Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards Kawai F-10H pedal have too much latency/delay

EDIT: At the end the problem was the ES100 all along. Kawai said that the first units has this problems and they fix it. For me, they repair the unit for free and now it works great.



Now, my original post:

After buying the Kawai ES100 digital piano, I was (and still am) very happy about it. I have nothing bad to say about the ES100, except very minor things I will post on a full review.

But the pedal that came with it (F-10H) although looks very well made, with a lot of metal parts, heavy, stable, silent, apparently working well... but after some little time after using it was clear something was wrong about that pedal. Was a long story to figure out what was the problem, but now I know. Problem is latency (delay). When you press fully, the signal send to the digital piano have around 0.115 to 0.172 seconds of delay, and even 1 second in some circumstances I will explain later.

People that play an instrument like the piano know that 0.115 of a second of input delay is quite a lot. The pedal shows (with my ES100 digital piano) 6 to 7 steps in total. Each step has a delay that adds on, from full-press to release.

This is the delay between values:

Delay with F-10H pedal:
MIDI message from 10 to 30 = 0.012 seconds of delay
MIDI message from 30 to 55 = 0.014 seconds of delay
MIDI message from 55 to 84 = 0.021 seconds of delay
MIDI message from 84 to 116 = 0.032 seconds of delay
MIDI message from 116 to 127 = 0.093 seconds of delay

Total delay with F-10H pedal from 0 to 127: 0.172 seconds

Notes that delay is bigger the more the pedal is press.


Delay with a switch pedal:
MIDI message from 10 to 30 = 0.002 seconds of delay
MIDI message from 30 to 55 = 0.001 seconds of delay
MIDI message from 55 to 84 = 0.002 seconds of delay
MIDI message from 84 to 116 = 0.002 seconds of delay
MIDI message from 116 to 127 = 0.005 seconds of delay

Total delay with switch pedal from 0 to 127: 0.012 seconds



Note: For those who don't know about MIDI, values are 0, 1, 2, 3...until reach 127 (that is full)

Using a switch pedal proves the problem is the kawai F-10H pedal itself since is obvious the digital piano electronics can process pedal signal very fast (0.012 seconds maximum from 0 to full).




What problems in real life this cause?

Issue "late pedal": Is the most obvious problem I had that actually make me start investigating about all of this. Is quite important problem because when pressing a key, and releasing the key at the same time when pressing pedal, sound should sustain normally all the way. But because of the latency on the pedal, the key is partially damped (called "late pedalling"). As most of you should know, pedal usage is all about timing. This slight delay can really ruin a song quite often. Even pressing pedal a little earlier than releasing key, late pedalling issue is present dampening a little the sound. You are force to leave the key press for longer time, and this is not always possible, and is not realistic.



Issue "full pedal delay": Sometimes when fully pressing the pedal, the MIDI monitor (digital piano is connected to a computer) shows an important delay on sending the FULL (127) state. The delay is sometimes even over a second. It freeze on 116 value and after a while it sends 127 (full).


Here is a short video showing both issues in action: https://www.mediafire.com/?dp41hw2w3mm2w9a




Issue "pedal damper noise effect": The digital piano Kawai ES100 supports this effect. The pedal noise is louder the faster you move the pedal, like on an acoustic piano. But since the F-10H pedal sends data with such latency between steps, the digital piano always thinks the pedal is moving slowly so the damper noise is always too soft, no-matter how fast/hard the pedal is press. This issue is not super important, and that is why I mentioned last. But still, was something I notes even before I knew about all of this pedal-latency thing.



Obviously the delay also happens when releasing the pedal, but I didn't notes that when playing so I don't mention that as an "real life issue", although technically it is.



After I discover this, I remove the cover on the F-10H pedal, and looks like it use some kind of optic sensor. Not sure what cause the latency in there. My bet is the sensor itself, but the problem could be some other component.

Other kind of pedals like the Yamaha FC3 use potentiometers to send half-pedal position. Problem with potentiometers is that they wear out after some use. But at least potentiometers send data practically in real time (no delay). The initiative of kawai to use optic sensor is very good to prevent wearing (the pedal in theory could last dozens of years), but sadly they fail on making them latency-free. Everything about input on a instrument most have as little delay as possible. 0.001 of a second is very good; 0.01 of a second is acceptable; 0.1 of a second is terrible.






Maybe I have a defected unit?

What about other pedals like the Kawai F-350 triple pedal unit?

I don't know since I didn't have a chance to test other pedals, and I don't have an answer from Kawai yet. I will post any news about this here in the future, if there is anything to report.




If any of you are interested on the schemat of the pedal, my friend dewster write it down here (thanks dewster):





For now, I will see how solve this problem for real. I want half pedal feature so will test with potentiometer pedals and see if any of them is compatible. Until then, switch pedal should be enough.

