╔═════════════════════════════════════╗
                 ║ Programming the SoundBlaster 16 DSP ║
                 ║      Written by Ethan Brodsky       ║
                 ║     (ericbrodsky@psl.wisc.edu)      ║
                 ║            Version 3.3              ║
                 ║              5/18/96                ║
                 ╚═════════════════════════════════════╝

┌────────────┬───────────────────────────────────────────────────────────────
│ Disclaimer │
└────────────┘

This information is distributed AS IS.  The author specifically disclaims
responsibility for any loss of profit or any consequential, incidental,
or other damages resulting from the use or misuse of this information.
This information may be freely distributed in any form as long as this
disclaimer remains intact.

┌──────────────┬─────────────────────────────────────────────────────────────
│ Introduction │
└──────────────┘

The Sound Blaster 16 is capable of both FM and digitized sounds.  Digitized
sound capibilities range from 8-bit mono 5000 HZ sound all the way up to
16-bit stereo sound at 44khz.  This FAQ documents programming the SB16 DSP
CT1341 chip for recording and playback of digitized audio.  Prior knowledge
on programming earlier Sound Blaster sound cards is necessary.

┌────────────────────────────────────┬───────────────────────────────────────
│ The Sound Blaster 16 DSP I/O Ports │
└────────────────────────────────────┘

The SB16's DSP chip is programming using several I/O ports at a base I/O
address determined by jumper settings.  On the SB16, there are 16 I/O ports
which are used for FM synthesized music, mixer settings, DSP programming and
CD-ROM access.  Five of these ports are used in programming the DSP.  They
are listed below.

  2x6h - DSP Reset
  2xAh - DSP Read
  2xCh - DSP Write (Command/Data), DSP write-buffer status (Bit 7)
  2xEh - DSP Read-buffer status (Bit 7), DSP interrupt acknowledge
  2xFh - DSP 16-bit interrupt acknowledge

┌───────────────────┬────────────────────────────────────────────────────────
│ Resetting the DSP │
└───────────────────┘

You have to reset the DSP before you can program it.  The DSP can be reset
using the following procedure:
  1)  Write a 1 to the reset port (2x6)
  2)  Wait for 3 microseconds
  3)  Write a 0 to the reset port (2x6)
  4)  Poll the read-buffer status port (2xE) until bit 7 is set
  5)  Poll the read data port (2xA) until you receive an AA

The DSP usually takes about 100 microseconds to initialized itself.  After
this period of time, if the return value is not AA or there is no data at
all, then the SB card may not be installed or an incorrect I/O address is
being used.

┌────────────────────┬───────────────────────────────────────────────────────
│ Writing to the DSP │
└────────────────────┘

To write a byte to the SB16, the following procedure should be used:
  1)  Read the write-buffer status port (2xC) until bit 7 is cleared
  2)  Write the value to the write port (2xC)

┌─────────────────┬──────────────────────────────────────────────────────────
│ Reading the DSP │
└─────────────────┘

To read a byte from the SB16, the following procedure should be used:
  1)  Read the read-buffer status port (2xE) until bit 7 is set
  2)  Read the value from the read port (2xA)

┌────────────────────────────────┬───────────────────────────────────────────
│ Programming the DMA Controller │
└────────────────────────────────┘

The DMA (Direct Memory Access) controller controls data transfers between
I/O devices and memory without using the CPU.  An Intel 8237 DMAC integrated
circut is used to control this.  An IBM compatible computer has two DMA
controllers, one for 8-bit transfers and one for 16-bit transfers.  The DMA
controller, coupled with an external page register, is capable of transfering
blocks of up 64k to the SB16.  Here is information on I/O ports and register
settings necessary for sound card I/O:

