LABS wrote:

[...]

Nevertheless, I just disassembled and decoded to a human-readable form the original firmware of Sound Blaster 2.0 (DSP 2.01). Many unknown things revealed and there is a room for perfection for BB's future firmware releases.

this is truly amazing news, kudos! 😲

LABS wrote:

Noisy playback in Windows (I assume you are running W95/98) does not depend on cpu speed. You don't hear a bit-to-bit accurate playback because Windows apply internal mixing and sample conversion in real time and the result sounds like crap. You should try DirectSound games like Diablo, Starcraft, Hexen II and hear how they sound, outputting directly:)
High CPU speed does not degrade the output under DOS, as ISA bus speed remains the same. Some games just may not have sound at all on faster machines, like early SCUMM-based games, which use finite loops when waiting for DSP response and just give up too early. The same is with OPL2 - the sirens you talked about earlier. Anyway, it is a user's responsibility which software to run on which hardware.

I can see what you are getting at, if we are talking about WDM drivers, as they remix sound up, which is not a particularly bright idea on an 8-bit output. However, VXD drivers shouldn't be doing too much to the output signal, and I think the Sound Blaster compatible driver is actually a VXD driver.

I'll definitely fire up Diablo though at some stage, see how that sounds though. Still many tests to do before I can narrow down what is causing playback to be significantly worse than I expected.

Unfortunately, I don't have the ability to respond to PMs, as I haven't posted enough. Labs, I'm definitely interested, so please put me on the list for the next batch.

Apologies, all, for the silly post.

- Alex

LABS wrote:

Nevertheless, I just disassembled and decoded to a human-readable form the original firmware of Sound Blaster 2.0 (DSP 2.01). Many unknown things revealed and there is a room for perfection for BB's future firmware releases.

That sounds a little intense. How long did that take? Anything immediately fascinating worth mentioning?

Wow, awesome ! Are you planning to document any of the findings somewhere ?
I do have a question, where did you find information about the ADPCM formats ? I haven't actually found any resource that tells how to decode or encode PCM into the Creative's ADPCM flavor.

If you can decode these DSPs maybe you can decode the SB16 ones as well, so that we can possibly figure out what the MIDI bug really is.

Tiido wrote:

Wow, awesome ! Are you planning to document any of the findings somewhere ?
I do have a question, where did you find information about the ADPCM formats ? I haven't actually found any resource that tells how to decode or encode PCM into the Creative's ADPCM flavor.

Is Creative's ADPCM really different? If not, sox has you probably covered:

https://github.com/chirlu/sox/blob/master/src/adpcms.c

MrSmiley381 wrote:

Anything immediately fascinating worth mentioning?

First noticeable thing was that its command jump table has no dummy code for unimplemented commands. Every function 00-FF has some code to run. However, the jump addresses are repeated. For example Set Time Constant function (0x40) is valid for commands 0x40 - 0x47. I bet some games have their command numbers written incorrectly and still work on the original thing.

A first half of a jump table looks like this (=== are documented command numbers, Lxxxx - labels not yet decoded):

