MT-32 vs. CM-32 et al.

[Cloudschatze]

I seem to recall the LAPC-N ROM being soldered. It's probably the same situation for the CM-32LN as well, but I'll double-check later.

Both have socketed ROMs.

[NewRisingSUn]

What CPU-relevant differences exist between the CM-32L and the CM-32LN/CM-500? The CPU controls the MIDI IN/OUT ports and the LA and Reverb chips. Have any of those been changed?

The SpaceAlvin issue is a difference between the old-type MT-32 and the new-type MT-32 and later models, not between the CM-32L and the CM-32LN/CM-500. (If you look at my update to the Wikipedia article, you'll see that I divided the models into three "generations", to get some structure in this. What do you think?)

Also, since you seem to have a digital camera: Wikipedia is notoriously anal about using images from other sources, so if you can, please take some images of the MT-32 and the original MPU-401 breakout box and add them to their respective | articles. Thank you.

[Cloudschatze]

What CPU-relevant differences exist between the CM-32L and the CM-32LN/CM-500? The CPU controls the MIDI IN/OUT ports and the LA and Reverb chips. Have any of those been changed?

Not to my knowledge. The PCM ROM changed though - would that matter? I haven't a CM-32L, else I'd chance the ROM swap.

The SpaceAlvin issue is a difference between the old-type MT-32 and the new-type MT-32 and later models, not between the CM-32L and the CM-32LN/CM-500.

I know. My thought-process was still on possibly contacting Roland, and I figured it would be good to have a complete list of everything in need of repair.

(If you look at my update to the Wikipedia article, you'll see that I divided the models into three "generations", to get some structure in this. What do you think?)

I think what you've done is great, and I appreciate that you've taken the initiative.

Also, since you seem to have a digital camera: Wikipedia is notoriously anal about using images from other sources, so if you can, please take some images of the MT-32 and the original MPU-401 breakout box and add them to their respective | articles. Thank you.

I'll borrow a light-box this weekend. What type of photos are you hoping to see in the article? External shots? Mainboard shots?

I don't own an original MPU-401, so I can't help there.

[NewRisingSUn]

Not to my knowledge. The PCM ROM changed though - would that matter?

Why did they change the PCM ROM? There doesn't seem to be an audible difference. I don't know how that ROM is addressed, but if the control ROM tells the LA chip to "start playing the sample from ROM position 0x43D21", it will matter if the samples' starting addresses have changed.

I haven't a CM-32L, else I'd chance the ROM swap.

The hex dump of the CM-32L ROM suggests that the ROM is *exactly* the same for the LAPC-I and CM-32L --- there's a string that reads "CM32/LAPC1.02 891205". Therefore, you might be able to use the one from the rev 1 LAPC-I.

What type of photos are you hoping to see in the article? External shots? Mainboard shots?

Just a nice external shot.

I don't own an original MPU-401, so I can't help there.

I thought you had everything and a hacksaw. Too bad.

[Cloudschatze]

Not to my knowledge. The PCM ROM changed though - would that matter?

Why did they change the PCM ROM? There doesn't seem to be an audible difference. I don't know how that ROM is addressed, but if the control ROM tells the LA chip to "start playing the sample from ROM position 0x43D21", it will matter if the samples' starting addresses have changed.

I'm not sure that the addressing has changed, but was referring to the physical chip itself.

I haven't a CM-32L, else I'd chance the ROM swap.

The hex dump of the CM-32L ROM suggests that the ROM is *exactly* the same for the LAPC-I and CM-32L --- there's a string that reads "CM32/LAPC1.02 891205". Therefore, you might be able to use the one from the rev 1 LAPC-I.

Hmm. Rev 1 cards have soldered ROMs, but I have a rev 0 card that I can borrow from. There shouldn't be any difference between the two... (Famous last words.)

I won't have time to try this tonight, but will give it a shot tomorrow and record the output.

