I’ve recently been given an S90 with a bit of an issue.  When I power it up after it’s been sitting a while, it either gives me an error message or a pulsing LCD.  If I try it again (sometimes once, sometimes more), then it’ll power on properly and works great.  At that point I’m also able to power it off and on normally.  All keys, buttons, LED’s, sounds, everything works great, once it decides to work.

The error message it gives me is:

I’ve gone into the test mode by holding [Category Search], [Remote Control], and [Mute], and the self tests for ROM and RAM both pass perfectly.  I’ve run the test multiple times, no failure.  Note that if the keyboard doesn’t boot up, it also won’t go into test mode.

What I figure is that there’s a power supply issue.  I have a spare (used) power supply, though that didn’t fix the problem.  I also replaced a couple capacitors on the main CPU board, which didn’t help either. 

Testing the power supply seems to give me good results.  Voltages are well within 10%. 

Pinout is:

Does anyone have a service manual for the S90, or can help me decipher the error code or fix the power-up problem?  It’s running Firmware 1.03 if that helps at all.

Welcome to the forum.

I’ll preface my reply by stating that I don’t have an S90 service manual, and I don’t have knowledge of the specific error message. Therefore, my comments will be of a general nature.

tertiumsquid - 04 February 2017 06:49 AM

I’ve recently been given an S90 with a bit of an issue.  When I power it up after it’s been sitting a while, it either gives me an error message or a pulsing LCD.  If I try it again (sometimes once, sometimes more), then it’ll power on properly and works great.  At that point I’m also able to power it off and on normally.  All keys, buttons, LED’s, sounds, everything works great, once it decides to work.

The error message it gives me is:

unregistrated interrupt
error code     00000180
occurrence at  8c24f896
content of sr  40000000

That’s a “system down” error message. It’s funny to see “unregistrated interrupt” - it seems the person who programmed that didn’t know how to spell “unregistered”. I’m not sure what value the message would have in troubleshooting the problem. For example, SR (Status Register) contents might reveal what flags are set, but unless you find someone who’s seen that particular message before, I can’t see it pointing to a particular component failure.

By the way, nice usage of the “code” tags.

 

tertiumsquid -

I’ve gone into the test mode by holding [Category Search], [Remote Control], and [Mute], and the self tests for ROM and RAM both pass perfectly.  I’ve run the test multiple times, no failure.  Note that if the keyboard doesn’t boot up, it also won’t go into test mode.

What I figure is that there’s a power supply issue.  I have a spare (used) power supply, though that didn’t fix the problem.  I also replaced a couple capacitors on the main CPU board, which didn’t help either.

I tend to agree that the problem is related to power supply, and capacitors would be my primary focus. Caps that are used for decoupling in digital circuits need to have low ESR, so hopefully the ones you used as replacements fit that requirement.

 

tertiumsquid -

Testing the power supply seems to give me good results.  Voltages are well within 10%. 

Pinout is:

1: +5v
2: +5v
3: Gnd
4: Gnd
5: +12v
6: Gnd
7: -12V
8: NC
9: NC

[...]

+/- 10% on the 12 volt supplies is probably OK, but at worst the +5 volts should be within 5%, and preferably within 2% or so. Slightly high is better than low.

The average DC voltage is important, but it’s also important that the supplies be clean. If you have access to an oscilloscope, what peak-to-peak ripple voltage are you seeing?

5pinDIN - 04 February 2017 11:38 AM

Welcome to the forum.

5pinDIN - 04 February 2017 11:38 AM

That’s a “system down” error message. It’s funny to see “unregistrated interrupt” - it seems the person who programmed that didn’t know how to spell “unregistered”. I’m not sure what value the message would have in troubleshooting the problem. For example, SR (Status Register) contents might reveal what flags are set, but unless you find someone who’s seen that particular message before, I can’t see it pointing to a particular component failure.

Yeah, I kinda assumed the error wouldn’t say much, but I also assume it’s still possible that something beyond the PSU is not working.

5pinDIN - 04 February 2017 11:38 AM

By the way, nice usage of the “code” tags.

Hah!  I had to guess that there were code tags!  Shame there isn’t a button for that on here.
 