1	; 00 - Resets DSP_BUSY flag
2	AJMP	CMD_00_03				
3	AJMP	CMD_00_03
4	AJMP	CMD_00_03
5	AJMP	CMD_00_03		
6	; 04 - Outputs byte 0x21 to host and starts Timer 0
7	AJMP	CMD_04_07				
8	AJMP	CMD_04_07
9	AJMP	CMD_04_07
10	AJMP	CMD_04_07
11	; 08 - Halts DSP
12	AJMP	CMD_08_0F				
13	AJMP	CMD_08_0F
14	AJMP	CMD_08_0F
15	AJMP	CMD_08_0F
16	AJMP	CMD_08_0F
17	AJMP	CMD_08_0F
18	AJMP	CMD_08_0F
19	AJMP	CMD_08_0F
20	; === 10 
21	AJMP	CMD_10_Direct_PCM_Out		
22	AJMP	CMD_10_Direct_PCM_Out
23	AJMP	CMD_10_Direct_PCM_Out
24	AJMP	CMD_10_Direct_PCM_Out
25	; === 14
26	AJMP	CMD_14_PCM_Single_DMA		
27	AJMP	CMD_14_PCM_Single_DMA
28	; === 16
29	AJMP	CMD_16_ADPCM82_Single_DMA
30	; === 17	
31	AJMP	CMD_17_ADPCM82_Single_DMA_REF
32	; === 18
33	AJMP	CMD_1C_PCM_Auto_DMA
34	AJMP	CMD_1C_PCM_Auto_DMA	
35	; 1A
36	AJMP	L0152	
37	; 1B
38	AJMP	CMD_1F_ADPCM82_Auto_DMA_REF	
39	; === 1C L0143 
40	AJMP	CMD_1C_PCM_Auto_DMA	
41	AJMP	CMD_1C_PCM_Auto_DMA	
42	; 1E
43	AJMP	L0152	
44	; === 1F L016D - ADPCM 82 Auto REF
45	AJMP	CMD_1F_ADPCM82_Auto_DMA_REF
46	; === 20 L018B - PCM Direct Input	
47	AJMP	CMD_20_PCM_Direct_Input 
48	AJMP	CMD_20_PCM_Direct_Input	; 21
49	AJMP	CMD_20_PCM_Direct_Input ; 22
50	AJMP	CMD_20_PCM_Direct_Input ; 23
51	; === 24 L0192 - PCM Single DMA Input
52	AJMP	CMD_24_PCM_Single_DMA_Input
53	AJMP	CMD_24_PCM_Single_DMA_Input ; 25
54	AJMP	CMD_24_PCM_Single_DMA_Input ; 26
55	AJMP	CMD_24_PCM_Single_DMA_Input ; 27
56	; 28
57	AJMP	L01C1 
58	AJMP	L01C1 ; 29
59	AJMP	L01C1 ; 2A
60	AJMP	L01C1 ; 2B