[Cloudschatze]

Well, I've just tried the CM-32LN with LAPC ROM, and... there's nothing to report. Literally.[/b]

The CM-32LN failed to power-on entirely. I did test it in this condition though, thinking that perhaps the code for the indicator lights was different. No joy.

Normally, there is a ~1 second power-delay, which is supposed to protect against surges. I wonder if that might have something to do with the failure to work; perhaps some trigger that is supposed to be executed from ROM.

[Laust]

Does it have a dedicated power LED? does it light up?

[Cloudschatze]

Does it have a dedicated power LED? does it light up?

Yes.
Not with the LAPC ROM.

[Laust]

Then it sounds like something went seriously wrong. A power LED normally doesn't require triggering but is connected directly (through a current limiting resistor) to the voltage and ground lines and will turn on as soon as the unit does.

Both sound modules still work after this attempt?

[Cloudschatze]

Then it sounds like something went seriously wrong. A power LED normally doesn't require triggering but is connected directly (through a current limiting resistor) to the voltage and ground lines and will turn on as soon as the unit does.

Both sound modules still work after this attempt?

I placed the ROMs back into their respective devices, and they work just fine.

The following is from the CM-500 manual, and is likely applicable to the CM-32LN:

"This unit is equipped with a circuit protection device. A brief interval after power up is required before the unit will operate."

The same notice is not given in the CM-32L manual...

[Great Hierophant]

So much for replacing Control ROMs, they must be hacked into submission, which is probably far beyond the abilities of most people that regularly read this board.

[NewRisingSUn]

The hacking is not the real problem. If I knew where to get an Intel 8095/8098 assembler/disassembler, I would even try it myself. The problem lies in modifying the CM modules to accept EPROMs and burning them.

[Great Hierophant]

The hacking is not the real problem. If I knew where to get an Intel 8095/8098 assembler/disassembler, I would even try it myself. The problem lies in modifying the CM modules to accept EPROMs and burning them.

Those chips with the stickers on them in the sockets are standard EPROMs where the stickers protect the window from light.  That way they would be doubly easy to replace if the units needed to be recalled or upgraded.  I assume they contain the Control ROM code, and most, if not all CM modules have them.  MT-32s may be tougher because the early ones may contain two EPROMs, but the later ones may have soldered EPROMs.  

Those LAPC-Is that have ROMs, like mine, would be toughest of all.

[Cloudschatze]

...but the later ones may have soldered EPROMs.

I believe this is what NewRisingSun is getting at. The control ROM is soldered on the CM-500 mainboard.

Would it be any more/less difficult to retrofit a CM-32L with a PCM55 DAC than the proposed ROM modification, assuming the end result is the same?

[NewRisingSUn]

I assume they contain the Control ROM code, and most, if not all CM modules have them.

My CM-32L only has soldered mask ROMs, no EPROMS at all. The CM-500 has one EPROM, but that one is not the LA board's control ROM, but just houses the code that sets up both the LA and GS parts depending on the Mode A-D switch. The LA board's control ROM is a soldered mask ROM as well.

Would it be any more/less difficult to retrofit a CM-32L with a PCM55 DAC than the proposed ROM modification, assuming the end result is the same?

It will be a lot easier to just add the trimpot circuit to the PCM54 DAC.

The tough part about adding a trimpot is not adding the parts, but adjusting it properly (after all, a trimpot is basically a potentiometer for the purpose of adjusting things). The PCM54 data sheet has instructions on how to adjust the trimpot. You basically have to adjust it during operation while applying certain digital input values (either 8000 and 7FFF, or the digital representation of a -60 dB sine wave), measuring the analog output at the same time and adjusting the result using the trimpot's potentiometer.

With that in mind, adding the PCM55 might be easier after all. The pin assignment is different, of course, so that'll be a lot of fun.

[Laust]

Are you saying the PCM55 actually has a built-in trimpot sitting right there in the epoxy casing next to the substrate? that would be a first...

In fact, as far as I can gather from the datasheet, the PCM54 and 55 are both equally adjusted from the factory (laser trimming perhaps?), and beyond that the PCM55 simply cannot be adjusted. At all.

