VOGONS


First post, by Tiido

User metadata
Rank l33t
Rank
l33t

CleanWave32 is Dream's sample set found on some waveblaster modules like MIDITEMP DS48, but it seems to have a bug that causes digital overflow effect resulting in awful cracking noises when sound gets loud enough.

Here's the distortion :
FLAC : http://www.tmeeco.eu/BitShit/PCschit/ElkmoonBad.flac
OGG : http://www.tmeeco.eu/BitShit/PCschit/ElkmoonBad.ogg

SAM9733 chip has its audio output format in software control, and software lives in the sample ROM so I hooked up an oscilloscope to the DAC to see maybe it is outputting wrong format to the DAC and I found that there actually is a problem. Namely BCLK polarity is wrong, new samples are output on rising edge instead of falling edge, no DAC out there can accept this format and the result is that data changes while DAC latches it. Due to delays stemming from analog factors it essentially means the data the DAC sees is shifted left by 1 bit position, losing its MSB and that results in the cracking sounds as sort of digital overflow happens.

CW32wrongBCLK.jpg
Filename
CW32wrongBCLK.jpg
File size
58.79 KiB
Views
484 views
File comment
Wrong bitclock
File license
Public domain

A fix is to add an inverter to the BCLK line, it will cure the distortion and will also reduce the loudness of the output by half, massively reducing clipping that happens in the analog stage after DAC on the WB card. Here's the fix :

CW32fix.jpg
Filename
CW32fix.jpg
File size
100.19 KiB
Views
484 views
File comment
inverter fix
File license
Public domain

And the result :

CW32correctBCLK.jpg
Filename
CW32correctBCLK.jpg
File size
68.89 KiB
Views
484 views
File comment
Correct bitclock
File license
Public domain

FLAC : http://www.tmeeco.eu/BitShit/PCschit/ElkmoonGood.flac
OGG : http://www.tmeeco.eu/BitShit/PCschit/ElkmoonGood.ogg

MIDI file used as a test : http://www.tmeeco.eu/BitShit/PCschit/ELKMOON.MID

CleanWave8 and GS sets do not suffer from this bug, it is strange such a thing went unnoticed or at least didn't get fixed...

Bonus fix on the MIDITEMP board is to change R11 and R21 to 4.7kohm, it reduces gain, clipping and also noise. I don't know why all these WB cards have so cranked up volume levels, there's almost nothing out there than won't clip with the insane levels used by most of these things...

T-04YBSC, a new YMF71x based sound card & Official VOGONS thread about it
Newly made 4MB 60ns 30pin SIMMs ~
mida sa loed ? nagunii aru ei saa 😜

Reply 1 of 4, by keropi

User metadata
Rank l33t++
Rank
l33t++

thanks for sharing this info Tiido - yes it is strange this went unfixed..!
then again Dream is known to *forget* many things including cool things they did in the past so it doesn't surprise me... this issue could be right when the company is sold and staff moved around
anyways, a late fix is better than no fix! 🤣

🎵 🎧 PCMIDI MPU , OrpheusII , Action Rewind , Megacard and 🎶GoldLib soundcard website

Reply 2 of 4, by kinkong

User metadata
Rank Newbie
Rank
Newbie

I have a Diamond Monster MIDI with the SAM9733 chip... I have noticed these unfortunate cracking noises/distortion happening particularly on loud notes. The difference is this board has a Dream GMS963200-B chip which is as I understand the Roland GS format 32 Mbit ROM and not the CleanWave32. Even though this is not CleanWave32 GMS973200, the crackling noises are very clear and consistent. Is it possible this bug exists on this card as well?

Is there any way I can modify my card similar to the fix above for the MIDITEMP DS48? But I'm not sure where the inverter would go.

Here is a picture of the board.

s-l1600.jpg
Filename
s-l1600.jpg
File size
329.89 KiB
Views
149 views
File license
Public domain

Reply 3 of 4, by darry

User metadata
Rank l33t++
Rank
l33t++
kinkong wrote on 2024-02-18, 06:34:
I have a Diamond Monster MIDI with the SAM9733 chip... I have noticed these unfortunate cracking noises/distortion happening par […]
Show full quote

I have a Diamond Monster MIDI with the SAM9733 chip... I have noticed these unfortunate cracking noises/distortion happening particularly on loud notes. The difference is this board has a Dream GMS963200-B chip which is as I understand the Roland GS format 32 Mbit ROM and not the CleanWave32. Even though this is not CleanWave32 GMS973200, the crackling noises are very clear and consistent. Is it possible this bug exists on this card as well?

Is there any way I can modify my card similar to the fix above for the MIDITEMP DS48? But I'm not sure where the inverter would go.

Here is a picture of the board.

s-l1600.jpg

You are likely hitting this other issue, which has a software fix also by Tiido.

Dream SAM9733 and SAM9407 based MIDI devices distortion fix

Reply 4 of 4, by kinkong

User metadata
Rank Newbie
Rank
Newbie

Running SysEx B0 63 37 62 07 06 4F as described in the fix from the other thread does reduce the volume level of the clicking, but the clicking is still present. I've also tried a lower value of 32 but it is still there albeit more quiet.

I've gone ahead and recorded a section of a MIDI which showcases the issue pretty clearly, compared to a Roland SCB-55 which does not have the clicking.

You do need to playback at around 100% volume with headphones, the popping sound happens every 1 second or so and it's sound is uncomfortable to the ears so I think it will be noticeable.

Setup:
- CHiLL V
- Roland UM-ONE
- MidiPlayer 6
- External USB DAC

Diamond Monster MIDI SAM9733 Roland clone soundbank GMS963200-B
- Clicking: (WAV)

Roland SCB-55
- No Clicking: (WAV)

MIDI File (found during 24 seconds to 34 second playback mark)
- JHELOM.MID

Possibly it is just my particular card... If anyone else has this card, or a Yucatan fx which I believe has the same chips, and can playback the Midi that would be helpful!