5pinDIN - 04 February 2017 11:38 AM

I tend to agree that the problem is related to power supply, and capacitors would be my primary focus. Caps that are used for decoupling in digital circuits need to have low ESR, so hopefully the ones you used as replacements fit that requirement.

I need to order more low ESR caps (Panasonic’s most likely) to do the power supply, that’s why I haven’t done it yet.  The couple I put in the main board were pretty generic cheapo’s.  Local stores here don’t sell much in the way of good caps.
 

5pinDIN - 04 February 2017 11:38 AM

+/- 10% on the 12 volt supplies is probably OK, but at worst the +5 volts should be within 5%, and preferably within 2% or so. Slightly high is better than low.

The average DC voltage is important, but it’s also important that the supplies be clean. If you have access to an oscilloscope, what peak-to-peak ripple voltage are you seeing?

DC voltages (today) are +4.99v, +11.98v, and -12.45v.  Ever so slightly low on the 5v, but less than 1% off, and it hasn’t budged an inch in the hour it’s been turned on.  I’m within 5% on the +-12v, but that’ll be for the audio section and not used at all for logic.

I have a crap oscilloscope, literally a DIY thing I built years ago that runs off an AVR microcontroller of some sort, so it’s not what you’d call “precise”.  That said, the ripple on my +5v is about 80mv peak to peak at first boot, then drops to ~50mv.  Hard to say.  I’ll post a pic of my scope so you can see what I’m working against…

The waveform’s a bit blurry thanks to my phone’s camera, but you can get the gist of what I’m looking at.  This is after a half hour of warm-up, with the S90 working properly.

I definitely need a better scope......

Click thumbnail to see full-size image

tertiumsquid - 04 February 2017 07:05 PM

Yeah, I kinda assumed the error wouldn’t say much, but I also assume it’s still possible that something beyond the PSU is not working.

A symptom of “works when warm or recently run” is typical of failing caps. It also depends on how you define “PSU”. For example, the main CPU (and other components} in the original Motif (Classic), ES, and XS run on 3.3 volts. That supply is derived via DC-to-DC conversion and/or voltage regulation from the 5 volts. I have no reason to doubt that the S90 has similar circuitry.

It’s also possible for a bad cap to change the timing of a CPU reset circuit, for example.

I’d strongly suggest that you get the service manual from Yamaha, if at all possible, so that you can verify voltages beyond the obvious ones, etc.

 

tertiumsquid -

I need to order more low ESR caps (Panasonic’s most likely) to do the power supply, that’s why I haven’t done it yet.  The couple I put in the main board were pretty generic cheapo’s.  Local stores here don’t sell much in the way of good caps.

Panasonic caps are usually a good choice. I’ve found many of the generic ones to be off-value, and have high ESR even when new. If you don’t have a means of measuring ESR, there are inexpensive (~$10) component testers with LCD readouts, that run on a 9 volt battery, available on eBay (and possibly elsewhere).

 

tertiumsquid -

DC voltages (today) are +4.99v, +11.98v, and -12.45v.  Ever so slightly low on the 5v, but less than 1% off, and it hasn’t budged an inch in the hour it’s been turned on.  I’m within 5% on the +-12v, but that’ll be for the audio section and not used at all for logic.

I have a crap oscilloscope, literally a DIY thing I built years ago that runs off an AVR microcontroller of some sort, so it’s not what you’d call “precise”.  That said, the ripple on my +5v is about 80mv peak to peak at first boot, then drops to ~50mv.  Hard to say.  I’ll post a pic of my scope so you can see what I’m working against…

The DC voltages appear to be well within expected tolerances, and the ripple on the +5v, based on the Vp-p you found, seems to be under 1%. If you get the service info, I’d look at supplies and caps beyond the main PSU.

5pinDIN - 04 February 2017 09:18 PM

A symptom of “works when warm or recently run” is typical of failing caps. It also depends on how you define “PSU”. For example, the main CPU (and other components} in the original Motif (Classic), ES, and XS run on 3.3 volts. That supply is derived via DC-to-DC conversion and/or voltage regulation from the 5 volts. I have no reason to doubt that the S90 has similar circuitry.

