VOGONS


Sound Blaster 16 Bugs and Deficiencies Summary

Topic actions

Reply 20 of 111, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

Thanks for the summary, James F! Hopefully the information makes it's way to places like Nerdlypleasures and Vogonswiki (with proper credit) so their information is brought up to date.

All hail the Great Capacitor Brand Finder

Reply 21 of 111, by James-F

User metadata
Rank Oldbie
Rank
Oldbie

Only if the bugs have been proven by more people than the sole me.
I've already seen some skeptical comment that has been deleted...

I call for anyone with a SB16 AWE32 AWE64 that has any further knowledge, want to help me fix or make the list in the first post more accurate, please inform this thread.
If you have found anything in the first post misleading or wrong please inform this thread.
The point of this thread is to be a source of correct information that people can learn from and address other people too, NOT be a single mans bullshit post.

I know people like Great Hierophant, Ace, gerwin, and many more can be of great benefit to this thread, but they are becoming less frequent visitors these days.


my important / useful posts are here

Reply 22 of 111, by gerwin

User metadata
Rank l33t
Rank
l33t

* MPU-401 Hanging Note Bug.
May not be DSP related as I have a CT2290 DSP v4.13 that behaves much better hanging-note-wise, compared to my CT2800/CT2940/CT2950 cards with DSP v4.13.

* MPU-401 Stuttering with high sampling rates.
True, X-Wing may be another example I heard (not tested myself, but it figures).

* Single-Cycle DMA Clicking (Non Vibra).
Cannot say much about it. I mainly used CT2504 Vibra16S and later a CT2502 Vibra based card.

* Ringing/Hissing artifacts, Vibra only.
True for the above mentioned vibra cards. I found this myself years ago, without anyone pointing it out. By the way, There is a third Vibra chipset, the first release, which I do not own.

* Vibra CT2800 series Audio Distortion.
Please explain more.

* CQM vs OPL3.
Many opinions on Vogons already. Cannot say much about CQM, I gave up on it too quickly.

* Self Noise
I find all Sound Blasters kinda noisy, maybe even all ISA sound cards. I remember having difficulty with the mixer settings to find a balance between noise-floor and loudness of game sounds, even with the relatively silent CT2502 Vibra.
Another example: The Crystal CS4232 should be as silent as it gets. But the programming guide expects DOS software to mute the DACs after program exit. You can still hear the difference when the DACs are muted or when they are not.
What complicates the above: The waveblaster midi daughterboard interface is analog, not digital. In my experience this is a significant noise-floor already. (The later Crystal CS4236B and CS4237B chipsets introduced a digital synth connection, for their own synth chip. They could see the problem)
Either way these are anecdotes, numbers (dB) are better.

--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul

Reply 23 of 111, by keenmaster486

User metadata
Rank l33t
Rank
l33t

I worked up a spreadsheet based on this.

https://docs.google.com/spreadsheets/d/1xrjRR … dit?usp=sharing

I would note that there are far more SB16 models than are usually considered, probably because some are far more common than others. Here's the full list as far as I can tell:

CT2890      SoundBlaster 16
CT2860 Sound Blaster 16 (HP Menuet)
CT2840 SoundBlaster 16
CT2810 SoundBlaster 16
CT2806 SoundBlaster 16
CT2805 SoundBlaster 16
CT2804 SoundBlaster 16
CT2802 SoundBlaster 16
CT2801 SoundBlaster 16
CT2800 SoundBlaster 16
CT2772 SoundBlaster 16
CT2771 SoundBlaster 16
CT2760A Sound Blaster AWE32 CSP (Multi-CD)
CT2709 Sound Blaster 16 (Mitsumi)
CT2700 Sound Blaster 16 CSP (Mitsumi)
CT2600 Sound Blaster Pro 2 (Mitsumi)
CT280 SoundBlaster 16
CT2290 Sound Blaster 16 (Panasonic and IDE)
CT2239M Sound Blaster 16 (Mitsumi)
CT2230 Sound Blaster 16 (Mitsumi)
CT1799 Sound Blaster 16 Sony
CT1789 Sound Blaster 16 LMSI
CT1780 Sound Blaster 16 CSP LMSI
CT1779 Sound Blaster 16 SCSI-2
CT1759 Sound Blaster 16 MCD
CT1750 Sound Blaster 16 MCD CSP
CT1770 Sound Blaster 16 SCSI-2 CSP
CT2771 Sound Blaster 16 Value
CT1740 Sound Blaster 16 CSP
CT1748 Sound Blaster 16 CSP
CT1730 Sound Blaster 16
CT1299 Sound Blaster 16 Pro
CT1291 Sound Blaster 16 Value IDE
CT1290 Sound Blaster 16 Pro CSP
CT1262 Sound Blaster 16 VIBRA
CT1261 Sound Blaster 16 VIBRA
CT1260 Sound Blaster 16 VIBRA MCD
CT1239 Sound Blaster 16 MCD
CT1239c Sound Blaster 16 MCD
CT1239s Sound Blaster 16 MCD
CT1231 Sound Blaster 16
CT1230 Sound Blaster 16 MCD CSP
CT1230C Sound Blaster 16 MCD CSP
CT1230S Sound Blaster 16 MCD CSP