61	AJMP	L01C1 ; 2C
62	AJMP	L01C1 ; 2D
63	AJMP	L01C1 ; 2E
64	AJMP	L01C1 ; 2F
65	; === 30 L01E4 - Polling MIDI Input
66	AJMP	CMD_30_Polling_MIDI_Input 
67	; === 31 L023E - Int MIDI Input
68	AJMP	CMD_31_Int_MIDI_Input 
69	AJMP	L0286 ; 32
70	AJMP	L030B ; 33
71	; === 34 L0394 - UART_Poll_MIDI_IO
72	AJMP	CMD_34_UART_Poll_MIDI_IO
73	; === 35 L03E3	
74	AJMP	CMD_35_UART_Int_MIDI_IO 
75	; === 36 L0436
76	AJMP	CMD_36_UART_Poll_MIDI_IO_Time
77	; === 37 lL04C6	
78	AJMP	CMD_37_UART_Int_MIDI_IO_Time 
79	; === 38 L01D2
80	AJMP	CMD_38_MIDI_Output 
81	AJMP	CMD_38_MIDI_Output ; 39
82	AJMP	CMD_38_MIDI_Output ; 3A
83	AJMP	CMD_38_MIDI_Output ; 3B
84	AJMP	CMD_38_MIDI_Output ; 3C
85	AJMP	CMD_38_MIDI_Output ; 3D
86	AJMP	CMD_38_MIDI_Output ; 3E
87	AJMP	CMD_38_MIDI_Output ; 3F
88	; === 40 L055A
89	AJMP	CMD_40_Set_Time_Constant 
90	AJMP	CMD_40_Set_Time_Constant ; 41
91	AJMP	CMD_40_Set_Time_Constant ; 42
92	AJMP	CMD_40_Set_Time_Constant ; 43
93	AJMP	CMD_40_Set_Time_Constant ; 44
94	AJMP	CMD_40_Set_Time_Constant ; 45
95	AJMP	CMD_40_Set_Time_Constant ; 46
96	AJMP	CMD_40_Set_Time_Constant ; 47
97	; === 48 L056B
98	AJMP	CMD_48_Set_DMA_Block_Size 
99	AJMP	CMD_48_Set_DMA_Block_Size ; 49
100	AJMP	CMD_48_Set_DMA_Block_Size ; 4A
101	AJMP	CMD_48_Set_DMA_Block_Size ; 4B
102	AJMP	CMD_48_Set_DMA_Block_Size ; 4C
103	AJMP	CMD_48_Set_DMA_Block_Size ; 4D
104	AJMP	CMD_48_Set_DMA_Block_Size ; 4E
105	AJMP	CMD_48_Set_DMA_Block_Size ; 4F
106	AJMP	L0582 ; 50
107	AJMP	L058D ; 51
108	AJMP	L0582 ; 52
109	AJMP	L058D ; 53
110	AJMP	L0582 ; 54
111	AJMP	L058D ; 55
112	AJMP	L0582 ; 56
113	AJMP	L058D ; 57
114	AJMP	L0596 ; 58
115	AJMP	L05BE ; 59
116	AJMP	L0596 ; 5A
117	AJMP	L05BE ; 5B
118	AJMP	L0596 ; 5C
119	AJMP	L05BE ; 5D
120	AJMP	L0596 ; 5E
121	AJMP	L05BE ; 5F
122	AJMP	L05E8 ; 60
123	AJMP	L05E8
124	AJMP	L05E8
125	AJMP	L05E8
126	AJMP	L05EC	; 64
127	AJMP	L05EC
128	AJMP	L05EC
129	AJMP	L05EC
130	AJMP	L05FE	; 68
131	AJMP	L05FE
132	AJMP	L05FE
133	AJMP	L05FE
134	AJMP	L05FE
135	AJMP	L05FE
136	AJMP	L05FE
137	AJMP	L05FE	
138	AJMP	L0605	; 70
139	AJMP	CMD_7D_ADPCM84_Auto_DMA_REF	; 71
140	AJMP	L0668	; 72
141	AJMP	CMD_7F_ADPCM83_Auto_DMA_REF	; 73
142	; === 74
143	AJMP	CMD_74_ADPCM84_Single_DMA
144	; === 75	
145	AJMP	CMD_75_ADPCM84_Single_DMA_REF 
146	; === 76
147	AJMP	CMD_76_ADPCM83_Single_DMA 
148	; === 77
149	AJMP	CMD_77_ADPCM83_Single_DMA_REF 
150	AJMP	L0605 ; 78
151	AJMP	CMD_7D_ADPCM84_Auto_DMA_REF ; 79
152	AJMP	L0668	; 7A
153	AJMP	CMD_7F_ADPCM83_Auto_DMA_REF ; 7B
154	AJMP	L0605	; 7C
155	; === 7D
156	AJMP	CMD_7D_ADPCM84_Auto_DMA_REF 
157	AJMP	L0668	; 7E
158	; === 7F
159	AJMP	CMD_7F_ADPCM83_Auto_DMA_REF 

ajacocks wrote:

Unfortunately, I don't have the ability to respond to PMs, as I haven't posted enough. Labs, I'm definitely interested, so please put me on the list for the next batch.

Apologies, all, for the silly post.

- Alex

Added you to the list.

Tiido wrote:

Wow, awesome ! Are you planning to document any of the findings somewhere ?
I do have a question, where did you find information about the ADPCM formats ? I haven't actually found any resource that tells how to decode or encode PCM into the Creative's ADPCM flavor.

Yes, this type of document will be useful. I would gladly share the source code with anyone who is interested as well and wish to collaborate in translating its darkest corners.
I did not get to the details of ADPCM code yet, so can't say anything about it, root42. BlasterBoard uses this one https://wiki.multimedia.cx/index.php/Creative_8_bits_ADPCM, it works great. DosBox uses tables instead, but I traced both methods with one ADPCM83 source and got to the same waveform at the end.

