Are there Bass/Treble settings on your SB16 that maybe are turned all the way down?

Maxxarcade wrote:

For some reason, Jazz Jackrabbit and Epic Pinball refuse to play at full quality on either card. It sounds like the sample rate is around 22Khz, despite the settings being on Ultra quality and SB16 in the setup. The problem persists with the setting on SB Pro too.

This may be normal. I vaguely remember that I had a similar issue with Jazz Jackrabbit on AWE64.

Yes, I've had this issue with Jazz Jackrabbit before too.

I was never able to fix it. And yes, I haven't had the issue on clones, only real SB16 cards and AWE64.

Well darn... I was hoping there was a workaround, but it seems I'm stuck trying to find a decent clone card. I wish more of them had jumpers instead of flaky software initialization.

Interesting, I never noticed that issue back in the day when I played it (the shareware version).
Guess I was lucky beeing a PAS16 owner! 😁
Anyway, is there any recording of that issue available ?
I'm curious about that issue and wonder if it also affects the CT17xx models..

I might do a recording in the near future if I get time. But it's basically dull and muffled on the SB16, and CD quality on the clone card.

Epic titles were aimed at GUS soundcards more then soundblaster, which may be why it sounds muffled.
I've never really noticed it but then I had a SB16 back when I played these games new and on terrible speakers so anything sounds better these days

Yeah, the underlying issue is that it restricts the sample rate to the lowest setting on a real SB16 no matter what you select, but it has no issue with clones.

Does this also happen in DOSBox or PCem ? Sorry, I know that question might be silly, but I have no comparison. So I have to ask. 😅

Jo22 wrote:

Does this also happen in DOSBox or PCem ? Sorry, I know that question might be silly, but I have no comparison. So I have to ask. 😅

You could compare DOSBox SoundBlaster vs DOSBox GUS.
Just remember to set both `rate=` and `gusrate=` to 44100.

Not a silly question, but no, the behavior with DOSBox is the same as with clones; i.e. it works perfectly.

I took recordings of One Must Fall and Epic PInball and Pinball Fantasies with a GUS, an SB Pro and a PAS16. With the Epic games, the SB Pro sounds terrible compared to the PAS16 and the GUS : http://nerdlypleasures.blogspot.com/2016/10/r … ultrasound.html

Someone commented on an IRC channel that the reason for this was because Epic MegaGames hated Creative Labs for dominating the market and shutting out their beloved GUS, so they took revenge by making their games sound terrible on true CL hardware while not sounding terrible on the clones. If this was the case there should be code somewhere in these programs that detects real CL hardware and could conceivably be patched out. I find the idea a bit farfetched, but there is some support in the empirical evidence that, for whatever reason, their drivers don't like CL products. Honestly, I did not record the SB Pro through a tin can on a string, but it sure sounds that way!

Thank you very much Great Hierophant! Your article is excellent, as usual! 😁
I'd just want to add that the PAS16 recording is perhaps a bit overdriven.
This happens easily, because the mixer has an extreme loud volume default setting.
Changing it to something lower may give you clearer audio.
Sample: PAS SET VOLUME TO 70 IN 2 SECONDS

@Maxxarcade

Are you the same Maxxarcade from Youtube?

gdjacobs wrote:

@Maxxarcade

Are you the same Maxxarcade from Youtube?

Yep!

Cool! I love your audio system repair videos, both pro and consumer.

Also, welcome to the forum. We have a few threads on stuff like recapping, bad vs good PSU design, etc. I think based on your vids and the skill level you demonstrate, your experience and opinions would be especially appreciated!

Hi,

Maxxarcade wrote:

For some reason, Jazz Jackrabbit and Epic Pinball refuse to play at full quality on either card. It sounds like the sample rate is around 22Khz, despite the settings being on Ultra quality and SB16 in the setup. The problem persists with the setting on SB Pro too.

As I've noticed that on a SB16 too, I decided to investigate. The Epic/Jensen Soundblaster driver checks the DSP version and sets it to 2.0 (Soundblaster 2.0) if it is >5.00. It seems there is a bug in the code requesting the version from the card (I guess timing issues) and the driver always assumes it is >5.0 and thus limits it to 2.0.

I've just modified the fallback to set 4.00 as DSP version and Epic Pinball started to sound great when selecting SB16 in the setup.