I put a few in that spreadsheet and tried to fill out what the different chips are, what bugs they have, etc. What do you guys think?

It's editable so you can continue to add cards, fix errors, add missing information, etc.

World's foremost 486 enjoyer.

Reply 24 of 111, by BloodyCactus

User metadata
Rank Oldbie
Rank
Oldbie
keenmaster486 wrote:

your spreadsheet needs it axis reversed. cards down, bugs across the top, for it to really be readable. otherwise, nice spreadsheet.

--/\-[ Stu : Bloody Cactus :: [ https://bloodycactus.com :: http://kråketær.com ]-/\--

Reply 25 of 111, by keenmaster486

User metadata
Rank l33t
Rank
l33t

Well, maybe you're right - I was thinking it would be easy to determine how many bugs a card has simply by reading the colors down each column.

World's foremost 486 enjoyer.

Reply 26 of 111, by James-F

User metadata
Rank Oldbie
Rank
Oldbie
gerwin wrote:

* MPU-401 Hanging Note Bug.
May not be DSP related as I have a CT2290 DSP v4.13 that behaves much better hanging-note-wise, compared to my CT2800/CT2940/CT2950 cards with DSP v4.13.

Is has the CT1747 chip, which is considered to be free of the hanging note bug.
But not from the stuttering MPU-401 bug.

gerwin wrote:

* Ringing/Hissing artifacts, Vibra only.
True for the above mentioned vibra cards. I found this myself years ago, without anyone pointing it out. By the way, There is a third Vibra chipset, the first release, which I do not own.

Ace clearly states that this is also the case with the first revision of Vibra CT2501 chip of the SB16 CT2260 card:

Ace wrote:

I've also noticed a rather annoying ringing noise on my SoundBlaster Vibra16 model CT2260. It seems to happen all the time when the sound card is first initialized, and sometimes also happens when digital sound stops playing or when the YMF262 outputs sound.

gerwin wrote:

* Vibra CT2800 series Audio Distortion.
Please explain more.

The CT2800 and CT2890 (might be all Vibra CT2504 family?) distorts when digital the audio reaches maximum input at the preamplifier stage inside the chip.
Some very loud DOS games distort the preamplifier without the user being able to do anything about it because the volume slider is AFTER the preamplifier so the audio will always distort.
Obviously there is not "preamplifier" in there but I used the terminology to explain the phenomenon.
This is the reason I consider it as a bug, because I could not do anything to resolve/fix this.

Example of clean vs distorted signal:
Re: My Sound Blaster 16 CT2230 Review

keenmaster486 wrote:

I worked up a spreadsheet based on this.

This is great.


my important / useful posts are here

Reply 27 of 111, by gerwin

User metadata
Rank l33t
Rank
l33t
James-F wrote:

* MPU-401 Hanging Note Bug.
Is has the CT1747 chip, which is considered to be free of the hanging note bug.
But not from the stuttering MPU-401 bug.

Yeah you already had that written, So my remark was unnecessary.
Though I still wonder sometimes, was there ever the observation that the same card CT number was found with different DSP versions? Like a CT1740 with DSP 4.06 and the bug? Is the DSP version the cause or just a correlation?

James-F wrote:

* Ringing/Hissing artifacts, Vibra only.
Ace clearly states that this is also the case with the first revision of Vibra CT2501 chip of the SB16 CT2260 card:

Ace wrote:

I've also noticed a rather annoying ringing noise on my SoundBlaster Vibra16 model CT2260. It seems to happen all the time when the sound card is first initialized, and sometimes also happens when digital sound stops playing or when the YMF262 outputs sound.

