VOGONS


First post, by James-F

User metadata
Rank Oldbie
Rank
Oldbie

So, if anyone wonders what the sound difference is, I made a comparison that combines OPL and PCM audio so one can clearly hear the balance and filtering between these cards.

SB1.5 CT1320C - Prince of Persia 2
SB2.0 CT1350B - Prince of Persia 2
SBPro2 CT1600 - Prince of Persia 2
SB16 CT2230 - Prince of Persia 2

SB1.5 CT1320C - Gods
SB2.0 CT1350B - Gods
SBPro2 CT1600 - Gods
SB16 CT2230 - Gods

Edit: All tracks are downloadable so you can properly compare them in your favorite player.
I recommend listening with a good pair of headphones.

I should add that SB1.5 and SB2.0 are VERY sensitive to CPU speed and many games break the OPL2/Adlib music in a Pentium.
That includes popular games like Doom, Prince of Persia 1/2, Mortal Kombat 2, DOTT, etc..

I had to underclock the Pentium 200 (non-MMX) CPU to 90MHz (60MHz FSB), which at that point the YM3812 (OPL2) started to behave properly in games like PoP1/2 and DOTT.
BUT, at 60MHz FSB the SB1.5 CT1320C goes into buffer overflow at some point and stop playing auto-initialized DMA sounds altogether in games like Doom, this does not happen at 66MHz FSB.
50MHz FSB breaks SB1.5 and SB2.0 DSP completely and no sound can be heard after a second or so...
So not only speed sensitive games have problems detecting the SB cards at "higher" clocks, but the DSP of the SB1.5 and SB2.0 in itself is sensitive to clock, which is a completely different issue than the YM3812 (OPL2) sensitivity to clock.

So in general;
SB1.0 - SB2.0 should live in a 486, nothing above that.
SBPro2 and SB16 are much better paired with a Pentium.

EDIT: important links to shorten my signature.
SB1.5, SB2.0, SBPro2, SB16 - Audio Comparison <- You're Here
Sound Blaster 16 Bugs and Deficiencies Summary
Yamaha YMF71x SB Pro Mixer Bug Issue Fix
FM and PCM Volume Balance of SB Cards
Sound Card Lowpass Filtering Measurement
SBPro1 CT1330A, lets solve the Reversed Stereo myth!

Last edited by James-F on 2020-01-09, 06:31. Edited 11 times in total.


my important / useful posts are here

Reply 1 of 30, by James-F

User metadata
Rank Oldbie
Rank
Oldbie

My personal subjective opinion on the sound of these cards:
The SB2.0 PCM audio is the less filtered of the bunch and may even sound "dirty" in comparison to others.
The SB1.5 is slightly more filtered than the SB2.0 but less than SBPro2, it sounds just right.
The SBPro2 is the most filtered and has the thickest ballsiest sound which we all familiar with, perfect for games with explosions and firing.
The SB16 has automatic brickwall filtering which moves according to to sampling rate like in a modern audio card, sounds very natural without any aliasing at any sampling rate because this kind of filtering.

My personal choice: SBPro2 for being Stereo, OPL3 and a bug-less workhorse in a "fast" Socket 7 machine.


my important / useful posts are here

Reply 2 of 30, by Scali

User metadata
Rank l33t
Rank
l33t
James-F wrote:

I should add that SB1.5 and SB2.0 are VERY sensitive to CPU speed and many games break the OPL2/Adlib music in a Pentium.
That includes popular games like Doom, Prince of Persia 1/2, Mortal Kombat 2, DOTT, etc..

That is probably because of the OPL2, which requires delays between commands. Most games probably didn't use proper delay routines, so they are speed-sensitive (nobody could imagine a CPU of the speed of a Pentium back when those games were developed).
The OPL3 doesn't require these delays, so it's not as speed-sensitive.

James-F wrote:

So not only speed sensitive games have problems detecting the SB cards at "higher" clocks, but the DSP of the SB1.5 and SB2.0 in itself is sensitive to clock, which is a completely different issue than the YM3812 (OPL2) sensitivity to clock.