[Great Hierophant]

If you build the circuit referenced in the datasheet for the PCM54, which seems to be quite easy to do, a one-time test would give the proper value/position for the potentiometer.  Or you could adjust the value manually while a song that shows the effects of the Lineal Differential Error plays until the error isn't audible.

[NewRisingSUn]

Are you saying the PCM55 actually has a built-in trimpot sitting right there in the epoxy casing next to the substrate? that would be a first...

The original statement was based on a misunderstanding of the word trimpot. Correct would be to say that the PCM55 is factory-trimmed, while the PCM54 is not.
I understand that the data sheet claims that the PCM54 is factory-trimmed in the same way; in actual practice it seems that it isn't; that the absence of an external trimpot makes it sound worse than the PCM55.

On the other hand, knowing that adding the trimpot circuit without adjusting it properly would likely make the output sound worse, not connecting it at all might expose the PCM54 to some kind of interference with causes it to sound worse than the PCM55.
The first-generation MT-32 seems to be prone to a lot of interference --- first, I find it to be a lot more picky about using the correct AC adaptor. Also, the high-pitched noise in the MT-32 recording I posted a while back sounds exactly like the digital time slot select data (SH1-SH3) applied to the HD14051 demultiplexer. For some reason it bleeds through to the analog output in my unit.

[Laust]

I understand that the data sheet claims that the PCM54 is factory-trimmed in the same way; in actual practice it seems that it isn't; that the absence of an external trimpot makes it sound worse than the PCM55.

Or the absence of an appropriate grounding circuitry. The PCM54 datasheet says to ground the Trim/VPOT pin via an appropriate capacitor, and at least on my CM-32L and a new-model MT-32, this simply isn't done. It's just left dangling, although the pin is routed to a tinplated hole where a capacitor _could_ be put. Actually, both PCBs have unused holes and traces near the DAC, which makes you wonder if Roland had considered the trim circuit. Go figure...

On the other hand, knowing that adding the trimpot circuit without adjusting it properly would likely make the output sound worse, not connecting it at all might expose the PCM54 to some kind of interference with causes it to sound worse than the PCM55.

The biggest problem with adding the trimpot circuit is that it is not possible to get the MT-32 to output the necessary -60dB sine wave, so you're left doing the adjustment by ear. I wonder if it is possible to generate the appropriate sine via MIDI and a custom patch...

The first-generation MT-32 seems to be prone to a lot of interference --- first, I find it to be a lot more picky about using the correct AC adaptor. Also, the high-pitched noise in the MT-32 recording I posted a while back sounds exactly like the digital time slot select data (SH1-SH3) applied to the HD14051 demultiplexer. For some reason it bleeds through to the analog output in my unit.

How many MT-32 are you basing this observation on? If it's less than two, I would not rule out component failure. I don't have any first-generation (Large PGA LA32 chip, EPROMs for ROMs) units, but Maxime's old run-of-the-mill MT-32 had a high pitched noise problem which disappeared when I replaced the faulty 7805 voltage regulator.  Bad capacitors (electrolytic go bad with age) can cause audible interference too.

I searched the forum for the recordings of yours, but only found some done with an SB-16, all of which were horribly horribly noisy (not just the MT-32 ones) and didn't show the problem particularly well

[NewRisingSUn]

but only found some done with an SB-16, all of which were horribly horribly noisy (not just the MT-32 ones) and didn't show the problem particularly well

Yes, they do. The noise you're hearing in the KQ5 recordings is the actual LA units'; it has nothing to do with the SB16 (as bad as that one is). The noise is so loud because the MIDI file being played back is so quiet; the recordings have been heavily amplified to illustrate the units' different noise floor.
The high-pitched whistling can be heard on the MT-32 recording, not on the CM ones. Here are the links again:
MT-32 revision 0, ROM 1.07
CM-32L
CM-500, ROM 1.00

The "real-world" modification documentation mentions the whistling problem as well, so it might well be an inherent issue.