You hit the nail on the head dude.  I found a small handful of voltage regulators on the main PCB: three at 3.3v, two at 5v.  DC voltage out of them is within 2%.  I checked them for ripple, and two of them are way high.  IC2 is giving me 400mv p-p and IC46 is giving me 600mv p-p!  That’s 12-18%!

FYI, the others (IC36, IC47, and IC57) have low enough ripple that I’m unable to measure it on my scope.  IC2, IC46, and IC47 are all the 3.3v regulators, all the same type of reg, and all fed the same 5v supply (with the same ripple on input), so they should have the same ripple.  I’m going to try and parallel a cap over the ones at the voltage reg outputs in the morning (3am here now, girlfriend getting grumpy).

Also interestingly, I’m getting more ripple out of the main 5v rail once it gets to the main PCB.  I’m still getting ~50mv measured at the PSU, but ~200mv p-p at the main board.  Might need to check the cap at the main board end of the 5v rail…

5pinDIN - 04 February 2017 09:18 PM

Panasonic caps are usually a good choice. I’ve found many of the generic ones to be off-value, and have high ESR even when new. If you don’t have a means of measuring ESR, there are inexpensive (~$10) component testers with LCD readouts, that run on a 9 volt battery, available on eBay (and possibly elsewhere).

I set up a bit of an ESR testing rig with my scope and an audio generator.  Poked around in-circuit testing all the caps in the PSU and they all seem good.  Poked at the electrolytics on the main board as well, didn’t find any suspect, though the crappy ESR testing rig might be to blame.  I’ll check again tomorrow.

I also JUST found the S90 service manual (even after much googling).  ElektroTanya is the best site ever.  I’ll read it to see if there are error codes.

tertiumsquid - 05 February 2017 08:34 AM

I set up a bit of an ESR testing rig with my scope and an audio generator.  Poked around in-circuit testing all the caps in the PSU and they all seem good.  Poked at the electrolytics on the main board as well, didn’t find any suspect, though the crappy ESR testing rig might be to blame.  I’ll check again tomorrow.

I’ve used a scope and square wave with good rise time at 1kHz or 100 Hz (depending on cap value). Comparing the display of the suspect cap with a known good one can help. High ESR can show as a “step” in the displayed waveform.

 

tertiumsquid -

I also JUST found the S90 service manual (even after much googling).  ElektroTanya is the best site ever.  I’ll read it to see if there are error codes.

Thanks for that information. I did a Google search, which usually reveals if ElektroTanya has the manual, but didn’t get a hit. Based on your post, I went to their site, and found both the S90 and S90 ES service manuals.

5pinDIN - 05 February 2017 09:51 AM

I’ve used a scope and square wave with good rise time at 1kHz or 100 Hz (depending on cap value). Comparing the display of the suspect cap with a known good one can help. High ESR can show as a “step” in the displayed waveform.

Yeah, the problem I had was finding a known bad cap to see what it would look like on my scope!  I tested all the electrolytics on the main board (service manual calls it the DM board so that’s what I’ll use from now on) but they all looked good.  Probably my scope doesn’t have enough resolution to see the difference in marginal caps - the smallest voltage that this scope can display is 10mv, which if my math is right, using a 200mv p-p test signal (for in-circuit testing) out of a 50ohm output would end up being 4mv per ohm of ESR.

I thought I had found a bad cap in the PSU when the scope gave me a triangle waveform, but it was the smallest cap at 1uf and the same thing happened with a brand new 1uf cap.  Increasing the frequency to something like 10khz made it look good again.  From a couple Youtube vids I watched on DIY ESR testers, theirs usually was something like a 200khz test signal.
 

5pinDIN - 05 February 2017 09:51 AM

Thanks for that information. I did a Google search, which usually reveals if ElektroTanya has the manual, but didn’t get a hit. Based on your post, I went to their site, and found both the S90 and S90 ES service manuals.

Yup, all I did before was google, but searching within ElektroTanya found it.