Good find, thanks!

James-F wrote:
* Vibra CT2800 series Audio Distortion. The CT2800 and CT2890 (might be all Vibra CT2504 family?) distorts when digital the audi […]
Show full quote

* Vibra CT2800 series Audio Distortion.
The CT2800 and CT2890 (might be all Vibra CT2504 family?) distorts when digital the audio reaches maximum input at the preamplifier stage inside the chip.
Some very loud DOS games distort the preamplifier without the user being able to do anything about it because the volume slider is AFTER the preamplifier so the audio will always distort.
Obviously there is not "preamplifier" in there but I used the terminology to explain the phenomenon.
This is the reason I consider it as a bug, because I could not do anything to resolve/fix this.
Example of clean vs distorted signal:
Re: My Sound Blaster 16 CT2230 Review

Okay, thanks for explaining. I don't recall this problem, but maybe, back then, I was too focused on the ringing problem. Here are these posts:
SB Vibra 16S / Vibra Pro / AWE64 output quality
What I also found there is that the CT2860 card is like a terrible Vibra16S variant because all output goes through an additional TV-grade amplifier. I also found it to have more ringing sound problems then my CT2800 card, even after removing the undesirable amplifier. Which may just be because of random manufacturing variations.

In general I cannot really motivate myself to go test these CT2800/CT2860 cards again, because I already replaced them anyways. My CT2940/CT2950 with OPL3 are unused too. Someone else can go defend them. The only Creative card I still have an ISA slot for is the AWE CT1920 goldfinch midi card.

--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul

Reply 28 of 111, by badmojo

User metadata
Rank l33t
Rank
l33t

I’ve really enjoyed your sound card examinations James-F and am not here to refute any of your findings, but this thread prompted me to revisit my choice of sound hardware in my P166 (a CT2290) and I’ve come to a different conclusion regarding the SB16.

As you found with the CT2230 (basically the same card) the CT2290 is quiet and has a real OPL3. Any number of clones meet those requirements too, but after cycling through my box of tricks (ES1688, ES1868, ES688, PAS16, YMF718, Audio Excel 3D), the CT2290 has once again won out for these reasons:

  • It sounds great. No modding required, it sounds like a Sound Blaster. The "it’s not SB Pro compatible" issue is often raised, but it doesn’t need to be SB Pro compatible. It’s SB16 compatible!
  • Great software for DOS and Windows 95. This is a big deal for me – some clones have good software, but others are awful in this regard. Creative did software well IMO – user friendly DOS mixer, no memory hogging TSR, and a comprehensive Windows suite. No wondering "what does this slider do?". No having your AUTOEXEC file updated on each boot. No wasting your time hunting through 'driverguide.com' looking for the right version.
  • Mixer levels are correct and stay put b/w games. Mixer levels on clones are often wonky I find, and having to tweak the mixer to get the voice and FM to sensible levels is doable, but once you add in wavetable / line-in / CD-in to the mix then it turns in to an exercise in frustration – particularly if the mixer software isn't user friendly. And then a game like DOOM comes along and messes with your settings and has you scrambling for the volume nob because you're waking up the kids – this doesn't happen with the SB16.
  • The SB16 is a full-height card with mounting holes in the right spots for a full sized DB. Running cables to a DB from a half-height card is fine, but not ideal.
  • No nasty pop when you power on the system. Not all clones do this, but a lot do. The CT2290 does make a noise, but it’s not offensive.

All of that makes the better SB16s (CT2230, CT2290, etc) pretty tough cards to beat for a straight up voice + FM solution for later DOS games, rivalled only by the SB Pro and PAS16 in my experience, but then of course there's the GM issue. If you choose the right SB16, then the MPU-401 is kinda-sorta bug free(ish). It works fine for most games, and to deal with the stutter in Duke3D, Tie Fighter (any others…?), it’s the Creative Goldfinch to the rescue! (Or any second card with a working MPU-401 I guess, but the Goldfinch is cooler).

