[SOLVED] Stuck on a dead end trying to fix a Roland D50 keyboard

[gkmaia]

Another chapter on my d-50 fix quest… trying to find the source for the cracking in sounds

If both custom ICs reverb and chorus are set to ZERO output I still get the noise at TP3 (test point).

I did also probed every pin on chorus and it's RAMs and they either high, low or pules within 0-5v peak. Haven't got anything lower than 5v or something that did not look like a digital pulse. Most at the micro second scale.

So far I have replaced all RAM, PCM54 and HC174(low bit DAC).
Swapped op amps IC12 and IC13 op amps to make sure it wasan't IC12 the bugger.
Did check all the 5v and 12v rails also checked if reverb, chorus, rams and all ICs, diodes, transistors on the DAC block we being fed the right voltages. And they are all fine.
Did also check for any ripples on CN1 and CN9. And power seems to be stable at the right voltage. Does not look like a power issue.
Swapped IC17 (HC139) with IC21 in order to eliminate an issues with it. Still same noise.

Any ideas that does not involve buying a new board?

I would like to go as far as I can here so I could also learn a bunch more.

Did lots of probing. Here is some logic data flow scoping:
1 - pd4066 - switch input from q4
2 - hc139 CH1 output to q4
3 - hc139 CH1 out to chorus latch clock
4 - hc139 - CH1 Output - to hc174
5 - x1 - 12mhz out (clock for CPU)
6 - x2 - 32.7 MHz out (clock for synthesizer chip)
7 - reverb ram RAS

[shakalnokturn]

Cracking only in sounds or also cracking when silent?
Could you better describe the symptom? Maybe give a waveform of the noise at TP3 with and w/o sound played.

[gkmaia]

Cracking only in sounds or also cracking when silent?
Could you better describe the symptom? Maybe give a waveform of the noise at TP3 with and w/o sound played.

Here you go!

There is noise when it is silent, but very little. Although could be picked up by the scope.
If you increase the reverb and chorus balances when silent the noise increases considerably.

But removing all the chorus and reverb does not fix the noise completely, neither fix the cracking when playing keys.

Video with cracking sound



Also a couple scans.

First scan has the chorus, reverb and main output set to 00. Obviously no sound at TP but the noise wave is there. It is this amount of noise acceptable, meaning normal?

Second scan has the chorus, reverb and main output set at default levels. The noise is amplified and when a key is pressed the cracking sounds.

[poot36]

If possible see if the issue is still present with the MIDI input.  Also try playing a square, sine, or triangle wave if possible to see where in the wave it is distorted.  I would also check the direct output of the DAC to eliminate the op-amp as the issue.  Also check all transistors and resistor pullup packs in that area.

[gkmaia]

Also try playing a square, sine, or triangle wave if possible to see where in the wave it is distorted. I would also check the direct output of the DAC to eliminate the op-amp as the issue.

I have designed a square wave sound that was only using Upper Partial 1. So it is a clean patch and it sounds 99% perfect.

Did then a bunch of scans comparing the signal out of TP5 Upper signal with some key points.

I also found the following noise pattern with Chorus. If I select options 3 or 4 the noise spikes considerably as per the video bellow.

[poot36]

Those are some interesting wave forms.  I would hook the logic analyzer up to the DAC input and see if there are any weird looking signals when you play the square wave.  The frequency of the digital signal should decrease by half or so on each pin depending if the data is LSB or MSB first.

[gkmaia]

thanks poot36!

I will probe that next! In the mean time I did probe the PSU outputs +5v and +12v that feed the motherboard.

Some images of the scan. As my probe attenuates 10x. So each scale should represent 50mv. I am getting up to 100mv ripples on both 12v and 5v outputs.

I have no knowledge to determine if that is satisfactory.

[shakalnokturn]

100mV pk on the +5V is certainly more than I'd like to see, besides it could actually peak quite a bit more than your scope shows.

Have you checked the -12V rail, I don't think you have mentioned that one at all unless I read too fast.

At this point I'd give it a try replacing a couple of decoupling caps, say some around the IC4, IC10, IC13 area and some around the ones that consume the most current (the warmer ones) maybe IC27, IC28, IC31?

Maybe have a close scoping at what's going on at pins 5 and 13 of IC4 before and after replacing some decoupling caps. When you fix something it's always nicer to actually understand how it was fixed and what was causing the problem.

