VOGONS


Sound Blaster 16 Bugs and Deficiencies Summary

Topic actions

First post, by James-F

User metadata
Rank Oldbie
Rank
Oldbie

Known Bugs and Deficiencies:

* MPU-401 Hanging Note Bug.
All Sound Blaster 16 and AWE32 that don't have a CT1747 chip have this bug.
DSP versions 4.04, 4.05 and 4.16 (AWE64) will not suffer from this bug.
The hanging note occurs randomly and sounds like a high pitched note that is not part of the midi music signal.

* MPU-401 Stuttering with high sampling rates.
In Duke Nukem 3D or Tie Fighter the game will stutter when using the MPU-401 midi interface.
It occurs with all SB16 to AWE64 sound cards, including Vibras, and those with the CT1747 chip.
Happens only when using high sampling rate digital audio (above 11kHz) along with the MPU-401 General Midi music.

* Single-Cycle DMA Clicking (Non Vibra).
A click will be heard after a sample is done playing with games that use Single-Cycle DMA mode for audio.
All DSP 4.0x to 4.13 are affected including the AWE32.
DSP 4.16 of the AWE64 is not affected, and is free of this bug.
The Vibra cards are not affected by this bug, but have the ringing bug instead.

Great Hierophant from Nerdly Pleasures has some examples of these "clicks and pops" as he calls them too:
http://nerdlypleasures.blogspot.co.il/2015/05 … t-playback.html

In this "SB16 programming" link the author states "This causes a click between each block," when talking about single-cycle DMA.
http://homepages.cae.wisc.edu/~brodskye/sb16doc/sb16doc.html

* Ringing/Hissing artifacts, Vibra only.
A faint high pitched ringing will be heard after a sample is done playing.
Sometimes the card will hiss instead of ring, but the cause it the same.
In any case, these "artifacts" can be heard only on the Vibra cards.

* Vibra CT2504 series Audio Distortion.
High volume sounds, mainly in DOS games, will distort the CT2800 and CT2890 Vibra cards.

* CQM vs OPL3.
If the card has a CT1978 chip on board, it has the Creative clone of the Yamaha OPL3 chip which sounds different.
Technically not a bug but a deficiency if one is looking an authentic OPL3 FM sound.
All AWE64 have the Creative clone chip built in.

* Self Noise
Self noise primarily comes from the DAC chip of the SB16.
CT1701 is the oldest and noisiest mainly on the SB16 CT1740 cards.
CT1703-T is slightly better but still considered noisy.
CT1703-TBS and CT1703-A are noise free an appear on the later SB16 cards including the AWE32 and AWE64.

It seems that most of the noise comes from the MIDI slider in the Mixer with cards using the CT1703-T DAC chip.
The MIDI slider is responsible for the Wavetable header and OPL volumes combined.

I attached an audio sample package with all the known bugs, so you can hear the them yourself.
SB16_Bugs_Sample_Collection.7z (15MB).

As a side note, the SB16 has dynamic brickwall lowpass filtering which changes depending on the sampling rate frequency,
unlike the SB2.0 and SBPro which have a simple static lowpass filter and is the same for all sampling rates.

Lowpass Filters.png
Filename
Lowpass Filters.png
File size
83.26 KiB
Views
18690 views
File license
Fair use/fair dealing exception

I welcome more findings, knowledge and sound samples for archival purposes on this thread.
Many thanks to Great Hierophant of Nerdly Pleasures Blog which was the main source of a lot of information in this post.

Last edited by James-F on 2016-11-05, 05:40. Edited 14 times in total.


my important / useful posts are here

Reply 4 of 111, by keropi

User metadata
Rank l33t++
Rank
l33t++

^ allright, I want to "fix" my CT2230's DSP and DACs , do you have the info necessary for that?

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

Reply 5 of 111, by awgamer

User metadata
Rank Oldbie
Rank
Oldbie

As far I understand the "DSP" is just an intel microcontroller with hard coded "DSP" software, so my immediate idea would be to simply try swapping in a 4.16 chip from a donor card. Of course, would hope it's the same micro controller, else the swapping becomes complicated getting the voltage and pinout hooked up correctly. Same deal with DACs.

Reply 8 of 111, by James-F

User metadata
Rank Oldbie
Rank
Oldbie

Turn it up.

I listen through flat frequency response studio monitors or in-ear monitors so I can hear everything.
The click is definitely there and clearly audible.

Last edited by James-F on 2016-09-24, 09:05. Edited 1 time in total.


my important / useful posts are here

Reply 9 of 111, by awgamer

User metadata
Rank Oldbie
Rank
Oldbie

It's supposed to be at the end right? It's very slight to me, and when you take the low quality recorded sound effects into account, it doesn't seem the biggest issue to me, but obviously it's annoying for those that pick it out.

Reply 10 of 111, by firage

User metadata
Rank Oldbie
Rank
Oldbie

Great to have some good examples!

A CQM vs. OPL difference is apparent in Doom E1M1 - the CQM's emphasis really grates after just a little while. The instrumentation is pretty simple, though (not as such, but one of the main pieces they likely tuned it for). There are probably some more glaring cases to compare out there, but I don't have a CQM card to make my own comparisons.

awgamer wrote:

It's supposed to be at the end right? It's very slight to me, and when you take the low quality recorded sound effects into account, it doesn't seem the biggest issue to me, but obviously it's annoying for those that pick it out.

No, the clicking is there throughout the Apogee fanfare in the Alien Rampage (edit: Alien Carnage 😀) example. Like small pops or clipping artifacts or something.

Last edited by firage on 2016-09-27, 12:59. Edited 2 times in total.