 I/O port addresses for the DMA Address and Count Registers
  ╔════════════╤═════════════╤═══════════════════╗
  ║ Controller │ I/O address │     Function      ║
  ╠════════════╪═════════════╪═══════════════════╣
  ║   DMA 1    │     00      │ Channel 0 address ║
  ║   8-bit    │     01      │ Channel 0 count   ║
  ║   Slave    │     02      │ Channel 1 address ║
  ║            │     03      │ Channel 1 count   ║
  ║            │     04      │ Channel 2 address ║
  ║            │     05      │ Channel 2 count   ║
  ║            │     06      │ Channel 3 address ║
  ║            │     07      │ Channel 3 count   ║
  ╠════════════╪═════════════╪═══════════════════╣
  ║   DMA 2    │     C0      │ Channel 4 address ║
  ║   16-bit   │     C2      │ Channel 4 count   ║
  ║   Master   │     C4      │ Channel 5 address ║
  ║            │     C6      │ Channel 5 count   ║
  ║            │     C8      │ Channel 6 address ║
  ║            │     CA      │ Channel 6 count   ║
  ║            │     CC      │ Channel 7 address ║
  ║            │     CE      │ Channel 7 count   ║
  ╚════════════╧═════════════╧═══════════════════╝

 I/O port addresses for the control registers
  ╔═════════════╤═══════════╤══════════════════════════════╗
  ║   Address   │ Operation │           Function           ║
  ║ DMAC1 DMAC2 │           │                              ║
  ╠═════════════╪═══════════╪══════════════════════════════╣
  ║  0A    D4   │   Write   │ Write single mask register   ║
  ║  0B    D6   │   Write   │ Write mode register          ║
  ║  0C    D8   │   Write   │ Clear byte pointer flip-flop ║
  ╚═════════════╧═══════════╧══════════════════════════════╝

 I/O port addresses for lower page registers
  ╔═════════╤════════════════════════════╗
  ║ Address │  Function                  ║
  ╠═════════╪════════════════════════════╣
  ║   81    │  8-bit DMA channel 2 page  ║
  ║   82    │  8-bit DMA channel 3 page  ║
  ║   83    │  8-bit DMA channel 1 page  ║
  ║   87    │  8-bit DMA channel 0 page  ║
  ║   89    │  16-bit DMA channel 6 page ║
  ║   8A    │  16-bit DMA channel 7 page ║
  ║   8B    │  16-bit DMA channel 5 page ║
  ╚═════════╧════════════════════════════╝

 Mode register bit assignments
  ╔═══════════╤═════════════════════════════════╗
  ║ Bit/Value │ Function                        ║
  ╠═══════════╪═════════════════════════════════╣
  ║ Bits 7:6  │ Mode selection bits             ║
  ║    00     │  Demand mode selected           ║
  ║    01     │  Single mode selected           ║
  ║    10     │  Block mode selected            ║
  ║    11     │  Cascade mode selected          ║
  ╠═══════════╪═════════════════════════════════╣
  ║   Bit 5   │ Address increment/decrement bit ║
  ║     1     │  Address decrement selected     ║
  ║     0     │  Address increment selected     ║
  ╠═══════════╪═════════════════════════════════╣
  ║   Bit 4   │ Auto-initialization enable bit  ║
  ║     1     │  Auto-initialized DMA selected  ║
  ║     0     │  Single-cycle DMA selected      ║
  ╠═══════════╪═════════════════════════════════╣
  ║ Bits 3:2  │ Transfer bits                   ║
  ║    00     │  Verify transfer                ║
  ║    01     │  Write transfer (To memory)     ║
  ║    10     │  Read transfer (From memory)    ║
  ║    11     │  Illegal                        ║
  ║    **     │  Ignored if bits 7:6 = 11       ║
  ╠═══════════╪═════════════════════════════════╣
  ║ Bits 1:0  │ Channel selection bits          ║
  ║    00     │  Select channel 0 (4)           ║
  ║    01     │  Select channel 1 (5)           ║
  ║    10     │  Select channel 2 (6)           ║
  ║    11     │  Select channel 3 (7)           ║
  ╚═══════════╧═════════════════════════════════╝

 Write single mask bit assignments
  ╔═══════════╤══════════════════════════════════╗
  ║ Bit/Value │ Function                         ║
  ╠═══════════╪══════════════════════════════════╣
  ║ Bits 7:3  │ Unused (Set to 0)                ║
  ╠═══════════╪══════════════════════════════════╣
  ║   Bit 2   │ Set/clear mask bit               ║
  ║     1     │  Set mask bit (Disable channel)  ║
  ║     0     │  Clear mask bit (Enable channel) ║
  ╠═══════════╪══════════════════════════════════╣
  ║ Bits 1:0  │ Channel selection bits           ║
  ║    00     │  Select channel 0 (4)            ║
  ║    01     │  Select channel 1 (5)            ║
  ║    10     │  Select channel 2 (6)            ║
  ║    11     │  Select channel 3 (7)            ║
  ╚═══════════╧══════════════════════════════════╝