By sneaking a Goldfinch into the slot next door, you get AWE for games that support it (Tie Fighter happens to sound fantastic on the AWE), GM with the default samples (admittedly quite crappy), and the ability to load sound fonts in Win 9x. If you drop 32MB of RAM onto your Goldfinch then you can load some pretty impressive sound fonts, and alongside a SB16 (and only a SB16 – no clone allows this), you can tell the Goldfinch to take over the MPU-401 duties in Windows, which of course means that you have access to your loaded sound font from within games. So that’s the Duke3D issue resolved too!

Anyway, I've banged on about this setup before but thought it was worth mentioning again in the context of this thread (I can refer to this next time I start questioning my loyalties). With a good SB16 + Goldfinch + optional DB, you get:

  • A great sounding, easy to use, and compatible sound card
  • A genuine OPL3
  • The option for wavetable via a DB the SB16s header
  • AWE32 capability
  • Sound fonts!

DSCN2820_zpshude9ylh.jpg

Life? Don't talk to me about life.

Reply 29 of 111, by SquallStrife

User metadata
Rank l33t
Rank
l33t

Does the MIDI "stuttering" affect using SoundFonts under Win9x on AWE cards?

As I understand it you're not using the card's MIDI UART, but rather a Win9x VXD that talks to the EMU chip. (Happy to be corrected on this)

If that's the case then the AWE cards present a fine solution for late DOS gaming on a Win9x rig, where having real OPL isn't very important. (And you can disengage your "But what if I want to .... one day?" gland 😜)

VogonsDrivers.com | Link | News Thread

Reply 30 of 111, by badmojo

User metadata
Rank l33t
Rank
l33t
SquallStrife wrote:

Does the MIDI "stuttering" affect using SoundFonts under Win9x on AWE cards?

Nar you're right, there's no stuttering if the Goldfinch takes control of the wheel. But I'll just say again that the Goldfinch can only take control of a real SB16's MPU-401 - I've tried it with ~6 clones (including an AudioExcel 3D which is a for reals SB16 clone) and it doesn't work.

Life? Don't talk to me about life.

Reply 31 of 111, by shamino

User metadata
Rank l33t
Rank
l33t

The "single cycle DMA mode" bug confuses me a bit.
Unless I'm misinterpreting [edit: Yes, I was misinterpreting, so ignore the rest of this post!]
, I think this is referring to what I always thought was called "single buffered DMA" as opposed to "double buffered DMA". The difference there being that with single buffering, the application will overwrite the contents of the whole buffer while it's being played so you'll always get an audible pop. I've never heard of this being a hardware glitch but rather an inevitable result of how single buffering works.
With double buffering you can overwrite one part of the buffer while another part is being played. You never overwrite the part that's playing, so you don't get a pop. IIRC support for this mode was added in later models of the 8-bit Sound Blaster, maybe it was the Sound Blaster 2.0 or something like that. It is possible to detect in software whether this is supported, so a well written application should use double buffering when possible.

A long long time ago I was learning to write some DOS Sound Blaster code and used double buffered DMA to play an audio file of any length. I tested it on an SBPro CT1330A (the older one) and an AWE32 (either a CT3990 or maybe a CT2760, I don't remember which I had at the time).
As far as I noticed there wasn't any popping/clicking issue, but with single buffering it was expected to happen because single buffering is dumb.

If a piece of software is overwriting new sound data at the same point in the buffer that's currently playing, then I don't really see how the pops and clicks could be avoided at the hardware level or blamed on it.

Last edited by shamino on 2016-10-07, 14:32. Edited 1 time in total.

Reply 33 of 111, by James-F

User metadata
Rank Oldbie
Rank
Oldbie

Great Hierophant from Nerdly Pleasures noticed the Sincle-Cycle DMA bug a long time ago, this should me common knowledge by now.
Just search for "pops" in these links:

http://nerdlypleasures.blogspot.com/2010/06/i … d-blasters.html
http://nerdlypleasures.blogspot.com/2012/07/s … ibulations.html
http://nerdlypleasures.blogspot.com/2012/09/4 … ng-machine.html
http://nerdlypleasures.blogspot.com/2015/05/1 … t-playback.html


my important / useful posts are here

Reply 34 of 111, by firage

User metadata
Rank Oldbie
Rank
Oldbie

Information about the bug has been super vague in the past, to be fair. Even in one of his links there, GH associates the popping and clicking with the first generation of SB16's only, promising that they're "fully gone" with the CT2230's CT1703-equipped models.

