Weird limit on Yamaha SY35/SY22 mod wheel resolution.

[Lostgallifreyan]

If you have either the SY35 or SY22, please can you tell me if the mod wheel resolution is unusually low, stepping 0>4>8 etc..?

Reason for asking:
I bought a cheap SY35 as a test keyboard over a year ago, I'd got tired of an ailing Alto Live49, I like Yamaha key actions and the MIDI spec is usually great. I found the pitch and vector control stick to run 0~127 as expected (with odd use of the first byte in pitch, when bending upwards), but mod was in steps of four, as the aftertouch appears to be too, but it's harder to tell with that one, it's very sensitive to tiny changes in pressure once there's enough to register. Proportional foot pedal only controls local volume, does not transmit MIDI.

The SY35 service manual says nothing about limited res, it just says range of 0~7F as expected, so testing a real instrument is the only way to know. I doubt mine is broken, everything else works perfectly, and the potentiometer action is clean.

If someone can tell me the results of an SY35 mod wheel test output to a MIDI monitor, please do. I want to know what the SY22 does too, because it's possible that the 'lesser' device might have a slightly better MIDI spec. Originally, even the cheapest Yamaha gear would not have limited the control res like this.

[Folderol]

Yes. Seeing the same here on my SY35. Hadn't noticed this before even though I've had it for donkey's years. Don't use it much anyway. More modern keyboards seem OK.

[Wonks]

Presumably a limitation of the rotary encoder used (or the A/D converter a pot value is fed into). 32 steps, not 128.

[Lostgallifreyan]

Yes. Seeing the same here on my SY35. Hadn't noticed this before even though I've had it for donkey's years. Don't use it much anyway. More modern keyboards seem OK.

Thanks for that confirmation. VERY helpful. The only reason I noticed it right away is I was looking to see exactly what kind of control res I was getting.. Can you recommend a modern Yamaha board that is fitted with enough controls for general use and testing of software, while beign cheap, easy to find second hand? I don't care about the sounds, so if people sold it off hating the sounds while neglecting a nice MIDI interface device, that might be exactly what I need to explore.

It is odd because the normal method is to multiplex a binary multiple of inputs. Consider the SY35: two axes in a vector control, a pitch wheel, a mod wheel, and aftertouch. Five.

I'm going to guess that they got clever with a 4-input scanner and somehow got a low-res scan for aftertouch and mod wheel, likely 4 bits for each (it would be 8-bit scan, internally), hence the adapting by upshifting by 2 bits to fit the MIDI control range. How they get the fifth bit per half-input I don't know. I read of some method to add a bit with discrete parts but can't remember it.

A neat trick, but I wish they hadn't done it with the mod wheel. An 8-input scanner and a proportional foot pedal socket, perhaps adding their (in)famous breath control, and they could upped the spec for minimal cost and sold a lot more of those synths!

If anyone seeing this can test the SY22, please do. I expect the same result, but it will help to know for sure.

[Eddy Deegan]

I don't have an SY22 but out of curiousity I tested an SY85. The resolution on the mod wheel is as one would hope, all the values from 0x00 to 0x7F in steps of 0x01 so it looks like something Yamaha did on some of the smaller models in the SY series.

[Folderol]

I don't have an SY22 but out of curiousity I tested an SY85. The resolution on the mod wheel is as one would hope, all the values from 0x00 to 0x7F in steps of 0x01 so it looks like something Yamaha did on some of the smaller models in the SY series.

I rather suspect is more a matter of age rather than size. 22,35,85 came out in that order over several years.

@Lostgallifreyan
I'm rather in the same boat of wanting another keyboard with a nice synth type action - with aftertouch

[Wonks]

The SY35 service manual circuit diagram shows the the mod wheel pot and the pitch bend pot feeding into an analogue multiplexer. The output from that feeds into an A/D port of the main CPU (not the scan processor). The A/D converter has a 10 bit resolution, so no theoretical reason why it can't output an 8-bit value for these two controls. They can throw away the noisy bottom two bits and still have an 8-bit value (7-bit plus sign).

[Lostgallifreyan]

I don't have an SY22 but out of curiousity I tested an SY85. The resolution on the mod wheel is as one would hope, all the values from 0x00 to 0x7F in steps of 0x01 so it looks like something Yamaha did on some of the smaller models in the SY series.

I rather suspect is more a matter of age rather than size. 22,35,85 came out in that order over several years.

@Lostgallifreyan
I'm rather in the same boat of wanting another keyboard with a nice synth type action - with aftertouch

Yes, exactly, they ARE useful, even to the point of picking a cheap one with lousy sounds just to have that control gear..

Thankyou Eddy, that is useful info too. I wonder if SY85 might be a contender, now that sample-based work is best done on a computer where long and diverse samples are more easily had, and the overheads to play them there are relatively low. Cost of SY85 might still be too high though, especially if rarer..

Re age of gear, the newest gear is better in res from what you said. Original gear was too, high end stuff likewise, hence my speculations about the special case of trying to put a vector control into what is otherwise a synth of very limited spec. I think that for whatever reason Yamaha just didn't want to use an 8-input scanner... Similar ingenious ways to fit more into a few bits, is a thing that Yamaha good good at in data storage and in things like LFO delay controls and other auxillary functions, so it seems likely they thought it more efficient to do this to get five inputs to a 4-input scanner somehow. Great for what it is, but it does rule out the SY35 (likely SY22 as well) as a general test board.

