Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards Escapement and double escapement in digital piano actions

Ah, yes. I never tested to see whether or not that is genuinely easier on a PX-3 than on other (2-sensor) DPs, but at least in theory, I could see where that should be a benefit.


Yes... though that also depends on whether the tone generator being triggered includes the kind of string resonance feature that actually does alter the sound of a key depending on what other keys (if any) have been depressed and have had their virtual dampers raised. AFAIK, the PX3 does not have that feature, it only has "pedal down" string resonance, not "pedal up, some keys down" resonance, which is what you would need for this to create an audible difference. However, it would presumably work very well if you were to use the PX3 to trigger a piano sound that did have that kind of string resonance, like in a Nord Stage 2.

In fact, as I was discussing on another forum, I think the most "authentic" Nord Piano might be achieved by triggering it from something like a PX3, not only because I think the Nord keyboards don't have the PX-3 style third sensor to permit the "pedal-up but damper-up" retriggering (can someone confirm?), but also because the Nord action does not trigger easily enough to permit the kind of light thumbnail glissando you can do on most boards. OTOH, someone else pointed out that there is one kind of real piano functionality that you can do on a Nord keyboard directly that you can't do on a Nord over MIDI, which is depressing the keys so slowly that they make no sound (but could still be used to create resonances for other notes). Again, I don't know how much repertoire calls for that, but it's cool that it works. It really should work over MIDI as well... the fact that it doesn't means that you can record a performance form the Nord into a sequencer that will not play back the same way it sounded when you first played it. It really should produce no tone at MIDI Velocity 1 received externally, just as it does internally. I suspect the fact that it doesn't is a bug.

Anotherscott,
I suspect that all the Clavinova does is overlap voices for partial repeats, just as it would do for a normal sustain pedal. I believe that almost all DPs do this, at least, aside from any fancy sympathetic resonance processing, for the sustain pedal, and it would be trivial do replicate this for partial repeats. If you repeat a single note, with the sustain pedal, I'd be surprised if ANY DP you were to test would not produce a subtle "ding dang dong" sound, due to the phasing effects of multiple voices sounding simultaneously.

So, I'll be surprised if my Casio PX-330 doesn't behave similarly to the Clavinova I tested.

I think the overlapping voice technique, although very crude, just seems to work well.

You can SEE this behaviour in software pianos that display the voice count in real time.

Greg.

If money is no object, I'd look into a digital with real grand piano action (long wooden keys, with real hammers), rather than just something with simulated escapement or triple sensors.

While triple sensors help with fast repetition as others have noted, the resistance at the bottom of the key makes it feel more like a grand piano. I find playing softly very difficult on a grand piano. I think this is because when you are depressing the key slowly, the extra resistance at the bottom of the stroke can slow down your movement even further, to the point where the hammer is not thrown high enough to reach the strings, and no sound is produced. The digital escapement simulation makes you compensate for this and focus on playing through to the bottom of the key evenly, and while I occasionally fail to produce tone on my digital with escapement, it is still far easier to play softly on it than on a grand.

My teacher is a bit of a piano purist as well. She always forgets that I play on a digital and bemoans it when it comes up... but I guess if she always forgets about it, it must not be negatively impacting my technique, right?

This thread has been technically very interesting. I play on a Roland DP-990, which has escapement, but I believe only 2 sensors. I know a note can be repeated without the key fully returning to the top, but now I'm curious if the damper engages between... I will have to experiment.

Greg, I don't often disagree with you, but I would be rather surprised if most DP sound generators overlapped notes of the same pitch like this. Once the looping hit it could statistically sometimes cause drastic phase issues, and it could massively eat into the polyphony. Software pianos have no looping, and so much polyphony to spare (or at least they act like they do ) that they can afford to throw some at note repetition.


Her toes would curl if she chould see some of the things my wife's students practice on - old home organs, toys, etc. Your Roland DP-990 should be the least of her worries.