My big-red-switch 486

Reply 35 of 111, by PhilsComputerLab

User metadata
Rank l33t++
Rank
l33t++
firage wrote:

Information about the bug has been super vague in the past, to be fair. Even in one of his links there, GH associates the popping and clicking with the first generation of SB16's only, promising that they're "fully gone" with the CT2230's CT1703-equipped models.

I'm with you on this one. James, I also wasn't aware of this issue 😊

And I admit I never noticed it either. Not back in the day, not in modern day. So this is really interesting, but be patient with everyone "catching up" with this new world order 🤣

YouTube, Facebook, Website

Reply 36 of 111, by James-F

User metadata
Rank Oldbie
Rank
Oldbie
firage wrote:

Information about the bug has been super vague in the past, to be fair. Even in one of his links there, GH associates the popping and clicking with the first generation of SB16's only, promising that they're "fully gone" with the CT2230's CT1703-equipped models.

The links are ordered by date.
In 2015 (last link) GH says about CT2230:

Great Hierophant wrote:

Surprisingly, despite the later DSP and CODEC, this card is even noisier than its predecessor the CT-1750. It also has the same amount of pops and clicks. Its actually the noisiest of the bunch.

His CT2230 has the CT1703-T DAC, hence the complaint about noise.
Latter CT1703-TBS and CT1703-A which also come in the CT2230 are as silent as the AWE64.

Right after the CT2230 he has the CT2940 Vibra CT2502 which does not have the "pops" but has the "rings" which he hasn't noticed.


my important / useful posts are here

Reply 37 of 111, by badmojo

User metadata
Rank l33t
Rank
l33t

Yep I'm not saying the single cycle thing isn't a thing, but Mortal Kombat was mentioned as a game impacted by it and I haven't noticed any pops or clicks with any of the SB16's I've used, and I play a lot of MK 😈

Life? Don't talk to me about life.

Reply 38 of 111, by shamino

User metadata
Rank l33t
Rank
l33t

What I said earlier about single vs double buffering wasn't relevant, so strike that.

My confusion was/is with the exact meaning of the term "single cycle DMA" and where exactly the bug is, beyond the general mention of pops and clicks occurring. The definition for the term earlier in the thread talks about interrupts and double buffering across page boundaries and I think that threw me off. After some research I think those details aren't really relevant to what it means.

I looked up some old code and some docs from the late 90s. I also tried doing a modern web search but it turned up 2 apparently conflicting meanings for the term. However, the usage of the term in the context of Sound Blaster cards seems to be consistent, so I think I get what "single cycle DMA" is now.

Single cycle DMA means that the card will play the designated sound buffer once and then stop. It will not play the buffer again until a new DMA transfer is set up. As a result, there is a gap between the time that the buffer is finished playing and when the application sets up another DMA transfer. This gap causes potential pops or clicks.
The alternative to this is Auto Initialized DMA mode, in which the buffer is automatically looped over and over without a gap. As long as the application keeps updating the buffer in an appropriate manner then it will play seamlessly. This is why you hear looping audio when many games crash.

Single cycle DMA is inherently flawed, but I'm surprised to hear that it exhibits worse audio artifacts on some cards than others. Seems like this issue would be affected by the programming of the particular game and possibly the system running it, but if the issue is more pronounced on different cards in exactly the same setup then that is indeed strange.
Has it been determined that the affected cards make a pop at the start or end of a single isolated transfer, when another transfer is not being started (and thus the gap between separate DMA operations is not a factor)?

I was wondering if there's a simple way to test for the single cycle DMA bug, but I just found this post in another thread:

James-F wrote:
To test the Single-Cycle DMA mode bug, start Alien Carnage (freeware) and listen to the music. In Wolfenstein 3D listen to openi […]
Show full quote

To test the Single-Cycle DMA mode bug, start Alien Carnage (freeware) and listen to the music.
In Wolfenstein 3D listen to opening doors sound, you might want to disable the music for better audibility.
In Prince of Persia there is a CLICK after each footstep sound.
Supaplex after a Ball (zonk) falls there is a click.

I don't have any of those games but I'll try to remember to at least get Alien Carnage and experiment with this on some cards when I get my DOS machine back up and running. Thanks.

Reply 39 of 111, by James-F

User metadata
Rank Oldbie
Rank
Oldbie

You don't have to assume the meaning of the term "single-cycle DMA" it is clearly documented in the official Sound Blaster Programming Guide.

Attachments


my important / useful posts are here