If any of you have the F-10H pedal, please do the test to see how much delay it has. Could be that I am just unlucky and got a defective unit. The way to test is connect the digital piano to the computer and install pianoteq (trial version works fine). In pianoteq there is a setup section where shows MIDI messages is receiving. That is what I use on the video I record. You simply write how much time it takes from having the pedal release to fully press.

Here is a warped/scaled/blended view of both sides of the PWB superimposed on one another:



U1 is probably an LED / phototransistor pair. Q1 is an emitter follower that buffers the output of the phototransistor. R2 & C1 form a low pass filter pullup, R5 and C3 form a low pass filter pulldown.

The phototransistor is more conductive the more light it gets. So when the pedal is up phototransistor is ON and pulling R2|C1 to ground. When the pedal is pressed quickly the phototransistor becomes a high impedance, and R2 must discharge C1 by itself, which can take 6800 x 0.000001 = 0.0068s or 6.8ms to reach 63% of the final value. Hard to know how the ES100 interprets this, but having most of the action near the top of the pedal movement likely aggravates things.

My advice to Daniel was to remove C1 and see what happens. If there is an improvement then C1 should perhaps be replaced with a smaller value, such as 0.1uF.

VR1 is marked "105" which usually means 1M Ohms, but that can't be right (way too high).

The "right" way to do this kind of optical thing is with a pulsed beam, because ambient light can then be largely removed from the equation.

Daniel, you might also try adjusting the potentiometer on the board, but first mark it so you can return it to the factory calibration point.

Thanks dewster for the suggestions, but I remove C1 and no change. Exactly the same latency, even with temporal freeze on 116 value like I mention before.

I also try to move the potentiometer on the board (VR1). Is clear is use to calibrate the position of the pedal. When move clockwise (reducing ohms) the pedal starts sending signal of press after pressing half-way or more. If move potentiometer too far on that direction, pedal always sends data as released, even when pressing fully the pedal. The opposite happens when moving the potentiometer counterclockwise (increasing ohms), sending signal like is press, but is actually released.

BTW I don't recommend relaying on marking the potentiometer so you can return it to the factory calibration point. Is not very accurate that way. Even a fraction of mm make a lot of difference on the ohms is setup. I prefer measure the ohms with a multimeter, write it down somewhere, and later when trying to restore to original position just move until is the same number.

I think I will stick with the idea of using a potentiometer as a pedal, like Yamaha FC3 pedal do. Tricky part will be adjust ohms and pedal-position to be realistically calibrated.

Question still remains: This latency is on every F-10H pedal or only on mine?

How do you know it's the pedal, and not delayed processing in the DP? Can you try another OEM pedal or visit a store and try another ES100 with pedal.

As I say on the main post, I test with a switch pedal. No significant delay with switch pedal, so the DP is not the problem. Read the post, that there is more details there about this.

I am searching for someone that have the F-10H pedal to test this. When find it, I will post the result.

Interesting post, and great analysis. It would be so tempting to just probe around in there with a scope, to see what's really going on. Maybe there's something wrong with some of the pedal components - like the 6.8k being open-circuited or, or Q1 not working as a transistor, with R3 open, so the voltage changes reach the output but only slowly. There's nothing I can see about the circuit that should make it so slow.

If someone else test this pedal and there is no delay, then the problem is a damage component for sure. We'll see.

Daniel, there may be some kind of calibration going on at power-up in the ES100. This could confound your testing. You might try adjusting the potentiometer a bit and then powering down, then powering back up.

Anything beyond this probably requires a scope and some inside knowledge of the ES100. Access to the former is likely much easier than the latter.

Actually I test that. What I found out is that ES100 don't calibrate in that way. Is always expecting a value from the pedal to detect half positions. Also means I most find a specific potentiometer to substitute this pedal and make my own.

So what calibrates is the pedal, not the digital piano. The calibration is that potentiometer VR1

Very interesting especially since I've been wanting to get this p. I really hope this is just a defective unit.

Would it be possible for you to replace the pedal through warranty? Perhaps the store you bought the dp from can do it. If the replaced pedal is still the same, then it's likely every F10H pedals are like that, Also, have you tried with a different digital piano?

First I most know if my pedal is defective or is a design problem, before considering sending it back for the warranty. I live in Venezuela so is not cheap ship things to US.

Can't try on different digital piano because no-one here have one. At least not Kawai. Anyway I know (from my test) that the ES100 is not the problem, so my focus is on the pedal itself.

I already contact the Kawai dealer and they contact kawai directly. Also James had inform this to kawai too. A response from them could take a lot of time, if there is any response.