You might try those repeated notes at various places on the keyboard. I bet there isn't much difference between two and three sensors on the low end, where (if brief partial damping is realistically implemented) the mass of the heavier strings keeps the note from damping much in between plays, but on the high end it may be trickier on a two sensor.

Dewster - we'll have to remain in disagreement then. DPs have done this practically FOREVER. Go and find the oldest one you can, and test it. You'll hear the subtle phasing. (ok, if it doesn't, then try to find a slightly less old one. ;^) ;^)

The reason they can get away with it is that piano notes fade away relatively quickly. Also, they may limit the number of overlapping voices too.

EWQLP doesn't do any voice aging. It will keep overlapping voices until it runs out of polyphony. (which I think it just silly). Yet, it still sounds ok.

From memory, Pianoteq does limit the number of overlaps. (can't remember the number, but I think it was about three)

The Lounge Lizard EP software is a rare example of one that does NOT overlap, and it sounds very bland. HOWEVER, the early versions MODELLED it, and it had a flaw - every now and then, repeated notes with the sustain pedal would produce a note that was far too loud. As far as I can tell, they simply disabled this aspect of the modelling, but did not implement voice overlapping. I have let them know about the issue.

I hacked in overlapping into Csound a very long time ago. I noticed that repeated notes with the sustain pedal sounded bland, so I coded it in.

Greg.

Quarkomatic, the DP-990 is actually a nice DP and I would consider to be one of the better DPs out for piano. Any of the higher end Yamaha, Roland, Kawai, and even the Casios are a heck of a lot better than many of the instruments students learn on, as dewster pointed out.

As far as the money being no object, the Yamaha AvantGrand N1, whilst still quite expensive, is the least expensive digital (hybrid) that comes with a full grand piano action. For around $8,000 it takes up no more space than an upright, sounds a heck of a lot better, plays even better, and costs less than many new uprights.

Yes, I'm quite happy with it. Digitals in general mimic the touch of a grand, which is great - I find the touch on most uprights to be far too light.

It may be that I'm just used to the DP-990's keyboard now, but I was at a piano store recently, and I found the actions on most of the other digitals felt completely different - even other Rolands. One might think the F110's "PHA alpha II" would be similar to the DP-990's "PHA II with escapement," but it seemed vastly inferior to me. There seemed to be more friction in the travel of the keys, whereas key travel on the DP-990 feels smooth and only resisted by the weight of the "hammer." The keys also looked and felt cheaper - more plastic-y and with slightly more rounded edges. My one complaint about the DP-990's action is that the key bed is rather hard, so it is a bit thumpy.

Anyway, it's certainly a good idea to try before you buy. When I made my purchase, I didn't really know what to look for, but looking back with more experience now, I think I made a good choice.

Oh yeah. Huge difference. I owned both the RD-700GXF and RD-300GX at the same time. The 700 having the PHA II Ivory Feel w/escapement, the 300 having the PHA II Alpha. The difference was night and day. After playing the 300, going back to the 700 was like going from an upright to a full on concert grand.

OK Greg, I'll definitely be on the lookout for this one!

The first time I noticed the effect was on my Roland RD-300SX (the old Structural Adaptive Synthesis one, about 20 years ago). I thought it may have been some fancy modelling, but someone told me it was likely to be simple overlapping. I'm sure they were right. I had a Kurzweil PX-1000 at the same time, and I'm pretty sure it did it too.

Note that in EWQLP, in general it takes a very long time to actually terminate a voice after a release, so that may be inflating the voice count.

The main piano sounds on my Casio PX-330 do it, however the EPs, and even the "GM" piano sounds, do NOT do it.

What you're listening for is a very subtle "jaw harp" ding-dang-dong when you play repeated notes with the sustain pedal. It sounds very natural - it doesn't spring out at you. However, when you try it with a sound that does NOT do this, it sounds cold and dead.

Greg.

I just tested Pianoteq again. I was wrong when I said it overlaps about 3 voices - it's 12. If I just play a single note with the sustain pedal, it only uses 1 voice, so it appears that the sympathetic resonance effects don't consume any extra voices - at least nothing that shows up on the performance tab.