[poot36]

If those scope readings are correct you do have some major issues.  The 12V line looks the worst out of all of them.  The scope is even measuring a frequency within the hearing range on it.  About how old is the keyboard?

[gkmaia]

hi shakalnokturn,

-12 was pretty much the same. Those caps are pretty old. Will need to check the ESR of them soon.

I did one interesting test. I removed R1 and C10. I cant seem to get a read of C10, seems faulty but not sure it that make much of a difference considering the issue I have. But will replace it anyway.

If you have seen my previous video you will notice there is considerable background noise while there is no sound. Just by changing chorus parameters. Of course when I produce a sound the sound gets "damaged" by whatever this cracking background noise is.

By removing R1 I stopped the demultiplexer from enabling the chorus IC8 to output data. There is no sound and background noise is totally gone while ICs 10 (DAC) is still receiving some random data (see photo) that seems to be regular pulses that cannot be translated into analog sound.

I could be wrong but I have a few insights:

• the whole circuitry around ICs 10(new), 11(new) & 12(new) (DACs & opamp) are likely not to blame as there is no background noise after chorus IC8 is disabled.
• as the synthesizer IC31 is still able to generate sound waves and all it's programming software parameters are still working and changes to it does seem to affect the sound she may be ok.
• as the programming software parameters of chorus IC8 and sometimes reverb IC9 seems to have a much clear impact on noise and cracking I am suspecting I may have an issue with one of these guys. More likely to have a problem with my chorus IC8.

I am attempting to re-enable chorus IC8 and then disconnect pins 25 to 42 hoping to re-enable sound output but at the same time disabling the connection between chorus and DAC, therefor, only getting DAC to convert data from reverb IC9.

The difficulty I am having with analysing the data from the logic analyser is that I cannot identify glitches, clocking issues or high or low shorts as it is hard to know how the IC should behave. Best I can do is look at the datasheet and see ok, that pin deals with data so comparing to other pins that do data it seems to be doing something similar.

[gkmaia]

hi poot36,

It is a classic Roland D50... pretty dam old. 40 years almost.

[poot36]

If this: https://en.wikipedia.org/wiki/Roland_D-50 is the keyboard we are talking about it is not 40 years old.  The DAC is just a resistor ladder with offset voltage applied to it.  There should be some sort of output from it (possibly ultrasonic) even with the chorus disabled.  You could also try disabling the chorus by removing the CS input and see if that results in different data on the DAC.  The cap you could not measure is 10pF so it is very hard to measure due to its low value.  It is most likely fine.  What ROM version do you have?  If it is below v1.07 that may explain your issues with the chorus sounds.  Here is a link to the service manual: http://r.duckduckgo.com/l/?kh=-1&uddg=http%3A%2F%2Fwww.polynominal.com%2Fsite%2Fstudio%2Fgear%2Fsynth%2Froland_d550%2Froland-d50-service-manual.pdf in it you can find a way to enter test mode to see if the ICs are at fault.  I hope this helps you.

[gkmaia]

Hi poot36,

I got the service manual that is where the schematic came from and have done all the troubleshooting procedures from the manual and no issues were found. 

Running the latest CPU with ROM 220.

I didi a bit more reading on CS - SS slave select.

Thanks for the tip! I think as per the japanese on the manual when it is active pin 7 goes LOW. If I disconnect will it think it is always active then?



Thanks

[poot36]

Any time you see a chip pin label with a bar over it it indicates that the signal is inverted from the standard signal inputs on the chip.  In this instance a low signal (0V) would activate the CS line.  Do you have access to a analog scope?  If so set the keyboard to loop a continuous sound and look at all the data lines going into the DAC for anything strange looking.  I guess that it is a good thing that it passes all self tests but I wonder how those are run.  It is weard having the DAC on the data bus like that.  I wonder if there could be some other chip on that bus that is been enabled or outputting when it should not be.  Have you checked the PCB for hairline cracks or bad solder joints?

[gkmaia]

So if LOW activates it then removing pin 7 would keep it stuck on LOWm therefor active right? Would not produce the results we are expecting?

Just did this test and it is smart. It did not boot. Connected it back and it booted fine.

I did check every single trace of the chorus all the way to its final destination and touched up solders on every component from Ics to SMD caps and res. Have replaced all RAM ICs. Checked traces as well.

There should be a way to by pass or hardware deactivate the Chorus so I can be 100% sure it is the bugger...

