Piano World Home Page Forums Our Most Popular Forums Piano Tuner-Technicians Forum Sanding hammers - weird question

I am making my own digital hybrid piano, using a grand action. Similar to what you can see at



and at



More of the former than the latter. A key difference is that I am using sensors at the hammer, rather than the shank (like the former) or key (like the latter). Well, I'll have also key sensors for the note/off, located at the back action, but here the question is for the hammers and hammer's sensors.

The mechanics of installing hammer sensors will be extremely simplified if all the hammers were originally striking strings at the same height, which obviously is not the case since the bass strings are higher in modern pianos (including the one I salvaged the action from). However I am thinking of sanding the hammers in the bass, enough to make them "shorter" and hence maintain the same blow distance from equally installed sensors, without need for unequal assembly hardware. I'd then regulate the let off and the backcheck distance to be the same, as necessary.

From the little piano regulation things that I know, this approach seems doable, but before I embark into that much sanding, am I overlooking a showstopper or something else that would make it more complicated than it appears to me?

If it sounds like a reasonable approach for this purpose, any suggestion for not spending too much time sanding? Obviously I'd do gang-sanding, and I could use power tools which I have, but I fear breaking the shanks or sanding too much and having to replace a few (if not all) hammers, which I really don't want to do.

Thanks

If you remove enough felt from the top of the bass hammers to reduce them to the same dimension above the shank as the treble hammers; the touch resistance in the bass will be reduced significantly. You could add a small copper wire to the shank on those hammers to compensate and glue it with E6000.

Thank Ed, good point. Adding some weight on the shanks sounds like an easy thing to do. The copper wire sound like a great idea.

Much more complicated (and I am sure I won't go that route with my limited experience), but just for the sake of the discussion.... Could removing some lead from the keys to maintain the same touchweight be better, given the reduced inertia? Or will be weird if done only on a section rather than the whole keyboard?

Forgive me if I'm missing something obvious but why not just regulate the hammer rest rail lower and reset the pilots to bring the hammers down? Then no need to butcher anything...?
Nick

Thanks for that suggestion. I thought about that and here is what went in my mind at the time.

On this action the rest rail is shared between the bass and the other registers (perhaps that's common), so I can't just lower it: I'd have to make it different between these two sections, by either replacing the rail with two pieces of different thickness (tricky to attach to the common post, but doable with some chiseling) or by adding more felt or other shims on the higher registers. I mean, not a big deal, but not "just regulate the rest rail lower".

The main reason why I discarded that option is because I thought that after "lowering the rest rail for bass only" as above, I would have needed some major regulation (which I've learned to do, but a bit sloppily and very slowly). And potentially be outside of the allowed range for some of these regulations. The reason I say that is because whippen and hammer shanks in bass do not look to be regulated differently than tenor/treble registers. It's just the hammers that are attached higher to the shank. So I thought: if I lower the shank to lower the hammer, then I have to lower the whippen, and hence the capstan, and hence the escapement and probably more. And how do I know that for example the whippen would not hit its rail if I attempt to regulate it so low? Are you saying that I would not need to regulate anything else besides the rest rail and the pilots?

I'm not sure what it entails, but can you not split your hammer sensor rail into two sections?

I'm sorry but I don't see your project going very far with your current understanding of grand action mechanics, it also looks a little like you are lacking practical skills and/or tools to do that, I don't want to discourage you, but you really should first understand physics of what you working with very well and be fluent at regulation. I'm talking from my own experience, I was very naive and very wrong when I first started working on piano actions, I'm robotics engineer so I know both "worlds" now and I'm pretty sure that separate bass rail for sensors IS the best idea, Ed's solution would work but it's just more and unnecesarily complicated, alternatively you can place the sensors on the shanks, they are on the same level.

Maybe show us your design here, we'll see if it's feasible or if there are any major flaws.

Of course I can, but it becomes very complicated. My action is housed inside a upright-looking cabinet (rather than "naked" like the two I posted). The rail is an aluminum extrusion like this: https://us.misumi-ec.com/vona2/detail/110302684530/
Having the hammers all on the same height, allows a single rail with single easy height adjustments.