Yes, the DSP is quite slow. I once fixed the detection routines in the Crystal Dream demo, which failed to detect a real SB/SB Pro on a fast 486 or Pentium. The code worked fine when you disabled the turbo, or used a clone card with faster DSP.
So it's probably similar to what happens with the OPL2: games send commands too quickly for the DSP to respond.
It's interesting that it is more troublesome at 50 and 60 MHz than at 66 MHz FSB. Might have something to do with how the buses operate asynchronously.

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/

Reply 4 of 30, by jesolo

User metadata
Rank l33t
Rank
l33t
James-F wrote:

So in general;
SB1.0 - SB2.0 should live in a 486, nothing above that.
SBPro2 and SB16 are much better paired with a Pentium.

Second Keropi's comment.

I tend to agree with your above statement.
I would even go so far (which is just my own opinion) as to say that a Sound Blaster 1.0 & 2.0 should be quite happy with anything up to a 386SX or slow 386DX, with a Sound Blaster Pro 2 better suited for a 386DX-33 up to a 486DX2-66.
By the time that your fast 486 & Pentium 1 CPU's became popular choices, the Sound Blaster 16 was already fully supported by most games and, under Windows with either Windows based games or other multimedia applications, will then also provide you with full 16-bit sound playback.

Reply 5 of 30, by PhilsComputerLab

User metadata
Rank l33t++
Rank
l33t++

We need to get you a proper 386 or 486 for your testings 😁

I love the Sound Blaster 1.5. It was my very first sound card and without technical explanations, it does a better job at "bringing back good emotions" than others. The CMS upgrade is also a nice bonus.

Sound Blaster Pro 2 is also a great card.

Also interesting that choosing the right sound card happens organically on Vogons. Meaning I don't think I have seen many use a Pentium with a Sound Blaster 1.5 or 2.0.

YouTube, Facebook, Website

Reply 6 of 30, by meljor

User metadata
Rank Oldbie
Rank
Oldbie

I have a ''quickshot sound machine'' card with a CT1336 chip on it, which card is that? Is it sb 2.0 or sb pro 2?

asus tx97-e, 233mmx, voodoo1, s3 virge ,sb16
asus p5a, k6-3+ @ 550mhz, voodoo2 12mb sli, gf2 gts, awe32
asus p3b-f, p3-700, voodoo3 3500TV agp, awe64
asus tusl2-c, p3-S 1,4ghz, voodoo5 5500, live!
asus a7n8x DL, barton cpu, 6800ultra, Voodoo3 pci, audigy1

Reply 8 of 30, by meljor

User metadata
Rank Oldbie
Rank
Oldbie

Thank you!

asus tx97-e, 233mmx, voodoo1, s3 virge ,sb16
asus p5a, k6-3+ @ 550mhz, voodoo2 12mb sli, gf2 gts, awe32
asus p3b-f, p3-700, voodoo3 3500TV agp, awe64
asus tusl2-c, p3-S 1,4ghz, voodoo5 5500, live!
asus a7n8x DL, barton cpu, 6800ultra, Voodoo3 pci, audigy1

Reply 9 of 30, by elianda

User metadata
Rank l33t
Rank
l33t

Are you sure you get the speed issues because of the Pentium and not due to a wrong configuration?

The recommended defaults are:
PCI Delayed Transaction to enabled
Passive Release to disabled
Also increasing 8-bit I/O Recovery Time and 16-bit I/O Recovery Time may help.

Retronn.de - Vintage Hardware Gallery, Drivers, Guides, Videos. Now with file search
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool

Reply 10 of 30, by James-F

User metadata
Rank Oldbie
Rank
Oldbie

Thank you all for the kind words.

elianda wrote:

Are you sure you get the speed issues because of the Pentium and not due to a wrong configuration?

I think so.
The SB1.5 (DSP v2.0 on mine) is even more sensitive than the SB2.0 (DSP v2.02) to the FSB frequency.
The SB2.0 can tolerate longer a 60Mhz FSB, while the SB1.5 breaks rather quickly.
It might be how the system behaves as a whole, so on a 50Mhz 486 the older SB cards may behave with no issues.