PS: Did anyone try any ADPCM-enabled games on BB?

You might be pleased to know that my blasterboard's DAC actually works perfectly fine on the 486 under windows 95. The level of noise and distortion is largely comparable to the level displayed by my Sound Blaster AE-5 when playing the same 8-bit content.

The AE-5 might still have a little bit of an edge over the blasterboard, but it is pretty close.

canthearu wrote:

You might be pleased to know that my blasterboard's DAC actually works perfectly fine on the 486 under windows 95. The level of noise and distortion is largely comparable to the level displayed by my Sound Blaster AE-5 when playing the same 8-bit content.

The AE-5 might still have a little bit of an edge over the blasterboard, but it is pretty close.

Oh, that's a bold comparison, thank you.
Is it W98 or W95 on your P3 installed where the sound is poor?

EDIT:
Can you check "Hardware version:" string under "Sound Blaster or compatible" Properties in Device Manager on both 486 and P3?

System specs:

Intel Pentium 3 600mhz Katmai (100fsb)
Asus P2B v1.1
512meg RAM
Voodoo 3-3000 GPU
Intel 100mbit PCI ethernet.
NEC USB 2.0 card
Windows 98 SE

LABS wrote:

Can you check "Hardware version:" string under "Sound Blaster or compatible" Properties in Device Manager on both 486 and P3?

I can't seem to find that particular detail on either computer.

In further testing, the P3 isn't as bad as I thought. If I play the sound clip in media player, it is quite badly distorted, while if I play the same clip in sound recorder, it is mostly OK. So it looks like windows media player, assuming that you would have a 16bit sound card, was munting the sound output, probably due to rounding errors.

I still get a lot of clicking on the P3 when playing though sound recorder though, I wonder if your newer firmware has already fixed that.

canthearu wrote:

System specs: […]
Show full quote

System specs:

Intel Pentium 3 600mhz Katmai (100fsb) Asus P2B v1.1 512meg RAM Voodoo 3-3000 GPU Intel 100mbit PCI ethernet. NEC USB 2.0 card Windows 98 SE

I have poor performance on W98 too. Sounds like W98 driver is downsampling the stream to like 4 or 8kHz.
Is it "Hardware version: Not available" on both machines?

LABS wrote:

I have poor performance on W98 too. Sounds like W98 driver is downsampling the stream to like 4 or 8kHz.
Is it "Hardware version: Not available" on both machines?

I get that hardware version: not avaliable.

However, have you tried playing the wave file through sound recorder rather than media player (the default player)?

canthearu wrote:

However, have you tried playing the wave file through sound recorder rather than media player (the default player)?

Yes, it is ok. Still need to get deeper into that later anyway. I have a P3-1GHz with ISA as well, but it is not at my place at the moment.

2 days of thorough SB2.0 firmware decoding answered a lot of questions concerning internal processes that affect the behavior of the card in different register programming situations and combinations. But still there is some code (about 5-6%) left with unknown yet functionality. Still needs some work.
I think I should document my findings later somehow - can be useful for emulator developers for example. At least I did not find any doc like that on the net.

Based on the acquired superpowers there will be a hardcore firmware upgrade for BB for sure.

I‘ve finally finished my card, too. Took me a while just soldering parts here and there when there was time on hand. It might not be the cleanest looking Blasterboard around, far from it, but it worked first try so thats good with me (well technically second try but I think jumpering it wrong doesn’t count here 😉 ). Soldering the card was pretty smooth sailing, spacing on the resistors was a a bit tight for my clumsy hands but all went well. Overall pretty impressed with the kit and the card, especially the noise level is quite impressive! Thanks again.

Which is the best Homebrew Soundcard? by root42

Just received a fresh batch of PCBs

However, still upgrading the firmware, which will take some time.