Having two rails differing by less than the width of the rail (which must be, unless I use rails too thin which will flex) makes it impossible to attach them with each other, unless I design and build a special way to attach them, in a machine shop (which I can do, but it's a can of worms). So the two rails "must" be separate (unlike the one from living pianos, but for him it's just a stop rail, not the place where to attach sensors, so he doesn't need much out of that rail). My rails cannot be attached at the action at the bass/tenor split from the action, because there is not enough horizontal space (I might be able to fit one, but no way I can fit two of them). They cannot be attached from the top, because I want/need a lid that can be opened. They could be attached from the front (where the pinblock would be in a acoustic grand), but I'd have to build "something" for them to "hang" on. Or they can be attached from behind. Either front or behind (or both) makes the attachment geometrically much more complicated to build, and the whole assembly more prone to flex. Should I go behind, I need to decide if the attachment should go to the back of the cabinet (which is not structural, so I'd have to change it and make it so) or to the keybed (where it would interfere with the back action, which is simplified compared to an actual grand, but still in the way). I will also need to calculate the strength of the design to make sure it would not move too much under hammer hits. The complicated geometry makes it more complicated analysis, and so it "nudges" me to "overkill" and build it stronger than it needs to be, which makes it even more complicated. I know this is mostly hand waving, but hopefully enough to satisfy your curiosity.

Sanding the hammers and adding the lost weight to the shanks is much easier than all of this. Assuming I can sand them at the right height or close enough to regulate them at the same height as the higher registers. And once I do that, a straight rail with simple supports is very easy to model, design and "build" is simply cut it to length. It will easily have the needed strength and flex (i.e. unnoticeable flex with the forces in questions). So that's why I plan to do it that way.

PS: LinkedIn tells me that your "profile is not available" perhaps you have a typo in your signature?

Thanks for being frank, I appreciate that. While I definitely lack the practical skills, since I have not done this as a job, I have certainly done lots of regulations on this one grand action following the Reblitz and the Igrec's books which I bought. If necessary, I will also be happy to hire a piano technician to do the final regulation, provided I don't make some huge mistakes which can't be fixed or are too expensive to fix (which is the reason for me asking here). So I'd appreciate it even more if you'd substantiate your comments with more specifics. Why Ed's solution is "complicated"? Sanding hammers and adding weight may be tedious and perhaps easy to get it wrong (is it?), but I don't see anything complicated. What do you think is complicated about it?

Also, can you elaborate why separate rails is "best"? Sure splitting rails is easy, just cut them. But they need to be solidly held in place, that's the problem. If they magically levitate and stay put where I need them, sure, that would be best. How would you attach them with all the areas where you'd place posts or other attachment devices is already taken by the action itself and the back action? I can do cantilever from the front or the back (as I wrote in my previous message) but that's more complicated for me than sanding the hammers. Especially because the rails will need to have some height adjustment, there is no chance that one would design and assemble something with perfect tolerances (the sensor have a limited range and it must be maximized for best output). As a robotics engineer you may have better ideas here and I'll appreciate them! I simply don't see anything easy to design and build which is also sturdy and adjustable like a single rail (see below for my "design" for that).

Measuring the shanks rather than the hammers is obviously an option (and what CyberGene did in the first video I posted above) and in fact I have not ruled out that option (yet). However it will cause the action to be stuck inside the cabinet (like if somebody installed a pinblock going too low with the action inside) and that would complicate regulation too. Not to mention other details, such as lots of other electronics boards to control the sensors and hence need to be nearby. I plan to install them in the area where you'd put the shank sensors instead. In this case they would need to be somewhere else, and there isn't a lot of space available, unless one stacks things on several levels and make servicing of both the action and the electronics even more difficult. For example, in the first video you see that CyberGene have them in front, blocking the access to the letoff regulation and complicating the access to the capstan. Having the hammer stop rail as being the same as the sensor rail removes one item (so that already reduce the volume of "stuff" to put inside the instrument) and solves all of these access issues, since the area where the pinblock would be in an acoustic is available for putting all those boards: easily accessible from the top to service them (like the tuning pins in an acoustic) and not blocking anything in the action for serving that. And close to the sensors as they need to be.

Regarding "show us the design", it's super simple. If I can have the hammers all at the same height, it's simply a misumi extrusion attached from the sides with L-brackets to another thinner extrusion attached to the structural part of the cabinet. The L-brackets allow final height adjustment.
If the hammers are at different heights, I have some ideas (mentioned in the previous message) but they are all unsatisfactory, so I have nothing to show at the moment.

Great you have those books, I don't know your space constrains but separate rails doesn't have to mean that main structure is separate, you can have one rail and just a spacer under tenor-treble section for sensor mounting, it's only couple of milimeters. What sensors you want to use? Are you aware of hammer speeds and at what distances you have to measure them? For example that you can't have first measurement point below drop distance (usually 3mm), and best would be at or after letoff (usually 1.5mm) so second point has to be almost at "string" contact and those points has to be stable. What level of performance you want to achieve? Is your action in decent condition? Is your action weighted properly and hammer strike weights are correct for action ratio?