The voices are terminated about 2 seconds after release without the sustain pedal, but for repeated notes with the sustain pedal, playing a note just twice will cause 2 voices indefinitely. So, it appears that Pianoteq is intentionally sustaining the earlier voices - it's not just the normal release fade time.

Without having any knowledge of Pianoteq internals, I can't be 100% sure that it's doing the overlapping in precisely the manner I have always assumed, but on the surface, it appears that it could be. (although 12 sustaining voices for a single note seems a bit excessive, if it really is doing that!)

Greg.

So where does "playing off the jack" come in?
With reference to this site, I surmise that it's the ability to play a note by pushing the key very slowly to the double escapement position without sounding a note, and then playing the note from there?

One thing I'm curious about is why manufacturers haven't made double escapement standard across all pianos. There should be no excuse with Digital Pianos. Even on Acoustic Pianos I don't understand why classical music pianists would want to buy uprights, or why manufacturers haven't found a way to incorporate the mechanism into uprights.

I don't buy the argument that it's "for advanced players" only. Classical music is full of trills and other ornaments, and having double escapement would seem to help greatly in playing them well. And how will you even learn to do that if your instrument doesn't allow you to repeat notes without fully lifting the key?

Anyway, this an older thread, so to update as of the end of 2012, my understanding of the premium keyboard actions on the market are:

Yamaha AvantGrand N1/N2/N3:
- full wooden keys with replica grand piano action right up to (just before) the hammer
- optical sensors
- DOES have double escapement
- slightly heavier action than a real grand

Kawai Grand Feel(GF) (CA65/CA95):
- Wooden keys with similar pivot point to a grand piano
- 3 optical sensors
- DOES simulate double escapement
- great feel, has simulated "let-off" bump

Roland PHA III (LX-15, RG-1F, FP-7F, etc)
- Plastic keys
- 3 optical sensors
- DOES simulate double escapement
- simulated "let-off" bump? (unsure)
- lighter action with a harder stop at the bottom of key travel

Would this be a fair assessment?

I'm interested in updating my DP and my priority is the key action since superior sound can always be generated by software as long as there is MIDI out. I was going to go with the N1, but was interested in the new Kawai GF keys at 1/3 the price until I found out it doesn't do double escapement... EDIT: I have now verified this to be untrue. In fact the GF keys really do feel good.

I need to reread this thread to see why you are saying the GF action in the CA line doesn't simulate double escapement. I think it does. As far as I know, every triple sensor action simulates double escapement in the sense that you can repeat a note without lifting the key fully and resetting the damper.

Also, the sensors in the Roland and Kawai models you mention are not optical. They are electromechanical--somewhat reminiscent of the switches under the keys in your computer keyboard. This type of switch is all but ubiquitous in digital pianos.

Lastly, the action in the AvantGrands are modified. There are no felt hammers. Just weights. So the "up to the hammer" part is incorrect.

Hmm thanks for the insight, gvfarns. It seems that I am a bit confused and underinformed about issues.

See for instance this post.

I don't quite understand how "let-off" comes into the equation since from my understanding from here, let-off is the distance that the hammer travels away from the strings after striking it... Obviously it does not make any sense in a Digital Piano, so obviously Kawai have to be using the term to mean something else.

I still don't quite grasp the difference between "play off the jack" and a normal double escapement playing though, because if you look at the mechanism of a grand piano, the hammer can only be in one of 2 positions at rest - fully retracted or sitting on the Back Check. So how can there be a difference between playing off the jack (if I've described it correctly above) and a normal half-depressed repetition of a key (aka double escapement)

As for the optical sensors - I was under the impression that the sensors measure the position and velocity of the key action, and the rubber domed switch (of which there is only 1 per key) is just the on/off indicator when the key is fully depressed.

I take your point about the N1 lack of hammers and have adjusted my list.

