damper cloth is no longer glued, but hold between the support and the jack. For this new arrangement you need:
- Longer dampers because they now go the full width of the jack.
- Longer screws because their thread now includes the damper cloth: instead of the original 2.7 millimetre-long thread you need anything between 5 and 6 millimetres.
- A tool to punch the damper cloth where the screw goes through.
Note that with the damper support reversed, the jack is more than 1 mm thicker (from left to right) in its top. This is not an issue with 8' jacks, because even in the OFF position there is usually at least 2 mm of space between the jack and the other 8' string. However, in this harpsichord model the 4' strings run offset with respect to the back 8' strings: as a consequence (in spite of the plectra being a slightly shorter) the 4' jacks are nearer to the front 8' strings , and therefore the reversed damper support may interfere with them when the 4' is OFF. Luckily, the issue of damper ungluing is much less frequent in the 4', where the string vibrations have much less amplitude and the dampers are shorter: they are thus subject to much lesser tension.
15. False lid and flap
When moving the instrument, unless it is kept horizontal, the lid and flap are best removed or else their weight may damage the spine where the hinges are screwed. Problem is, without lid and flap, the thin harpsichord strings can easily be damaged in transport. Therefore, for removals, I fit a “false lid” that covers the whole instrument: it can be made of any cheap material, with added felt pads to avoid scratching the rim of the case. It can have hinges in the middle, so that it can be stored folded.