derSammler wrote on 2020-06-09, 19:46:

Keep the 1.21 and sell the 1.01.

Dumping the ROM to use either version won't work, as the sample ROM and another IC are different, too. You can't just change the firmware only, apart from minor steps, like 1.20 to 1.21.

Good to know. Thanks .

Wish I could get my hands on an SC-55 MKII as that has 28 voice polyphony vs 24 the MKI (GS only) has. I updated the firmware to 1.21 and labeled both chips the firmware revision numbers.

Can this be identified from the outside? I have a me 1 and mk2 but rarely use them as I also have scva.

Wish I could get my hands on an SC-55 MKII as that has 28 voice polyphony vs 24 the MKI (GS only) has

I don't think the polyphony makes any difference for games. They knew that the original SC-55 was limited to 24 voices, so pretty much all composers took care of not using too many voices. When I did listen to a larger collection of MIDI files on my different modules a while ago, I even found some MIDI files that sounded wrong when the polyphony was too high, as some composers apparently forgot to stop notes, since those notes were cut off automatically on devices with only 24 voices and they never noticed them still playing.

Also, 24-voice polyphony is General MIDI v1 standard. While many (well, all apart from the SC-55 actually) MIDI modules and wavetables in the early 90s already offered more voices, that was still the value you had to work with, as GM only guaranteed 24-voices to be available.

Can this be identified from the outside? I have a me 1 and mk2 but rarely use them as I also have scva.

Yes, just read the beginning of the topic to see how you can display the firmware version.

Some of the MIDI files I have, like RuneScape's original theme song or the second version of it, either don't hold the notes like it's supposed to or no rhythm is playing if the instrument style shows as unknown.

RuneScape? Isn't that a MMORPG from 2001? A bit out of time for a SC-55, don't you think? For MIDI files that late, you need 32-voice polyphony and something better than the SC-55 anyway.

derSammler wrote on 2020-06-10, 06:29:

I don't think the polyphony makes any difference for games. They knew that the original SC-55 was limited to 24 voices, so pretty much all composers took care of not using too many voices.

Actually there are games where even the 28 voice polyphony of MKII is not enough and can cause problems. It's mainly because the "Voice Reserve" algorithm intended to distribute and prioritize the available polyphony across the channels is not always optimal. The default values reserve nothing for the 11-16 channels, making them cut off the sounds on many midi files if channels 1-10 take all 24/28 available sounds.

Warcraft 2 is an example for this problem on a real SC-55:
https://www.youtube.com/watch?v=0ndCckv-lww
It can be heard in the first 10 second that the Bell cuts off (it's on channel 13 ).
HUMAN1GM.MID file to test on SC-55:
download/file.php?id=25153

Are you sure that's not just a bug in the SC-55? There are only 5 channels playing and I can't quite believe these would utilize all 24 or even 28 voices.

derSammler wrote on 2020-06-10, 17:03:

Are you sure that's not just a bug in the SC-55? There are only 5 channels playing and I can't quite believe these would utilize all 24 or even 28 voices.

Of course it can be considered as a "bug" of the default voice reserve algorithm. But it's not a bug in the sense that it can be cured by modifying the default voice reserve settings of SC-55 (at least on MKII). If you distribute voices evenly among channels the problem is gone. Here's a SysEx by James-F that does exactly this:

F0 41 10 42 12 40 00 7F 00 41 F7
F0 41 10 42 12 40 01 10 02 02 02 02 02 02 02 02 02 02 02 02 01 01 01 01 13 F7

Yellow is default /Green is modifications by James-F

Also you can try to mute Channel 9 (Harp) or Channel 2 (Timpani) as a test to cure the Bell sound cut problem .

Cloudschatze wrote on 2016-09-26, 15:30:

Jolaes76 wrote:

...which SC-55 model is more compatible with more DOS games "outta box"...

With that criterion, and given currently known information, an SC-55 with either ROM version 1.20 or 1.21 provides the greatest degree of "out-of-the-box" compatibility with DOS games. These units shipped with "GS" logo faceplates, and with serial numbers greater than AC94800.

Sorry for the necroposting, but why SC-55 v1.2x and not v2.00? What does v2.00 "loses"? Is it the whole CTF fiasco? Or the GM reset? Or what?

Am386DX-40 wrote on 2020-09-18, 02:11:

Sorry for the necroposting, but why SC-55 v1.2x and not v2.00? What does v2.00 "loses"? Is it the whole CTF fiasco? Or the GM reset? Or what?

The program change behavior when selecting drum sets differs slightly between the v1.2x and v2.00 SC-55 units (thats v2.00, not mkII; the latter of which introduces even further change), breaking compatibility with the improperly-programmed music of at least two games that I've identified thus far - TFX and Inferno. Unless you happen to care about those two games, I wouldn't be concerned with this difference at all.

