Here is my info on the tape data:
/* start clip from file */
Korg Polysix Tape Format
Info from Old Crow's Synth Shop
The tape interface is a simple one, based on the Kansas City Standard
for FM data-to-tape. A tape dump looks like this:
1) 10,240 "1" bits of leader tone
2) A 2-byte header; the 2 bytes are 0x50 and 0x36 (ASCII "P6")
3) 512 bytes of patch data for 32 patches
4) One-byte checksum of data
5) 1,536 "1" bits of trailer tone
Bytes are framed by one "0" start bit and two "1" stop bits, making
a single byte actually use 11 bits. Also, note that bytes are sent
LSB *FIRST*. That is 0x50 = 0b01010000 is sent 00001010.
LSB^ ^MSB
The "0" and "1" bits are designated by waveforms with periods of 640us
(1562.5Hz) and 320us (3125Hz), respectively. Note that the waveform
polarity is irrelevant, for example a "1" bit can be represented by a
_ _ _
_| |_ or by a |_|. Only the interval matters. Thus, 0x50 encoded in
FM data format with 1 start and 2 stop bits can appear as either of:
_ __ __ _ _ _ _...
Case#1: |__| |__| |__| |__| |__| |_|
Start bit->0 0 0 0 0 1 0 1 0 1 1<-Stop bits
__ __ __ __ __ _
Case#2: _| |__| |__| |_| |_| |_| |_...
\LSB____ 0x50 ____MSB/
The checksum is just the lower 8 bits of the sum of all 512 bytes of
the RAM data (carry outs of the sum aren't even stored; they just
"fall off" the data byte each time it exceeds 0xFF).
/* stop clip from file */
I made a simple PIC circuit with the factory patch data coded in ROM so
I can reload repaired boards without fumbling with tape recorders and
levels. Just connect it to the tape in jack, set the tape load mode
on the P6 and hit the single button on the PIC widget to zap the data
across.
Crow
/**/
On Thu, 4 Sep 2003, riccardo231057 wrote:
> Do you know if the specs of the sound of the tape interface is
> available somewhere?
>
> I will like to build a patch editor for my Polysix (easy task), to
> group patches in set of 32 (easy too) and to self generate the wav
> for that bank (not so easy).
>
> So, what I need to know is the base sound (shape and frequency) and
> its modulation related to values encoding.
>
> Thanks a lot for any help.
>
> Riccardo