I think I will need to read up a bit more on these issues.

Lots of good questions. I'll try and keep them organized.

People say "double escapement" to mean a number of things in digital pianos. The most significant meaning is the ability to restrike a string without putting down the damper and resetting the action. That is something a grand can do (thanks to double escapement) but an upright cannot. In digital pianos, a two-sensor action behaves as an upright action does: you cannot restrike a note without lifting the key enough to reset the action, which also lowers the damper. This is because the higher sensor is used both to control the lifting of the digital dampers and to begin velocity measurement. In a three sensor action there's a separate sensor for the dampers and two to calculate the velocity, so you can repeat a note without lowering the damper.

Sometimes people are also referring to the physical sensation you get when escapement happens. It's a little jitter or bit of resistance. Roland has been simulating this for some time and some Kawai actions do it as well. It is more correctly termed "letoff simulation." It doesn't affect the ability of the piano to play repeated notes.

Playing off the jack means pushing down the key to where this letoff simulation kicks in, and then pressing down from there, as you can on an acoustic. Whether this works or not has to do with where, exactly, the manufacturer placed the sensor relative to the resistance. As far as I know, playing off the jack is not tied to anything else more significant in digital pianos (the AvantGrand has a real double escapement mechanism, but it cannot play off the jack), so it can safely be ignored if you do not use this technique. Roland does it, Kawai and Yamaha (at least the AG line) do not.

Most people don't play off the jack. The primary reason for concern is that for greatest accuracy velocity is measured as near the end of the strike as possible (as near the letoff point on an acoustic). The fact that some digitals have letoff simulation but cannot play off the jack might lead us to believe that the sensors are not placed very close together. For this reason, the reply pv88 got from Kawai is a little cryptic. I'm not sure anyone around here (James excluded) understands exactly why correct position of the sensors precludes playing off the jack. At least, I don't. The Kawai engineers appear to be action purists, though, so they probably know better than I do.

Digital piano sensors are typically on/off switches placed at different heights. Velocity is calculated by the time it takes from one switch being activated to the next being activated. Your computer keyboard has only one switch per key. But if it had one at 2mm of depression and one at 3mm, then it could calculate the velocity with which you pressed the key. This is how most digitals work. For each key there is a rubber dome with three switches that activate at different heights. The AvantGrand has an optical equivalent. It has two beams of light that are broken as the hammer passes them. The time between the two breaks is used to calculate the velocity. The third sensor on the AG is under the keys. We don't know the details of how they work but probably they shine a light up to the key and when the reflected light reaches a certain intensity, they are considered "on," which probably means the software dampers are lifted.

As far as I know none of these switches is continuous. They are all on or off. The reason Yamaha uses optical sensors (as far as I know) is not because they are more continuous or in other ways better, but because they don't affect the feel of the action. Since the AG already has a grand action, they wouldn't want to make you push down rubber domes and perturb the natural key sensation.

The super-expensive alpha piano apparently operates on a different principle than other digitals and has an actual pressure sensor. I have never heard of anyone around here owning or even trying one, so we may never know the details on how or how well this works.

Appreciate the detailed reply, gvfarns.

I had not appreciated the fact that the damper stays off the string during a double escapement action. However it still doesn't make sense that "double escapement" as defined by "in the sense that you can repeat a note without lifting the key fully and resetting the damper" can exist independently of "playing off the jack". As far as I can see, if you have one, you have the other. Playing off the jack is just a case of playing ONE note as opposed to repeating a note, under exactly the same circumstances (half-depressed key). Unless of course, the key "release" or "reset" point is at the halfway point of return travel, but in that case the damper would have come back on the string, and it would not be considered "double escapement".

This link demonstrates that there is only 1 type of motion possible with "double escapement", so there doesn't seem to be a different type of motion possible for "playing off the jack"

Anyway, are you sure about your assertion earlier that the Kawai GF keyboard "implements double escapement" as defined above? Might have to look around for a Kawai dealer to test it out after all...