Cloudschatze wrote on 2020-09-18, 18:12:

Am386DX-40 wrote on 2020-09-18, 02:11:

Sorry for the necroposting, but why SC-55 v1.2x and not v2.00? What does v2.00 "loses"? Is it the whole CTF fiasco? Or the GM reset? Or what?

The program change behavior when selecting drum sets differs slightly between the v1.2x and v2.00 SC-55 units (thats v2.00, not mkII; the latter of which introduces even further change), breaking compatibility with the poorly-programmed music of at least two games that I've identified thus far - TFX and Inferno. Unless you happen to care about those two games, I wouldn't be concerned with this difference at all.

I believe the 2.00 units fix the GM reset being interpreted as GS reset issue, resulting in issues with some Build engine games. However that's based on reading other posts here, not my own first-hand experience.

Also, my GS-marked 1.21 unit has serial AC94400.

Oetker wrote on 2020-09-18, 18:26:

I believe the 2.00 units fix the GM reset being interpreted as GS reset issue, resulting in issues with some Build engine games. However that's based on reading other posts here, not my own first-hand experience.

I'm not sure that I'm familiar with the Build issues. Hypothetically though, a sequence containing NRPNs but headed with just a GM System On message will fail to render properly, or at least per the composer's intent, on a v2.00 SC-55 or later Sound Canvas.

Also, my GS-marked 1.21 unit has serial AC94400.

My statement of ROM v1.2x applying to units with serial numbers >AC94800 is based off of information in a Roland service bulletin. It's entirely possible that their information is slightly off, or that the mainboard in your unit was swapped-out at some point. I have a v2.00 unit with a serial of ZC20975 and a "GS Standard" faceplate, incidentally.

Is there a Wiki or flow chart or something that has compiled all the great information in this thread to make it more understandable to the casual computer midi user like me? I am in the market for a SC-55 myself and have a chance to get a MK1 with GS logo only. Want to make sure I am not wasting my money. If all I do is play the usual DN3D, Doom, Blood, and Sierra/Lucasarts etc. will this be perfectly adequate then?

fsmith2003 wrote on 2020-12-24, 20:16:

Is there a Wiki or flow chart or something that has compiled all the great information in this thread to make it more understandable to the casual computer midi user like me? I am in the market for a SC-55 myself and have a chance to get a MK1 with GS logo only. Want to make sure I am not wasting my money. If all I do is play the usual DN3D, Doom, Blood, and Sierra/Lucasarts etc. will this be perfectly adequate then?

yes! go for it if the price is right
you can't go wrong with any SC-55 - also I am a little biased I prefer the old gs-only units my self 🤣
but seriously it will work just fine , you might not get breath noise in the sewers level of dn3d but you'll get the click one that sounds like water drops it's not like you will get french horn playing

Is the SC-155 firmware the same as the SC-55?
I have a normal SC-55 (GM/GS logo version) which has ROM version 2.00, but I also have a Roland SC-155 (which also has the GM and GS logo, i think all of them do) which has ROM version 1.00. I think I would have to test with the dn3d midi file between those to see what will happen.
I recently bought a second cheap SC-55 with the GS-only front panel. It hasn't arrived yet but I am curious which ROM version that will have.

EDIT:
I compared the SC-55 with ROM 2.00 and the SC-155 with ROM 1.00 and they seem to behave exactly the same with everything I've tested so far.

Something else,
There was also discussion about polyphony between the SC-55 (24 voices) and the SC-55mk2 (28 voices).
I have an example that proves that the latter is not always better. I don't know why this difference is there, but I tink it has to do with either note priority differences or maybe that some sounds are made up of multiple layers (therefore using more voices) in the SC-55mk2 when comparing it to the original model.

Here are two samples of a part of the Settlers II in-game soundtrack which clearly shows note cut-off on the SC-55mk2 version where it does not on the normal SC-55.

Does anyone have a copy of the old freeware MIDI Voice Reserve Analyzer so we can determine the optimal voice reallocation?

Salient wrote on 2020-12-24, 22:23:

Here are two samples of a part of the Settlers II in-game soundtrack which clearly shows note cut-off on the SC-55mk2 version where it does not on the normal SC-55.

Damn, that track sounds awful on the mk2.

I can confirm that it does play correctly on the SC-155. Here's my recording if anyone wants to listen to the entire song. For reference, an Avance Logic ALS100 (not plus) was used for MPU-401 duties.

I just picked up an SC-55 mkII and peculiarly it has ROM version 1.00 instead of 1.01, which I think the latter is more common. I wonder what would be changed between the two. From reading this thread, it's probably something that I'll never encounter. Looks like it's dated 03/17/1993 and the number sequence is 300300300. I'm assuming flashing another ROM chip with 1.01 would work, and it appears to be socketed. Wondering how worth the trouble it is.