Strange CM-64 Problem

[Alkarion]

Last night I played Wing Commander with Dosbox with the MIDI data being transferred via MIDI-Ox to my CM-64. After playing for half an hour or so, suddenly there was a screeching noise and the device fell silent (except for the noise). It's not a continous background hiss but sounds more like the distortion sound of a sound which is clipping. I switched off the CM-64 and switched it on again without leaving Dosbox. The problem stayed. I also remember that it then fell completely silent but that some instruments were played again after some time.

I don't know what exactly I did then since the thought of losing my CM-64 made me quite nervous. This morning I made some experiments. First of all I wanted to find out if Dosbox or perhaps even MIDI-Ox was the culprit. So I listened to the entire KQ5 soundtrack by playing the MIDI file from this site. Nothing unusual happened. Then I tried Dune 2 in Dosbox and played for over an hour when suddenly wrong instruments played. I guess it sounded as if suddenly all custom patches had been lost or replaced by standard instruments (Dune 2 has many custom sounds). Though I'm not certain, custom sounds may have been the only problem.

So nothing is ruled out yet and I have to further look into this matter. I'm a bit skeptical that it's Dosbox's fault, though (The only explanation could be faulty SYSEX messages sent by Dosbox at random points in the game). Can anyone experienced with defective Roland LA synths recall a similar problem? Is it possible that there is a hardware defect though it's playing allright most of the time? Are there components which may show their faults only after a longer running time?

[Alkarion]

I made a little experiment to test one of my suspections. I loaded the Wing Commander patches onto the CM-64 and left it on over night. And indeed, much to my dissatisfaction, when I played a Wing Commander Midi file, the custom patches were replaced by standard instruments.

So, it seems my fears have been confirmed and my CM-64 is indeed defective. Perhaps some refreshing mechanism for the RAM has gone bad.

[Tom]

I would think it's much more likely that this is a software problem, perhaps sending a reset or something, rather than the module "leaking" the custom patches.  If you send one of the SysEx banks on this site for a Sierra game, then load the appropriate MIDI file, does that work?  What software do you use to playback MIDI files?  Try something else, instead and see how that affects the module.

[Alkarion]

If you send one of the SysEx banks on this site for a Sierra game, then load the appropriate MIDI file, does that work?

Yes, absolutely. If I load the SysEx and play MIDI files everything is as it should be. However, I just discovered that this leaking of custom patches happens quicker than I thought. I can play a MIDI file after sending the SysEx, wait for some minutes and play the same file again and the memory patches are replaced by standard instruments. I did not transmit anything during the wait time.

Also, I tried to load the custom patches in Midi Quest XL.  I don't know if I'm doing anything wrong right now, because this still worked yesterday. But today, if I try to load the entire bank from the module, Midi Quest stops at 2070 bytes with the transfer.

What software do you use to playback MIDI files? Try something else, instead and see how that affects the module.

I'm using Windows Media Player right now, but looking at the Midi-Ox monitor, I can check if SysEx is transmitted. And if the same MIDI file is played right first and after some time, without doing anything in between, is played wrong, then I assume the module is the culprit.

[Tom]

I know there used to be some MIDI utilities and playback programs around that would send reset commands, or would otherwise affect the MT-32 in a way that when loaded, they cleared the memory of the MT-32.  Try using Cakewalk or some other MIDI sequencer software, instead of what you're using now...see if that helps.  I'm still leaning towards this being a software, not a hardware issue.

Here's an old version of Cakewalk Pro that should have no playback issues for the CM-64:

http://66.49.226.244/utilities/cakewalk3.zip

Is there a chance that there's a power issue with the CM-64; perhaps the memory is being cleared because the power isn't being maintained properly to the unit?

[Alkarion]

Well I haven't yet tried it with other software since I have gathered further evidence for a hardware problem. I uploaded the Wing Commander patches and then downloaded the entire memory bank in Midi Quest. So far there was no problem, the whole patch bank with all the samples showed up.

Then I waited half an hour, doing nothing in between (I ate my breakfast , and tried to download the memory bank again in Midi Quest. And, surprise, it was empty. So assuming that Midi Quest is working correctly, the CM-64 forgot the samples in the waiting time. Also, I had intermittent problems with downloading the memory bank. It would stop at random times, with the CM-64 stopping to send data.

