Re: Decoding kryoflux stream for HP9895A

2016-09-25 Thread Al Kossow 
OK, I'll see what I can do. You left some Intel floppies of them out here, and
I've got some Amigo OS disks that are pretty important to image.
Guess I need to get the Diskferret set up ;-)

On 9/25/16 5:34 PM, Eric Smith wrote:
> On Sun, Sep 25, 2016 at 6:25 PM, Al Kossow  wrote:
>> Is it doing better on Intel M2FM now?
> 
> No, but since it had no trouble with the HP M2FM, I'm starting to
> think that perhaps my PLL code is OK and the drive used to read those
> Intel disks might have been suboptimal.
> 
> I need more double-density flux transition disk images to test with.
>

Re: Decoding kryoflux stream for HP9895A

2016-09-25 Thread Eric Smith 
On Sun, Sep 25, 2016 at 6:25 PM, Al Kossow  wrote:
> Is it doing better on Intel M2FM now?

No, but since it had no trouble with the HP M2FM, I'm starting to
think that perhaps my PLL code is OK and the drive used to read those
Intel disks might have been suboptimal.

I need more double-density flux transition disk images to test with.

Re: Decoding kryoflux stream for HP9895A

2016-09-25 Thread Al Kossow 
Is it doing better on Intel M2FM now?

On 9/25/16 4:48 PM, Eric Smith wrote:
> fluxtoimd now successfully decodes HP M2FM double-density floppy
> format, as used by the HP 7902, 9885, and 9895, from either DiscFerret
> images or ZIP files of KryoFlux stream files.
> 
> https://github.com/brouhaha/fluxtoimd
>

Re: Decoding kryoflux stream for HP9895A

2016-09-25 Thread Eric Smith 
fluxtoimd now successfully decodes HP M2FM double-density floppy
format, as used by the HP 7902, 9885, and 9895, from either DiscFerret
images or ZIP files of KryoFlux stream files.

https://github.com/brouhaha/fluxtoimd

Re: Decoding kryoflux stream for HP9895A

2016-09-25 Thread Eric Smith 
On Tue, Sep 20, 2016 at 7:42 AM, Denise de Vries
 wrote:
> Does anyone know of documentation for the HP9895A format with its own M2FM 
> encoding?
> I have a kryoflux preservation stream but so far can make no sense of it.

On Wed, Sep 21, 2016 at 12:59 AM, Christian Corti
 wrote:
> I've already written a decoder for the HP 9885/9895 format. Denise already
> sent me his image, but I haven't had the time to try decoding it yet...

Denise sent me the image as well.

dfitoimd has now been renamed to fluxtoimd, as it now supports both
DiscFerret and KryoFlux image input. I've partially added HP M2FM
format; it now correctly decodes the address fields, but I have a bit
more work to do on the data fields. I expect to have it completed
tomorrow.

> - 0xD0: ECC data address mark

Any idea how they use that? Aside from that being included in the list
of address marks in the service manuals, it's not described at all.
Since the 7902/9885/9895 don't do ECC, perhaps it was reserved for
future use.

Re: Decoding kryoflux stream for HP9895A

2016-09-21 Thread Eric Smith 
On Wed, Sep 21, 2016 at 12:59 AM, Christian Corti
 wrote:
> I've already written a decoder for the HP 9885/9895 format. Denise already
> sent me his image, but I haven't had the time to try decoding it yet...

I'd still like to get a 9885/9895/7902 M2FM flux-transition disk
image, if anyone has one handy that they'd be willing to share.

Re: Decoding kryoflux stream for HP9895A

2016-09-21 Thread Eric Smith 
On Wed, Sep 21, 2016 at 11:19 AM, Al Kossow  wrote:
> I wonder if this is the same format the Amigo used?

The Amigo used the 7902A floppy drive, which supports the same formats
as the 9895A.  This includes HP M2FM, single or double-sided, and IBM
3740 FM single-sided single-density.

Re: Decoding kryoflux stream for HP9895A

2016-09-21 Thread Al Kossow 
HP300

On 9/21/16 10:24 AM, tony duell wrote:
> [HP9895 format deleted]
> 
>> I wonder if this is the same format the Amigo used?
> 
> Which was the Amigo? 
> 
> FWIW every 5.25" and 3.5" HP drive unit that I've seen (and the
> internal drives on the 9826 and 9836) use a Western Digital or
> compatible disk controller chip. The low-level format (sync
> bytes, headers, etc) are what you'd expect for such an IC.
> 
> -tony
>

RE: Decoding kryoflux stream for HP9895A

2016-09-21 Thread tony duell 
[HP9895 format deleted]

> I wonder if this is the same format the Amigo used?

Which was the Amigo? 

FWIW every 5.25" and 3.5" HP drive unit that I've seen (and the
internal drives on the 9826 and 9836) use a Western Digital or
compatible disk controller chip. The low-level format (sync
bytes, headers, etc) are what you'd expect for such an IC.

-tony

Re: Decoding kryoflux stream for HP9895A

2016-09-21 Thread Al Kossow 


On 9/20/16 11:59 PM, Christian Corti wrote:

> Here's a rough sketchup of what my decoder does:
> 
> - Find the sync pattern (several 0x55)
> - Swap the bits in each byte (i.e. MSB<-->LSB)
> - Look for an address mark (clock pattern always 0x0E)
>   - 0x70: ID address mark
> Calculate CRC16 (initial value 0x) over four bytes
>   - 0x50: data address mark
>   - 0xF0: defective track address mark
>   - 0xD0: ECC data address mark
> 
> - The ID address mark contains the following information:
>   0: cylinder
>   1: head in bit 7, sector in bits 5..0
>   2: CRC byte 1
>   3: CRC byte 2
> 

I wonder if this is the same format the Amigo used?

Re: Decoding kryoflux stream for HP9895A

2016-09-21 Thread Christian Corti 

On Tue, 20 Sep 2016, Eric Smith wrote:

On Tue, Sep 20, 2016 at 7:42 AM, Denise de Vries
 wrote:

Does anyone know of documentation for the HP9895A format with its own M2FM 
encoding?
I have a kryoflux preservation stream but so far can make no sense of it.


I've successfully decoded Intel M2FM disks.  The track-level format
isn't the same as the HP 9895A, but the channel code is.  I'd be happy
to take a look at a 9895A image.


I've already written a decoder for the HP 9885/9895 format. Denise already 
sent me his image, but I haven't had the time to try decoding it yet...


Here's a rough sketchup of what my decoder does:

- Find the sync pattern (several 0x55)
- Swap the bits in each byte (i.e. MSB<-->LSB)
- Look for an address mark (clock pattern always 0x0E)
  - 0x70: ID address mark
Calculate CRC16 (initial value 0x) over four bytes
  - 0x50: data address mark
  - 0xF0: defective track address mark
  - 0xD0: ECC data address mark

- The ID address mark contains the following information:
  0: cylinder
  1: head in bit 7, sector in bits 5..0
  2: CRC byte 1
  3: CRC byte 2



Christian

Re: Decoding kryoflux stream for HP9895A

2016-09-20 Thread Eric Smith 
On Tue, Sep 20, 2016 at 7:42 AM, Denise de Vries
 wrote:
> Does anyone know of documentation for the HP9895A format with its own M2FM 
> encoding?
> I have a kryoflux preservation stream but so far can make no sense of it.

I've successfully decoded Intel M2FM disks.  The track-level format
isn't the same as the HP 9895A, but the channel code is.  I'd be happy
to take a look at a 9895A image.