DMAC2 is used for 16-bit I/O and DMAC1 is used for 8-bit I/O.  The procedure
for starting a transfer is complicated, so I'll list the steps for starting
the type of DMA transfers used for sound I/O:

  1)  Calculate the absolute linear address of your buffer
        LinearAddr := Seg(Ptr^)*16 + Ofs(Ptr^));

  2)  Disable the sound card DMA channel by setting the appropriate mask bit
        Port[MaskPort] := 4 + (Channel mod 4);

  3)  Clear the byte pointer flip-flop
        Port[ClrBytePtr] := AnyValue;

  4)  Write the DMA mode for the transfer
      The mode selection bits should be set to 01 for single-mode.  The
      address inc/dec bit should be set to 0 for address increment.  The
      auto-initialization bit should be set appropriately.  I will discuss
      auto-initialized DMA later.  The transfer bits should be set to 10
      for playback and 01 for recording.  The channel select should be
      set to the sound card DMA channel.  Be aware that "read" means a read
      from memory (Write to sound card) and that "write" means a write to
      system memory (Read from sound card)
        Port[ModePort] := Mode + (Channel mod 4);
      Some often used modes are:
        48h+Channel - Single-cycle playback
        58h+Channel - Auto-initialized playback
        44h+Channel - Single-cycle recording
        54h+Channel - Auto-initialized recording

  5)  Write the offset of the buffer, low byte followed by high byte.  For
      sixteen bit data, the offset should be in words from the start of a
      128kbyte page. The easiest method for computing 16-bit parameters is
      to divide the linear address by two before calculating offset.
        if SixteenBit
          then
            begin
              BufOffset := (LinearAddr div 2) mod 65536;
              Port[BaseAddrPort] := Lo(BufOffset);
              Port[BaseAddrPort] := Hi(BufOffset);
            end
          else
            begin
              BufOffset := LinearAddr mod 65536;
              Port[BaseAddrPort] := Lo(BufOffset);
              Port[BaseAddrPort] := Hi(BufOffset);
            end;

  6)  Write the transfer length, low byte followed by high byte.  For an
      8-bit transfer, write the number of bytes-1.  For a 16-bit transfer,
      write the number of words-1.
        Port[CountPort] := Lo(TransferLength-1);
        Port[CountPort] := Hi(TransferLength-1);

  7)  Write the buffer page to the DMA page register.
        Port[PagePort] := LinearAddr div 65536;

  8)  Enable the sound card DMA channel by clearing the appropriate mask bit
      Port[MaskPort] := DMAChannel mod 4;

┌───────────────────────────┬────────────────────────────────────────────────
│ Setting the sampling rate │
└───────────────────────────┘

Unlike earlier Sound Blasters, the SB16 is programmed with actual sampling
rates instead of time constants.  On the SB16, the sampling rate is set
using DSP commands 41h and 42h.  Command 41h is used for output and 42h is
used for input.  I have heard that on the SB16, both these command currently
do the same thing, but I would recommend using the individual commands to
guarantee compatibility with future sound cards.  The procedure for setting
the sampling rate is:
  1)  Write the command (41h for output rate, 42h for input rate)
  2)  Write the high byte of the sampling rate (56h for 22050 hz)
  3)  Write the low byte of the sampling rate  (22h for 22050 hz)

┌─────────────────────┬──────────────────────────────────────────────────────
│ Digitized sound I/O │
└─────────────────────┘

To record or play back sound, you should use the following sequence:

  1)  Allocate a buffer that does not cross a 64k physical page boundary
  2)  Install an interrupt service routine
  3)  Program the DMA controller for background transfer
  4)  Set the sampling rate
  5)  Write the I/O command to the DSP
  6)  Write the I/O transfer mode to the DSP
  7)  Write the block size to the DSP (Low byte/High byte)

Upon interrupt when using single-cycle DMA:
  1)  Program DMA controller for next block
  2)  Program DSP for next block
  3)  Copy next block if double-buffering
  4)  Acknowledge the interrupt with the SB by reading from port 2xE for
      8-bit sound or port 2xF for 16-bit sound.
  5)  Acknowledge the end of interrupt with the PIC by writing 20h to port
      20h. If the sound card is on IRQ8-15, you must also write 20h to A0h.