I contact a friend that bought a Kawai ES100 and she will make the test. I have more hope on that to know if I have a defective unit or all F-10H pedals are like this.

"Stay tuned" for more news about this

My hunch is the pedal is not defective, and that the circuitry or software that interfaces with the pedal in the ES100 is the issue.

My hunch is that the problem is the optic sensor itself that is too slow, and that all F-10H pedals have the same problem.

Let's see.

Confirmed. All F-10H pedals have basically the same delay.

I ask a friend "therightmoment" to do the test with pianoteq showing the midi messages and when she press the pedal fast she get 0.116 seconds of delay from 0 to 127. Although the issue "full pedal delay" is not present. She only get "normal" delay of 0.116 seconds.




Her pedal shows 16 steps. Mine only 6 to 7 steps. But that don't bother me at all. What bothers me is that if I return this pedal I will have the same delay. Well, the "full pedal delay" would be solve, but I don't care if I still get 0.1 delay. So returning pedal is out of the question.

I have two options:
- Modify my kawai pedal to use switch, because I love the mechanics of this pedal, and the feel. My current switch pedal feels too artificial (what you expect from a cheap pedal). I want keep using this pedal. I only need to change what is inside.
- Or modify the pedal to use a potentiometer to send half positions. I assume the digital piano should read a potentiometer values like it was on the pedal. As far I know, that is basically what this pedal do, but digitally. I will try to make the same but analog.

I am not sure when I am going to do this mod. Is christmas and I don't have much rush to do this since I already am using a switch pedal. But when I get the time I will do it and report here how is going.

There is in theory one other option and is buy a F-350 3-pedal unit for the ES100. I am not sure yet, but I bet it doesn't have the same latency issues of the F-10H pedal. But F-350 pedal is too expensive for me. And overkill in a way because I don't use the other 2 pedals. It is problematic fit that pedal unit on a stand. Would need to buy the official kawai stand, or do a mod on my stand. But main reason is just too expensive, the unit and the shipping to my country.

If any of you have the F-350 pedal, would be great if you can test latency on that pedal.

Thank you "therightmoment" for doing the test.

Happy holidays everybody.

I would not assume this:

Right. If it's looking for a voltage, you'd have to rig the pot to produce the expected output voltage range, which would probably be a subrange of the supply voltage range applied to the pedal.
I think it's very unlikely that any lags are due to the optical devices.

Uh oh. Daniel, how big of an impact do you think this will be for people who have this pedal? Will it affect the playing experience or is it minor?

Actually many light sensors are quite slow. Depends on the type of sensor they use.



I think 0.1 seconds of latency on the pedal is notable for players that already play real songs on real pianos. If you are a learner that press pedal all the time or don't use pedal in the same way you would do in a song, don't see a problem. But I am experienced piano-player and notes this latency in day 1 since I got the digital piano.

I could still play songs fairly well. It doesn't make impossible play songs. But late pedalling happen quite often and normally that happens as a mistake of the player for pressing pedal too late. With this pedal happens even if you didn't make mistake.

So basically if you press pedal too near the limit, you will have late-pedalling even though you did it inside the limit. For complicated songs can happen all the time.

Any latency on input is bad, in my opinion. For instruments or video game controllers or computer screens, etc. Each 0.001 second count.

My recommendation for other owners of this F-10H pedal is buy a switch pedal. There are great ones like the M-Audio SP-2. Another option is buy the Kawai F-350 3-pedal unit, that I suspect should not have this issue. That one is expensive but at least you get half pedalling feature.

Of course, with a switch pedal you lose the half pedal capability, but having 0.1 seconds of latency is a lot worst.

I will explore the potentiometer pedal later to have half pedallling feature without latency.

Daniel:

I poked around DigiKey for a bit. Here is a datasheet of a part that may be similar to the one in the pedal (Omron EE-SY125):

www.components.omron.com/components/web/pdflib.nsf/0/3DF7E7E37F4CF53185257201007DD6AB/$file/D21EESY1250305.pdf

Response time is pretty quick, but can be improved by increasing the collector current. I would try lowering R2 somewhat and see what that does. Solder a 4.7k ohm resistor in parallel with the 6.8k resistor (4.7k || 6.8k = 2.8k) and recalibrate the potentiometer for correct range of operation. If that noticeably improves the response you can lower the resistance more, but not too much or the photo transistor could be destroyed.

It would help to know what the voltage being supplied to the pedal is. There is very likely a resistor or other current limiting device within the ES100 on the supply which could be playing into the equation.

The response times mentioned on that data sheet are 10 to 100 us, or 0.1 ms max. That's about 1000 times faster than the 100 ms lag times that are a problem.