Also my beginner knowledge does not help much. But anyway, just by fiddling with this thing and getting these tips I am already learning heaps.

I bough an analog but it is faulty. Got to fix it.

[shakalnokturn]

If you don’t want to go to the trouble of checking / replacing the decoupling caps just yet and your scope has 2 channels, you could at least check the audio glitches against the 3 power rail glitches. That may at least rule out the unclean power or confirm it is time to recap.

[gkmaia]

Any time you see a chip pin label with a bar over it it indicates that the signal is inverted from the standard signal inputs on the chip.  In this instance a low signal (0V) would activate the CS line.  Do you have access to a analog scope?  If so set the keyboard to loop a continuous sound and look at all the data lines going into the DAC for anything strange looking.  I guess that it is a good thing that it passes all self tests but I wonder how those are run.  It is weard having the DAC on the data bus like that.  I wonder if there could be some other chip on that bus that is been enabled or outputting when it should not be.  Have you checked the PCB for hairline cracks or bad solder joints?

I did some scoping on pin 7 the CS. And it is stuck on HIGH (5v) all time. No matter what I change in the software it is always high.

Does that mean anything useful?

[poot36]

That may indicate a issue depending on how the chip is used.  It is tied into the reverb so it may be used to communicate between the two chips.  I would boot it with the CS connected and then disconnect it after it has powered up and see if the high is coming from the gate array or the Chorus chip.  The CS line should toggle when you run the built in tests I would think.  I would think that it would pulse at least during start up of the keyboard due to the keyboard not booting if it is disconnected.  It is possible that it is to fast of a signal for your scope to capture and that is why it always looks like a high.  Another test you could try is reconnect every thing and check that all bits of the DAC are pulsing.  If one is not that could cause the sound you are hearing.  I don't know if this is useful but I will share this link as well: http://midibox.org/forums/topic/11435-inside-a-roland-d550/  Is it possible that the MSB adjust pot is not set correctly?  It could be corroded and not making good contact.  I don't know if you have done this or not but scope the output of the DAC and the output at TP3 and compare the waveforms.  This is turning into an interesting repair.

[gkmaia]

I reconnected everything already. I am a bit reluctant to raise the pins again. It is causing damage to the legs and pads. I am more inclined to severe the trace and reconnect again later instead of be fiddling with the IC legs. What are your thoughts on that? Is it worth the damage considering the outcome of the test?

I suspect the chorus clocks at 32Mhz and my scope can only see 10Mhz or 0.1us so I can't really see it. When I try to go under 1us I start getting some triggering issues.

The MSB adjustment has been done when I did all the troubleshooting procedures on the manual and the pot do respond well.

I will do the DAC out put and compare to TP3 as well as scope all DAC input pins and will post here.

And yes, it has been very nice for me as I am learning heaps.

[poot36]

At this point I would not mess with the pins on any of the SMD mounted chips due to the age and construction of the board.  You can always cut traces and jumper over them as long as the iron is set correctly and does not over heat the board when doing so.  It would be nice to have a working keyboard to check against.  I wonder if this is normal for this keyboard due to its design.

[gkmaia]

Here are some scans of the DAC input under different situations

sound 55 clean no chorus

Is a sound that do no use chorus. So chorus balance and depth are set to 00. It sounds perfectly clear.

sound 11 no sound

When you boot the keyboard without an internal battery it corrupts the internal memory patch the CPU cannot generate a sound. The question is there is no sound because:
- these pulses mean nothing to the DAC
- as the internal memory is corrupted the whole system goes to a halt

 sound 11 noise

That is a sound with lots of noise and chorus depth balance >0. Most of the sounds are like that and pulse to the DAC like that.

Can you make anything useful out of it?

[poot36]

Interesting how some of the inputs never seem to change.  I would try some different sounds to see if you can make them change.  If you cannot then you may have found the issue.  I would check all of the DAC data lines with your multimeter set to diode mode between the ground and each data line with the leads in both directions.  This will check the protection diodes in the chips.  You can also check to the power rails of the chips instead of ground as well.  If you see any difference between the data lines then you found your issue.

[gkmaia]

Can you pin point one screen and channel you would like to see changing with different sounds?

[poot36]

On both sets of measurements pins 2-9 and 10-17, channel 8 on the logic analyzer never changed.  Bad cable or connection perhaps?

[gkmaia]

Oh sorry. My bad forgot to tell you channel 8 was not plugged.

Any other screen and channel you would like to see changing with different sounds?