┌──────────────┬─────────────────────────────────────────────────────────────
│ DSP commands │
└──────────────┘

 D0 - Pause 8-bit DMA mode digitized sound I/O initiated by command Cxh.
      Applicable to both single-cycle and auto-initialized DMA I/O.
 D4 - Continue 8-bit DMA mode digitized sound I/O paused using command D0.
      Applicable to both single-cycle and auto-initialzied DMA I/O.

 D5 - Pause 16-bit DMA mode digitized sound I/O initiated by command Bxh.
      Applicable to both single-cycle and auto-initialized DMA I/O.
 D6 - Continue 16-bit DMA mode digitized sound I/O paused using command D5
      Applicable to both single-cycle and auto-initialized DMA I/O.

 D9 - Exit 16-bit auto-initialized DMA mode digitized sound I/O after the
      end of the current block.
 DA - Exit 8-bit auto-initialized DMA mode digitized sound I/O after the
      end of the current block.

 E1 - Get DSP version number.  After sending this command, read back two
      bytes form the DSP.  The first byte is the major version number and
      the second byte is the minor version number.  A SB16 should have a
      DSP version of 4.00 or greater.  Check this before using an SB16
      specific commands.

 Bx - Program 16-bit DMA mode digitized sound I/O
      Command sequence:  Command, Mode, Lo(Length-1), Hi(Length-1)
       Command:
        ╔════╤════╤════╤════╤═══════╤══════╤═══════╤════╗
        ║ D7 │ D6 │ D5 │ D4 │  D3   │  D2  │  D1   │ D0 ║
        ╠════╪════╪════╪════╪═══════╪══════╪═══════╪════╣
        ║  1 │  0 │  1 │  1 │  A/D  │  A/I │ FIFO  │  0 ║
        ╚════╧════╧════╧════┼───────┼──────┼───────┼════╝
                            │ 0=D/A │ 0=SC │ 0=off │
                            │ 1=A/D │ 1=AI │ 1=on  │
                            └───────┴──────┴───────┘
       Common commands:
         B8 - 16-bit single-cycle input
         B0 - 16-bit single-cycle output
         BE - 16-bit auto-initialized input
         B6 - 16-bit auto-initialized output

       Mode:
        ╔════╤════╤══════════╤════════════╤════╤════╤════╤════╗
        ║ D7 │ D6 │    D5    │     D4     │ D3 │ D2 │ D1 │ D0 ║
        ╠════╪════╪══════════╪════════════╪════╪════╪════╪════╣
        ║  0 │  0 │  Stereo  │   Signed   │  0 │  0 │  0 │  0 ║
        ╚════╧════┼──────────┼────────────┼════╧════╧════╧════╝
                  │ 0=Mono   │ 0=unsigned │
                  │ 1=Stereo │ 1=signed   │
                  └──────────┴────────────┘

 Cx - Program 8-bit DMA mode digitized sound I/O
      Same procedure as 16-bit sound I/O using command Bx
       Common commands:
         C8 - 8-bit single-cycle input
         C0 - 8-bit single-cycle output
         CE - 8-bit auto-initialized input
         C6 - 8-bit auto-initialized output

The FIFO is used to eliminate inconsistencies in the sample period
when the sound card is not able to get DMA when it needs it.  With
FIFO disabled, the card attempts DMA at precisely the instant that
it needs a sample.  If another device with a higher priority is
accessing DMA, the sound card waits for the sample and the sampling
rate may be decreased.  The FIFO allows the DMA sample period to be
more erratic without affecting the audio quality.  The FIFO is cleared
whenever a command is sent to the DSP.  In Single-cycle mode, the DSP
is constantly being reprogrammed.  The FIFO may still contain data which
has not been output when it cleared.  To avoid this problem, the FIFO
should be turned off for single-cycle mode.  When in auto-initialized
mode, the DSP is never reprogrammed.  The FIFO can be left on and sound
quality will be improved.