Anyway, I think I’ve worked out the issue.  I paralleled a cap across each of the ones at the outputs of the noisy 3.3v regulators, and the ripple is all but gone now, including the ripple from the 5v feeding those regs.  That was at IC2 and IC46, and the caps in question were C9 and C124, both of which were 47uf 16v SMD electrolytics.  I stuck a 100uf 16v across them and noise is gone.  I’ll order some proper replacements, but until then I’ll test with this.  Of course I’ll have to make sure the thing boots up reliably now, but I’ll report back.

Thanks a TON for the help 5pinDIN; pointing me towards the 3.3v sections made a huge difference in getting this far this quickly.

tertiumsquid - 06 February 2017 01:31 AM

Yeah, the problem I had was finding a known bad cap to see what it would look like on my scope!  I tested all the electrolytics on the main board (service manual calls it the DM board so that’s what I’ll use from now on) but they all looked good.  Probably my scope doesn’t have enough resolution to see the difference in marginal caps - the smallest voltage that this scope can display is 10mv, which if my math is right, using a 200mv p-p test signal (for in-circuit testing) out of a 50ohm output would end up being 4mv per ohm of ESR.

The source impedance of your generator is too good.  :-)

Some series resistance helps. See This.

tertiumsquid -

Anyway, I think I’ve worked out the issue.  I paralleled a cap across each of the ones at the outputs of the noisy 3.3v regulators, and the ripple is all but gone now, including the ripple from the 5v feeding those regs.  That was at IC2 and IC46, and the caps in question were C9 and C124, both of which were 47uf 16v SMD electrolytics.  I stuck a 100uf 16v across them and noise is gone.  I’ll order some proper replacements, but until then I’ll test with this.

I’m not a big fan of surface-mount devices, in particular electrolytic caps. I’ve often replaced them with radial lead ones with the leads trimmed short, as long as there’s room. It may not be original, but the problem doesn’t recur.

tertiumsquid -

Of course I’ll have to make sure the thing boots up reliably now, but I’ll report back.

Thanks a TON for the help 5pinDIN; pointing me towards the 3.3v sections made a huge difference in getting this far this quickly.

You’re welcome. I hope that nailed it, but please do get back to us, either way.

------- EDIT - The above was accidentally deleted, now restored -------

How are things going? In case the problem isn’t resolved yet…

In my experience when you find more than one SMD cap of a particular value and voltage bad, there’s often a good chance that others of the same type are also failing. So with two 47µF/16v ones apparently bad, I’d check the rest like them.

5pinDIN - 06 February 2017 02:36 AM

How are things going? In case the problem isn’t resolved yet…

In my experience when you find more than one SMD cap of a particular value and voltage bad, there’s often a good chance that others of the same type are also failing. So with two 47µF/16v ones apparently bad, I’d check the rest like them.

I’ve been testing and playing it throughout the week, and I haven’t had a single problem since replacing the caps.  It’s been working beautifully.

I did replace all 3 of the 47uf caps after the 3.3v regulators.  I didn’t check the rest of the board for more 47uf caps, but I did check ripple across all of the regulators and power supply feeds and it all looks good right now.  If I have any other issues in the future, I’ll know what to check, but yeah I do agree that like-value caps can tend to fail together, especially SMD.

I just checked the BOM in the service manual, and there’s apparently 13 more of that same part number in the S90.  From what I can tell, none of them are used in the main PSU, and none in the 5v regulator section of the DM board.  I found a bunch used as coupling caps in the audio in/out section, so those are not as likely to fail I think, though that only accounts for 9 of the 13.  I’m too lazy to look for the rest in the 6-page schematic… At least if I have to do any other work on this in the future, this is my own keyboard, and it is one of the easier keyboards I’ve worked on in a while - it comes apart quite easily.

tertiumsquid - 12 February 2017 05:47 AM

I’ve been testing and playing it throughout the week, and I haven’t had a single problem since replacing the caps.  It’s been working beautifully.

Great!

 

tertiumsquid -

I did replace all 3 of the 47uf caps after the 3.3v regulators.  I didn’t check the rest of the board for more 47uf caps, but I did check ripple across all of the regulators and power supply feeds and it all looks good right now.  If I have any other issues in the future, I’ll know what to check, but yeah I do agree that like-value caps can tend to fail together, especially SMD.