Last edited by James-F on 2017-05-09, 07:51. Edited 1 time in total.


my important / useful posts are here

Reply 11 of 30, by PhilsComputerLab

User metadata
Rank l33t++
Rank
l33t++

In Gods the 2.0 sounds quite a bit brighter than the 1.5. The SB Pro 2.0 is like you sounds heavily filtered, but sounds very clean and smooth. The 16 is a bit hard to place for me, it sounds brighter than the 1.5 though?

This is over headphones. Over speaker it is much harder for me to pick out differences. Here the 2.0 sounds quite bad to be honest. The SB Pro 2 sounds lovely, very clean and clear, but when you switch over to the 16 it brightens and opens up just a bit to sound even nicer. The 1.5 sounds very similar, but the 16 just a tiny bit livelier if that makes sense?

1. SB 16
2. 1.5
3. Pro 2
4. 2.0

Is my preference.

YouTube, Facebook, Website

Reply 12 of 30, by James-F

User metadata
Rank Oldbie
Rank
Oldbie

Yes the SB16 indeed sounds brighter, yet more filtered (smoother) at the same time.
The SB16 (also AWE32 and AWE64) filters everything above 5.5kHz (11kHz sampling rate), while the other cards have passive roll-off filter which is fixed.
The SBPro2 is the most filtered while the SB2.0 is the least, the SB1.5 somewhere in between (not shown).
SB16 and SBPro2 have almost 15db difference at 5kHz, while having the same frequency response below 1kHz on all of them.

The sound card indeed makes difference in how your DOS games will sound, something that DOSBox is not capable yet to emulate authentically.

Spectrum.png
Filename
Spectrum.png
File size
91.54 KiB
Views
10036 views
File license
Fair use/fair dealing exception

PS.
Please don't change Capacitors expecting "better low-end", in reality it does nothing.
All the cards are at least 20 years old, yet measure absolutely the same below 1kHz with stock capacitors.
😎


my important / useful posts are here

Reply 13 of 30, by PhilsComputerLab

User metadata
Rank l33t++
Rank
l33t++

Fantastic work! Especially the graphs show what's really going on.

Looking at the graph, the SB16 let's a bit more of the "good highs" through, and then filters out the "bad highs" very thoroughly.

YouTube, Facebook, Website

Reply 14 of 30, by James-F

User metadata
Rank Oldbie
Rank
Oldbie

One last comparison with PCM audio only where it's much clearer to hear artifacts and noise.
I also thrown in the YMF719 for comparison.

SB1.5 CT1320C - Prince 2 (Speech Only)
SB2.0 CT1350B - Prince 2 (Speech Only)
SBPro2 CT1600 - Prince 2 (Speech Only)
SB16 CT2230 - Prince 2 (Speech Only)

YMF719 Filtered (modified) - Prince 2 (Speech Only)
YMF719 Non-Filtered (stock) - Prince 2 (Speech Only)

PhilsComputerLab wrote:

Fantastic work! Especially the graphs show what's really going on.
Looking at the graph, the SB16 let's a bit more of the "good highs" through, and then filters out the "bad highs" very thoroughly.

Thanks!
Yes indeed, a brickwall filter is actually still used in ALL modern audio equipment from cheapest MP3/CD Player to most expensive studio equipment.
In 44.1kHz (CD quality) for example, the brickwall filters out anything above 20kHz at the edge of our human range.
This to avoid DAC (digital-to-analog conversion) aliasing.

Last edited by James-F on 2016-12-30, 15:59. Edited 2 times in total.


my important / useful posts are here

Reply 15 of 30, by Scali

User metadata
Rank l33t
Rank
l33t
PhilsComputerLab wrote:

Looking at the graph, the SB16 let's a bit more of the "good highs" through, and then filters out the "bad highs" very thoroughly.