My big-red-switch 486

Reply 11 of 111, by dondiego

User metadata
Rank Member
Rank
Member
James-F wrote:

All AWE32 except the very first model (CT2760) have the clone

Not true, i have a non pnp ct3900 awe32 with the ct1747 chip which contains a real yamaha opl.

LZDoom, ZDoom32, ZDoom LE
RUDE (Doom)
Romero's Heresy II (Heretic)

Reply 12 of 111, by James-F

User metadata
Rank Oldbie
Rank
Oldbie
dondiego wrote:
James-F wrote:

All AWE32 except the very first model (CT2760) have the clone

Not true, i have a non pnp ct3900 awe32 with the ct1747 chip which contains a real yamaha opl.

You're right, fixed.

firage wrote:

No, the clicking is there throughout the Apogee fanfare in the Alien Rampage example. Like small pops or clipping artifacts or something.

Right.
Here is some more info on how it works:

Single-Cycle DMA Mode.png
Filename
Single-Cycle DMA Mode.png
File size
70.82 KiB
Views
18601 views
File license
Fair use/fair dealing exception

Some games like Alien Carnage will use continuous audio stream in single-cycle DMA mode, you will hear a click each time a sub-block of sound ends and an interrupt is sent.
Same thing can be heard in the Wolfenstein 3D Door example where the door opening sound actually constructed of several sub-blocks hence the many clicking sounds DURING the audio stream.
At least this is how I understand how it works.

Last edited by James-F on 2016-09-24, 10:49. Edited 2 times in total.


my important / useful posts are here

Reply 13 of 111, by keropi

User metadata
Rank l33t++
Rank
l33t++

The wolf3d door sound is a soundcard test I do when I get a non-sb card... most of the times it just cuts off, ie you only hear the first part of the sample. I also found this is true to several old SB clones (chinese clones - not compatible name chipsets like opti/crystal etc)

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

Reply 14 of 111, by James-F

User metadata
Rank Oldbie
Rank
Oldbie
keropi wrote:

The wolf3d door sound is a soundcard test I do when I get a non-sb card...

If you read the Single-Cycle DMA mode description I've posted it reads:

At the end of a sub-block transfer, the DSP will generate an interrupt to the application.
On receiving the interrupt, the application reprograms the DMA controller...

If the DSP (or DMA?) is plagued with this bug, there will be a click at the end of the sub-block or a complete silence after the first sub-block interrupt in clone cards as you say.
So yes, it appears that Single-Cycle DMA mode is one buggy mode even in official Creative cards.


my important / useful posts are here

Reply 15 of 111, by awgamer

User metadata
Rank Oldbie
Rank
Oldbie
firage wrote:
Great to have some good examples! […]
Show full quote

Great to have some good examples!

A CQM vs. OPL difference is apparent in Doom E1M1 - the CQM's emphasis really grates after just a little while. The instrumentation is pretty simple, though. There are probably some more glaring cases to compare out there, but I don't have a CQM card to make my own comparisons.

awgamer wrote:

It's supposed to be at the end right? It's very slight to me, and when you take the low quality recorded sound effects into account, it doesn't seem the biggest issue to me, but obviously it's annoying for those that pick it out.

No, the clicking is there throughout the Apogee fanfare in the Alien Rampage example. Like small pops or clipping artifacts or something.

I guess it would help to hear the same recording without the pops.

Reply 18 of 111, by dr_st

User metadata
Rank l33t
Rank
l33t

Great work summarizing it all, and a nice selection of samples. 😎

Here are my subjective impressions:

  • The Vibra distortion in Skyroads sounds quite bad, but the ringing is somewhat bearable. Which distortion type is more common?
  • The single-cycle DMA clicking is clearly audible, but it seems that the relative volume differs. In the Duke2 sample it's pretty annoying, for instance, but in the PoP footsteps it's quite benign. I imagine it would be less noticeable, and thus less disturbing when if is MIDI playing in the background.
  • The MPU-401 hanging note is atrocious and is easily the worst artifact of everything demonstrated here. I can understand why people would find the affected cards totally unsuitable for passing MIDI to an external device. How frequent would it be in practice? I.e., what percentage of MIDI tracks is affected?
  • The CQM vs OPL3 is a good choice of track, and the description of CQM as more metallic and buzzy is pretty accurately demonstrated in this tune, and it is potentially grating. Based on other comparisons I've heard, this is actually probably one of the more obvious ones. In many other tracks, the difference will not be as pronounced.
  • For instance, there is a nice comparison of OPL3 vs CQM here, with 4 different tracks and several cards (3 OPL, 2-3 CQM). The differences are there, and I clearly prefer the OPL variants, but for some tracks the differences are fairly mild, to the point that I'm not sure I would have noticed them if I weren't looking for the tell-tale signs, and not even sure I would have correctly identified it in a blind test.

https://cloakedthargoid.wordpress.com/ - Random content on hardware, software, games and toys

Reply 19 of 111, by James-F

User metadata
Rank Oldbie
Rank
Oldbie
dr_st wrote:

The Vibra distortion in Skyroads sounds quite bad, but the ringing is somewhat bearable. Which distortion type is more common?

The Ringing is more common, and indeed bearable.
Sometimes it is louder than I would want, even with music in the background.
Same with the clicking, sometimes it is just annoying and I can't not hear it even with background music.

. How frequent would it be in practice? I.e., what percentage of MIDI tracks is affected?

More often than anyone would want, a single note is enough to destroy the whole composition (IMO).

As for CQM, I don't like it one bit.
ESFM is a great OPL clone and indistinguishable without a direct comparison, can't say the same for CQM.


my important / useful posts are here