This is the necessary patch:

1 0000240F  7205              jc 0x2416
2 00002411  3D0005            cmp ax,0x500
3 00002414  7603              jna 0x2419
4-00002416  B80002            mov ax,0x200
5+00002416  B80004            mov ax,0x400
6 00002419  A33B22            mov [0x223b],ax
7 0000241C  C3                ret
8 0000241D  8A0E4222          mov cl,[0x2242]

Just change the 0x02 at 0x2418 to 0x04 in MDRV004R.MUS.

This also works for Jazz Jackrabbit, where the offset is 0x24CA in MDRV004D.MUS. I'm using the GoG version of both games.

Cheers,
Michael

@stargo: Thanks that is really helpful. I patched my files.

Stargo, thanks for digging into this and pointing this out.
The same sound drivers are also used for one must fall 2097.
I can confirm that the same issue exists on my Intel Morrison motherboard with an AWE32 or a SB16 (can't remember the exact CT models out of my head)
I always wondered why the sound was so "dull" compared to PAS16 sound card in the same machine. After patching the driver the sound quality was comparable.
Other games might also use the same driver. Don't know if anyone has an exact list.
For now we have OMF2097/Epic Pinball/Jazz Jackrabbit.
I also tried the emulating this in PCEM with some other motherboard config and wasn't able to replicate the same bug in that environment.
So it has to do something with the hardware.
i did a quick disassembly of the related routines and it looks like the value is read out by using some IO DMA access to get that number.
I'm not familiar with how this precisely works so maybe someone else can take a look into this
Here's a small listing of the main routine and 2 subroutines that do the DMA work for the OMF driver

1seg000:2400                 mov     al, 0E1h ; 'ß'
2seg000:2402                 call    sub_2243
3seg000:2405                 call    sub_2257
4seg000:2408                 jb      short loc_2416
5seg000:240A                 mov     ah, al
6seg000:240C                 call    sub_2257
7seg000:240F                 jb      short loc_2416
8seg000:2411                 cmp     ax, 500h
9seg000:2414                 jbe     short 
10seg000:2416 loc_2416:                               ; CODE XREF: seg000:2408j
11seg000:2416                                         ; seg000:240Fj
12seg000:2416                 mov     ax, 200h
13seg000:2419
14seg000:2419 loc_2419:                               ; CODE XREF: seg000:2414j
15seg000:2419                 mov     ds:word_223B, ax
16seg000:241C                 retn
17
18
19seg000:2243 sub_2243        proc near               ; CODE XREF: seg000:2402p
20seg000:2243                 push    ax
21seg000:2244                 push    dx
22seg000:2245                 mov     ah, al
23seg000:2247                 mov     edx, ds:dword_24B
24seg000:224C
25seg000:224C loc_224C:                               ; CODE XREF: sub_2243+Cj
26seg000:224C                 in      al, dx
27seg000:224D                 or      al, al
28seg000:224F                 js      short loc_224C
29seg000:2251                 mov     al, ah
30seg000:2253                 out     dx, al
31seg000:2254                 pop     dx
32seg000:2255                 pop     ax
33seg000:2256                 retn
34seg000:2256 sub_2243        endp
35
36seg000:2257 sub_2257        proc near               ; CODE XREF: seg000:2405p
37seg000:2257                                         ; seg000:240Cp
38seg000:2257                 push    dx
39seg000:2258                 push    cx
40seg000:2259                 mov     edx, ds:dword_24B
41seg000:225E                 add     dl, 2
42seg000:2261                 mov     cx, 0FFFFh
43seg000:2264
44seg000:2264 loc_2264:                               ; CODE XREF: sub_2257+12j
45seg000:2264                 in      al, dx          ; DMA controller, 8237A-5.
46seg000:2264                                         ; channel 1 current address
47seg000:2265                 or      al, al
48seg000:2267                 js      short loc_226E
49seg000:2269                 loop    loc_2264
50seg000:226B                 stc
51seg000:226C                 jmp     short loc_2275
52seg000:226E ; ---------------------------------------------------------------------------
53seg000:226E
54seg000:226E loc_226E:                               ; CODE XREF: sub_2257+10j
55seg000:226E                 sub     dl, 4
56seg000:2271                 in      al, dx
57seg000:2272                 clc
58seg000:2273                 jmp     short $+2
59seg000:2275 ; ---------------------------------------------------------------------------
60seg000:2275

61seg000:2275 loc_2275:                               ; CODE XREF: sub_2257+15j
62seg000:2275                                         ; sub_2257+1Cj
63seg000:2275                 pop     cx
64seg000:2276                 pop     dx
65seg000:2277                 retn
66seg000:2277 sub_2257        endp