Yea, and that's where the 'problem' lies. It sounds different from the earlier SBs.
If you're used to how a game sounds on an early SB, then the extra high frequencies will sound 'wrong'. In fact, when creating the music and sound effects for the game, they probably tuned it for how an SB sounds, which may have made them boost the high frequencies somewhat.
On the SB16 it will sound 'overly bright'.

What I find interesting to see is that the SB16 has no signal at all above 15 Khz... The SB2.0 and Pro both seem to 'bleed' all sorts of weird aliasing in these high frequencies. Their designs are pretty cheap and flawed.

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/

Reply 16 of 30, by James-F

User metadata
Rank Oldbie
Rank
Oldbie
Scali wrote:
Yea, and that's where the 'problem' lies. It sounds different from the earlier SBs. If you're used to how a game sounds on an ea […]
Show full quote
PhilsComputerLab wrote:

Looking at the graph, the SB16 let's a bit more of the "good highs" through, and then filters out the "bad highs" very thoroughly.

Yea, and that's where the 'problem' lies. It sounds different from the earlier SBs.
If you're used to how a game sounds on an early SB, then the extra high frequencies will sound 'wrong'. In fact, when creating the music and sound effects for the game, they probably tuned it for how an SB sounds, which may have made them boost the high frequencies somewhat.
On the SB16 it will sound 'overly bright'.

Yeah, I concur.
That's why I run both the SB16 and SBPro2 side by side on different addresses for maximum flexibility.
For newer games using auto-init DMA with versatile setup utilities I direct them to the SB16 (240 5 0), while older games by default go for the SBPro (220 7 1).

Scali wrote:

What I find interesting to see is that the SB16 has no signal at all above 15 Khz... The SB2.0 and Pro both seem to 'bleed' all sorts of weird aliasing in these high frequencies. Their designs are pretty cheap and flawed.

The SB16 in this example is at 11kHz sampling rate, but at 44.1kHz all the frequencies are there 20-20kHz and it sounds (and measures) quite good in Win98SE playing MP3.
Part of the "secret sauce" of the early SB sound (with passive filtering) is the in-band aliasing when playing lower sampling rate audio like most DOS games.
Definitely cheap and flawed by todays standards but for a DOS sound card lover like myself, this is an authentic treasure.


my important / useful posts are here

Reply 17 of 30, by Cloudschatze

User metadata
Rank Oldbie
Rank
Oldbie
James-F wrote:

That's why I run both the SB16 and SBPro2 side by side on different addresses for maximum flexibility.

Wait, hadn't you declared the SB16 to be a "lousy," "buggy mass," a "really bad egg," and "the last card you want for DOS gaming" just a few months ago? 😉

Reply 18 of 30, by James-F

User metadata
Rank Oldbie
Rank
Oldbie

I did...except when it's not. 😎

Newer games that use auto-initialized DMA (not single-cycle DMA) don't have the clicks, that's where I (sometimes) use the SB16, but the majority of games are run on the SBPro.
Moreover, I mix the SBPro2 PCM and SB16 Wavetable or MIDI in games like Duke3D so there are no MIDI related bugs like when running both PCM and MIDI on the SB16.
The SBPro Out is looped back into the Line-In of the SB16 and carefully mixed.
Besides, I prefer how the SBPro sounds even in newer games.

The SB16 is still not a good card for DOS gaming when used solely, but it is a good companion to the SBPro to complete the SBPro where it lacks, mainly as a Wavetable host and MPU-401.
I still believe a YMF71x based card is the way to go as a single DOS card.


my important / useful posts are here

Reply 19 of 30, by Cloudschatze

User metadata
Rank Oldbie
Rank
Oldbie
James-F wrote:

The SB16 is still not a good card for DOS gaming when used solely, but it is a good companion to the SBPro to complete the SBPro where it lacks, mainly as a Wavetable host and MPU-401.

I'm actually very fond of this idea. Not so much the pairing of the SBPro2 with an SB16 proper, but rather, with an AWE32 or AWE64, where a wider breadth of compatibility is achieved by the duo (concerning the EMU8000).