I just checked the BOM in the service manual, and there’s apparently 13 more of that same part number in the S90.  From what I can tell, none of them are used in the main PSU, and none in the 5v regulator section of the DM board.  I found a bunch used as coupling caps in the audio in/out section, so those are not as likely to fail I think, though that only accounts for 9 of the 13.  I’m too lazy to look for the rest in the 6-page schematic… At least if I have to do any other work on this in the future, this is my own keyboard, and it is one of the easier keyboards I’ve worked on in a while - it comes apart quite easily.

The caps at the regulators see quite a bit of ripple current - since that tends to induce heating, they’re indeed more likely to fail than in other applications such as audio coupling. Of course, if you notice a loss of audio level, especially in the bass region, then the other 47µF SMD caps are certainly a good place to look first.

Have fun playing the S90…

Hi Tertium, Hi 5-Pin.

It seems I really need you’re help here.

I’ve got exactly the same problem.

I just bought a S90 from a guy who didn’t seem to have this problem before. (yeah right, got scammed here.)

As you seem to have solved this problem. Can you elaborate more on this? Is it SMT or through hole? I’m fine with soldering as long as it’s not SMT.

I really like to know what you suggest I do first, and how to maesure what cap I should replace.

If it’s SMT I will get somebody else to do it, I first will snag the service manual.

You guys will save my day, I looks like you totally know what you’re talking about and I just need some guidance here.

Reg,

Ksus, Amsterdam.

KSUS - 15 February 2017 08:34 AM

Hi Tertium, Hi 5-Pin.

It seems I really need you’re help here.

I’ve got exactly the same problem. [...]

Welcome to the forum.

Since this forum covers more than one model - the following applies only to the original S90, not the ES or XS versions.

The S90 SM is here:
https://elektrotanya.com/yamaha_s-90.pdf/download.html

The caps tertiumsquid replaced are SMD type, 47µF/16v, at the outputs of the 3.3 volt regulators on the DM board.

Although replacing those caps apparently fixed tertiumsquid‘s S90, that may or may not resolve the problem in yours. Best of luck.

Hi 5Pin,

Thanks for the quick reply, incredible.

It’s indeed the original S90 I’m having trouble with. Same errors, same way to surpass it.

I’d love to see a reply from tertiumsquid also how he exactly found out what caused the problem, but I can imagine since his S90 is operational he won’t check the forum lately.

Is there a way to detect the ripple with that $10 tool you referred to?

Reg,

Kees

KSUS - 15 February 2017 10:42 AM

Hi 5Pin,

Thanks for the quick reply, incredible.

You’re welcome.

 

KSUS -

It’s indeed the original S90 I’m having trouble with. Same errors, same way to surpass it.

I’d love to see a reply from tertiumsquid also how he exactly found out what caused the problem, but I can imagine since his S90 is operational he won’t check the forum lately.

Of course I don’t know if tertiumsquid will reply, or if that will be in a timely manner. (You might try sending a Private Message.) Although I can’t speak for him, I’ll provide as much information as I can.

 

KSUS -

Is there a way to detect the ripple with that $10 tool you referred to?

The link I provided was primarily to show how a scope and square wave generator could be used to check a cap’s ESR (Equivalent Series Resistance), which can be done with the S90 not powered on, and might find bad caps while they’re still in circuit. I didn’t mean to endorse the product, since I’m not familiar with it.

In my experience, the best way to evaluate the caps is by looking at how smooth the DC is at the regulator outputs, with an oscilloscope. The high ripple tertiumsquid found (see previous posts in this thread) was a good indicator that the caps were no longer up to the task.

I hope that helps.

Thanks again 5Pin.

I’m pretty happy with the SM, I will find it out.

I will replace the the caps at IC46, IC47 and IC2. (after measuring them.)

I have a feeling it will fix it. I will replace them with radial lead capacitors if it got the space. If not, youtube is providing enough help to fix it this weekend.

Can’t wait to rock the S90:) - it’s already working perfectly when it’s booted, but it’s doesn’t feel good knowing there’s a failing part(s) inside.