[Alkarion]

I just uploaded something which the CM-64 put out while I was not playing any MIDI files. It can be downloaded here. It's the strange noise I talked about earlier. Although this recording is less spectacular than the sounds which I encountered first. It's a rather muffled noise.

[HondaSiR]

Hmm, to me it sounds like a problem with the power adaptor. It seems the module is receiving a fluctuating electrical current. Perhaps if you try (borrow) another adaptor the problem may be solved?

[Alkarion]

Hmm, I checked the power adaptor, which is none-Roland, but of high quality, and it says it is able to deliver 1000mA at 9V which would not be enough to meet the CM-64 specifications. On the other hand, I used this adaptor for quite a while now with the CM-64 (over a year) and haven't encountered any problems so far.

I will see if I find another one giving capable of providing higher currents. Meanwhile, I recorded a bit of the Willy Beamish soundtrack as played by my CM-64.

Willy Beamish.mp3

[Great Hierophant]

Hmm, I checked the power adaptor, which is none-Roland, but of high quality, and it says it is able to deliver 1000mA at 9V which would not be enough to meet the CM-64 specifications. On the other hand, I used this adaptor for quite a while now with the CM-64 (over a year) and haven't encountered any problems so far.

I hate to be obvious, but the CM-64 requires an adapter that can output 9V, 1200mA.  All CM modules are very, very rugged, so I would first try replacing the power adapter with one that meets the spec.  The usual Roland adapter is that spec, and you can get random results when the system isn't getting what it needs.

[Alkarion]

I just did exactly that. Let's see if it helps. Though, having had no problems for a long time now makes it seem less likely.

[Alkarion]

All CM modules are very, very rugged, so I would first try replacing the power adapter with one that meets the spec.

You may be right since I had no problems so far with the module being powered by the new adapter. Though I can't be sure since the problem would often show up only after prolonged use. I'm still puzzled by this since the adapter and the module worked excellently so far together.

Do you think the CM series is more reliable than the MT-32? Because we have seen defective MT-32 units here, have we not?

[NewRisingSUn]

I would assume that you're seeing more defective MT-32 than CM modules because the MT-32 is far more common. Other than that, the CM modules should potentially last longer because they use mask ROMs, while the original MT-32 uses EPROMs, which tend to lose their data over time.

[Great Hierophant]

Other than that, the CM modules should potentially last longer because they use mask ROMs, while the original MT-32 uses EPROMs, which tend to lose their data over time.

Don't most, if not all Roland LA synth modules use an EPROM for the control ROM?

[NewRisingSUn]

The CM-32L doesn't, and neither does the MT-32 after serial #851400.

[Cloudschatze]

The CM-32L doesn't, and neither does the MT-32 after serial #851400.

The four "newtype" boards I've seen all had EPROMs. Given Roland's track record, I can understand why...

[NewRisingSUn]

Usually, you only change from EPROM to mask ROM once development is complete. So they used EPROM until serial #851400, then switched to mask ROM, then decided they wanted to do some changes after all, and switched back to EPROM, probably around serial #900000.

[Alistair]

Man, what's up with that CM-64?

That MP3 was terrible. Not to mention all the instruments are wrong (I guess you didn't send the SysEx).

- Alistair

[HondaSiR]

Man, what's up with that CM-64?

That MP3 was terrible. Not to mention all the instruments are wrong (I guess you didn't send the SysEx).

- Alistair

Simple. For quite some time Alkarion used a power adaptor that has a lower output rating the CM-64 needed. When he used it for extended periods of time, the adaptor overheated and could not provide the required voltage. The module was literally being turned on an off in a fast manner, thats why the patches were being lost.  Aren't you glad you upgraded your adaptor back then Alistair? It could really be a fire hazzard.  

[Alistair]

I still use the adaptor, I just use it for my SC-55mkII now. My MT-32 uses a higher level one.

Yes, I'm glad. One of the reasons I love this community, smart helpful folk who can provide timely advice!

- Alistair