I had not appreciated the let-off "feel" issue. I should pay attention to that in my testing.

Incidentally, I tested a Yamaha GB1K "silent" (acoustic) baby grand yesterday. Whatever sensors they put in there, it does NOT do double escapement in silent mode. It was plain that repeating a key with very short travel from the bottom would sound in "regular" mode and would NOT sound in "silent" mode. In any case, they went through all that trouble to put a silent module in, but neglected to include simple midi out, very disappointing.

Some of the Kawai plastic actions have the escapement feel feature. For example my old CN33 did, so I assume the newer range of x4 models do.

Depends on how you define "half." These digitals do not actually have jacks, so there's no reason to think that the point of resistance and fast repetition should be related in any way at all. All you need is for the resistance point to be placed below at least one of the velocity sensors and playing off the jack is impossible. This appears to be the case in the Yamaha AvantGrand and in Kawai's triple sensor actions. It's not called playing off the jack unless you can rest the key on the point of resistance and then play from there. But you can definitely replay a note on a triple sensor action without lifting it as far as you need to on a double sensor action, and that's the key characteristic of a grand action.

Notice that the placement of this resistance point has nothing to do with the sensor governing the dampers. The important characteristic of a double escapement is that you can play a note twice without lifting the key to the point where damping occurs (and the action resets). The point of resistance used when playing off the jack is a very unimportant characteristic of a double escapement action by comparison.



I am quite sure, but you can test it for yourself if you would like.

Be aware that in all pianos (acoustic and digital) the difference between the point at which you can replay a note in a single- and double-escapement action differ by only a couple of millimetres or so, so you have to be very careful in your testing. Under normal circumstances one can't easily tell if an action has double escapement or not. That's one reason so many very fine digitals have been made without this feature.

In other words, try it on an acoustic upright and grand before moving to the digitals. That way you will ensure that your testing methodology is sound.



That is disappointing and saddening indeed. I had no idea there were silent modules that did not export MIDI.

I can understand how a keyboard action with fake/simulated double escapement feel and sensors in the wrong place can give rise to the dichotomy of "double escapement for repeated notes but no playing off the jack". However the Avantgrand has EXACTLY the same mechanics as a grand piano. (I could not find a clear cross section picture to verify that, but articles like this make me assume that the whole shebang including double escapement is identical.)*
If this assumption is true, then the double escapement action on an Avantgrand must be generated by mechanical action identical to a grand piano. If so, it's impossible for it to have double escapement but no playing off the jack. It's not like the Avantgrand has a sensor for saying "oh, the key is now in double escapement phase"... it just happens mechanically same as in a real grand piano!

Any action that will cause the hammer to hit the strings in a real grand piano, should trigger the hammer sensor on the Avantgrand. And like I've said, in a real grand piano (and presumably in the Avantgrand), as far as I can see, the action that gives rise to double escapement is exactly the same as that for "playing off the jack".

Anyway, it's just something that puzzles me, but you're right, "playing off the jack", whatever it means, is not something I would need. However, the double escapement feature is definitely a requirement. I just don't see how the two are different in the grand piano mechanism.

*Edit: found this. You can clearly see there is a Back Check that the Hammer Tail rests on, so it's safe to assume the mechanical double escapement is all there, exactly like a grand piano.

You are right that the double escapement mechanism is there and being used, but when you say "Any action that will cause the hammer to hit the strings in a real grand piano, should trigger the hammer sensor on the Avantgrand" you make the mistake of assuming that the sensors on the AvantGrand are at the location of the string and infinitesimally close together. They are not.

Only the hammers passing the two velocity sensors are used in computing whether a strike has happened or not. If the first sensor is further up in the stroke than the jack is, no playing off the jack. You can make the hammer shoot up from the jack, but no note sounds.

It is a well-established fact that the AvantGrand cannot play off the jack but it does feature a triple sensor action, real double escapement, and fast repetition capability. Though I guess the repetition may not be as fast as that of an acoustic if the configuration is as I have described.