Nice to brainstorm this with you. Thanks for your feedback.

For this action (*) it's about 7mm. And that is the problem: the difference in height between hammers is too thick for a thin spacer and too thin for a whole rail. If for the sake of the argument that difference were 20mm (or perhaps even 12-13mm), I could attach one rail to the other and be done with it. Moreover, I can't install just a single long spacer, because that will close the slot of the rail which is where the sensorboards will be attached. See https://us.misumi-ec.com/vona2/detail/110302684530/ if you haven't already and this should become clear if it isn't. So I need to use individual spacers. Something like thick washers, perhaps nylon ones to not increase the load excessively (7mm is a heck of a thickness for a metal washer and I'd need dozens and dozens of them). I initially discarded this option because I did not find a suitable washer with the ID, OD and thickness I need. Plus I was not sure if that much of thickness would make the attachment less secure. But I'll follow your advice and test this option. I could investigate 3D printing the spacers if I still do not find anything suitable off-the-shelf.

Optical. Either Vishay CNY70 (more likely) or Everlight EAITRCA6. They can measure the distance of an approaching object in a range of distance from 8-10mm at the farthest to about 1mm at the closest.

Yes, I plan to place the sensor rail at a height in which the sensors themselves would be 1-3mm above the level where the strings would be. Why not "exactly the same level as the strings"? To account for the yield of the strings when hit (which this stop rail will not have) and for the fact that the sensor cannot measure the last 1mm of travel. If having letoff so distant from the actual point where the hammer stops will feel weird, I can put something in front of the sensors (with a small hole obviously).
From what you say it seems to me that you have in mind the typical mechanical switches in which the scan is at fixed positions and the time is measured to compute speed. This will be a bit different: as I said the sensors measure distance, so they are not triggered by a particular position, and instead they scan the distance at fixed times (to allow velocities dramatically different, repeated time-scans will be used and summed, so not strictly "fixed", but multiples of the shortest possible scanning time). I plan to ignore the distance measurements outside of the last couple of mm from from "the strings". I know that ideally I should ignore anything before letoff, but I am not sure yet if that is possible with the time and distance resolutions I have -- but I know that I'll be close.


Well, ideally I want performance similar to a decent/good old grand piano. I certainly can't expect a top-notch Renner or WNG both because of the action itself, but also because of the sensors and all related things. The action is in decent/good condition (*) and the touchweight analysis as described by Ingrec (which I have done for only a couple of keys per octave) is not bad either.

(*): this action is the original one which was inside a 1934 Mason & Hamlin A, whose plate broke (for reasons unknown to me). The owner, unable to sell the instrument with the broken plate, and not willing to put money on that repair, decided to trash the whole piano. I had the opportunity to purchase the action (including the lyre and pedals) and parts of the cabinet, including the keybed, at an affordable price. Before the incident, the instrument was in fine conditions.

Sure. I could think of one or two approaches to attempt a self-supporting, sufficiently rigid yet adjustable/shimmable joint that has about 7 mm offset, especially if there are machine nuts that can be fitted into the extrusion slots, but there are probably equally good counter-arguments.


You're right, it was outdated. I've removed it, as it stemmed from a time when there was lots of tension on the forum and even I as an amateur / part-timer / hobbyist got sucked up into the demand to provide disclosure. I've removed the link, but if you're interested, I can send it to you in a PM.

So basically CyberGene design, still you will calculate hammer velocity measuring time and/or distance between two points, second point theoretically should be at "string" contact but practically you need some margin, so if you want for example target regulation like this: 45mm blow distance, 3mm drop, 1.5mm letoff I think you should do something like that: 46mm blow distance, 4mm drop, 2.5mm letoff and second measurement point at 1mm (all distances from hammer stop rail) so basically your hammer stop rail will be just 1mm higher than it should be, this will not cause any problems with functioning and performance of the action and you will have margin for the sensing distance which is not really very stable for this sensor type (will fluctuate a bit with changing ambient light, changing color of the measuring surface (hammers will yellow over time), fluctuations of sensor diode supply current, errors of comparator/adc and degradation of the sensor diode over a long period of time)

Yes, more or less that's it. Unlike CyberGene's Cybrid I have a closed enclosure, which I plan to make it somewhat lightproof (within reason). Like him, my hammers are old and perhaps already as yellowed as they could be. I'll be ADCs, not comparators, so that should be a bit more stable. I also hope to use a super-stable supply current, again within reason, so some residual fluctuations will be there. The the rest that you mention applies.

Have you already built one? Interested in collaborating?