Thankyou Wonks, that's interesting. I wondered if it might be a 10-bit scanner (because somehow they seem to be getting extra bits to put in data byte 1 of the pitch bend messages when up-banding the wheel), so that confirmation is good, and it might well be that they did split mod/aftertouch across one line of a 4-input scanner.

[Lostgallifreyan]

The SY35 service manual circuit diagram shows the the mod wheel pot and the pitch bend pot feeding into an analogue multiplexer. The output from that feeds into an A/D port of the main CPU (not the scan processor). The A/D converter has a 10 bit resolution, so no theoretical reason why it can't output an 8-bit value for these two controls. They can throw away the noisy bottom two bits and still have an 8-bit value (7-bit plus sign).

That's sort of how I'd do it, even for pitch bend if I only wanted 0~127. Scanning with 8 bits per control not only reduces noise, but the code can read this to detect a jump of 2, not 1, in SAME direction, and only register that as a valid transition, i.e. it reduces not only analog noise, but digital, too. It won't vacillate just because it's sitting on what might otherwise appear to be a boundary between two output values. A lot of cheap MIDI controllers do NOT do this, so they send spurious signals a lot, unlike good ones.

Another nice thing about Yamaha and other quality makers, is they scan in TIME, sampling the value, instead of queuing up every change and trying to force them all out of a MIDI port given the time. This is another reason I gave up the Alto Live49 I mentioned in first post. While it helped me code a nice mod system with ways to optimise good responses for BAD controllers, there is no substitute for a GOOD controller!

[Wonks]

The aftertouch sensor is one of the five inputs (with the mod and pitch bend multiplexed and feeding a shared port) that feed directly into the main CPU's 4 A/D input ports. The other two inputs are the vector control joystick axis pot values.

[Folderol]

The aftertouch sensor is one of the five inputs (with the mod and pitch bend multiplexed and feeding a shared port) that feed directly into the main CPU's 4 A/D input ports. The other two inputs are the vector control joystick axis pot values.

They really were squeezing that CPU till the bits squeaked

[Lostgallifreyan]

The aftertouch sensor is one of the five inputs (with the mod and pitch bend multiplexed and feeding a shared port) that feed directly into the main CPU's 4 A/D input ports. The other two inputs are the vector control joystick axis pot values.

They really were squeezing that CPU till the bits squeaked

Yamaha did that a lot.. Probably CPU time wasn't the limiter, their schemes would use a bit more time at need (but the operations are fast). The real bottleneck seems to have been storage and bus width limits.

Wonks, I don't know why the aftertouch was on its own line. I assumed it might not be because its res is limited like the mod wheel. Their pitch bend messages are weird though, they have a 7-bit fullrange value in data byte 2, but a lower res fullrange in databyte 1 that only works on upward bends. This is a weird thing to do, I can't think of a reason for it.

Whatever they did it was tortuous, worth finding a different test keyboard just to avoid it. Adding DIY to a simple keyboard is not cost effective or convenient, so it boils down to finding overlooked gear that was widely sold, and is ideal to use now when its sounds don't matter. This was what led me to the SY35. (But I do like its pipe organ..)

[Lostgallifreyan]

The aftertouch sensor is one of the five inputs (with the mod and pitch bend multiplexed and feeding a shared port) that feed directly into the main CPU's 4 A/D input ports. The other two inputs are the vector control joystick axis pot values.

I found a PDF reader that could cope with the schematic on the machine I store my references on, and it does show that the details are at the mercy of the main CPU with its four inputs. It is odd stuff, the aftertouch gets its own line, so why they limited its res like that of the mod wheel I don't know. Given that the clock (line 'AC') just time-shares the mod and the pitch bend by sampling them for line 'AN3', there's nothing in the analog circuit that prevents full res for both. Whatever the reasons for any of the weirdness, it's locked away in the code Yamaha used to handle the signals after they enter that CPU.

I think this almost certainly excludes the SY22 as a contender. A quick look on Ebay suggests the SY85 might be one! I'd spend £200 or so to solve this. Eddy's test suggests it will. I think the main inconvenience would be the larger size.

[Lostgallifreyan]

Has anyone got an SY55? Please let me know the result of the res-test if so.. This is an interesting keyboard. It's compact, sells at low cost, but it has the breath control input, sends volume control from pedal to MIDI on CC007, it has aftertouch, and a Data Entry slider that sends on CC006. This is enough for a useful test controller, if all these outputs use the full resolution. Not having vector control is probably no disadvantage for this task.

[Lostgallifreyan]

^^^ Anyone? Please?

I'll shut up after I get a test result for the SY55 because that ends the likely range of Yamaha synths to test for the purpose.

Thanks to all those who have helped me with results of tests so far, it's extremely helpful to narrow this down that fast and that well..

[Wonks]

Maybe try on a Yamaha synth forum? e.g.

https://yamahasynth.com/community/

[Lostgallifreyan]

Maybe try on a Yamaha synth forum? e.g.

https://yamahasynth.com/community/

Possibly. I don't like signing up to lots of forums though (and social media gives me the heebies). I'd rather just buy one to try, given the likely low cost and the info already given allowing me to infer it's a fairly safe bet that it's good enough.

Still, I doubt it's too specialised for someone on SOS to know the answer, though I admit that there's a good chance no-one here has got one of those synths anymore.