┌──────────────────────┬─────────────────────────────────────────────────────
│ Auto-initialized DMA │
└──────────────────────┘

When single-cycle DMA is used, sound output stops at the end of each block.
The interrupt handler can start another transfer, but there will be a break
in output.  This causes a click between each block, reducing sound quality.
When auto-initialized DMA is used, sound output loops around at the end of
the buffer.  The DMA controller keeps transfering the same block of memory
that the DMA transfer was initiated with.  When the end of the buffer is
reached, it will start sending the buffer again by auto-initializing the
current offset and count registers with the values stored in the base offset
and count registers.  The usual method for achieving click-less output is to
allocate a buffer and divide it into two blocks.  Program the DMA controller
with the length of the whole buffer, but program the SB16 with the length of
a block. (Half of the buffer)  An interrupt occurs for each sound card block,
so two interrupts will occur each time the buffer is played, once at the
midpoint (Start of the second block) and once at the end (In effect, the
start of the first block)  The interrupt handler should copy data into the
block that was just finished so that the data is ready when it is needed for
output.  The programming procedure for an auto-initialized DMA transfer is
identical to the procedure for a single-cycle DMA transfer, except that bit
4 of the DMA mode register and bit 3 of the DSP command are set.

Upon interrupt when using auto-initialized DMA:
  1)  Copy next chunk into output buffer block that just finished
  2)  Acknowledge the interrupt with the SB by reading from port 2xE for
      8-bit sound or port 2xF for 16-bit sound.
  3)  Acknowledge the end of interrupt with the PIC by writing 20h to port
      20h. If the sound card is on IRQ8-15, you must also write 20h to A0h.

To stop sound immediately:
  8-bit  - Write DSP command D0h (Pause 8-bit DMA mode digitized sound I/O)
  16-bit - Write DSP command D5h (Pause 16-bit DMA mode digitized sound I/O)
  (Stops sound immediately, without an interrupt)

To stop the sound at the end of the currently block:
  8-bit  - Write DSP command DAh (Stop 8-bit auto-init DMA sound I/O)
  16-bit - Write DSP command D9h (Stop 16-bit auto-init DMA sound I/O)
  (These two commands will stop the sound at the end of the current
  block.  If your program is not prepared for an interrupt after output
  is finished, it may cause problems)

You can also end auto-initialized mode by reprogramming the DSP for
single-cycle mode.  The card then switches from A/I mode to S/C mode after
the next interrupt.  It will then contiue to play or record for the length
specified, generate an interrupt and stop.  This will allow you to stop
output exactly at the end of the data, without requiring the remainder of
the DMA buffer to be filled with silence.  This technique may or may not
be useful to you.  I would recommend using the pause commands documented
in in the immediate stop section unless another method is more suited to
your purpose.

┌────────────┬───────────────────────────────────────────────────────────────
│ References │
└────────────┘

 Title:         Developer Kit for the Sound Blaster Series, Second Edition.
 Publishers:    Creative Technology Ltd.

 Title:         Intel 486 Microcomputer Model 401 Board Technical Ref. Manual
 Publisher:     Intel Corporation
 Order-number:  504366-002

 Title:         8237A High Performance Programmable DMA Controller specs
 Publisher:     Intel Corporation
 Order-number:  231466-005

 Title:         8259A Programmable Interrupt Controller specs
 Publisher:     Intel Corporation
 Order-number:  231468-003

 Title:         SMIX Sound System
 Author:        Ethan Brodsky
 Location:      x2ftp.oulu.fi /pub/msdos/programming/mxlibs/smix*.zip
                (May be moved to another directory)

 Title:         SB16SND code
 Author:        Ethan Brodsky
 Location:      http://www.xraylith.wisc.edu/~ebrodsky/other/sb16snd.zip

Thanks to Douglas Kaden at Creative Labs for information on 16-bit DMA, 
FIFO mode, and numerous other topics.