Track reading (Sedoric / FDC)

Anything related to the tools Tap2Wav, Tap2CD, Tap2Dsk, Sedoric Disc Manager, Tape Header Creator, WriteDsk, and generaly speaking tools related to the management of Oric data files and devices.
User avatar
Symoon
Archivist
Posts: 2301
Joined: Sat Jan 14, 2006 12:44 am
Location: Paris, France

Re: Track reading (Sedoric / FDC)

Post by Symoon »

Thanks for all the replies guys ;)
Godzil wrote:Also, if the floppy have some defect, it is possible that the hardware try to resync (the $c2)
Well I tested 2 different controllers (mine and NightBird's one), 3 drives (including 3" or 3.5") and about 5 or 6 different disks. On each test I read the same track several times and compared the results, that's how I saw those $C2 bytes.
I haven't been further so far in the reliability of the reading, this was already a too low reliability for me ;) Even with multiple reads and comparison it would bee complicated (because of both inserted or crushed bytes)
Godzil wrote:There is one thing I'm not sure to understand, some bytes are sometimes replaced by $C2, sometimes between valid bytes, there is some $C2 added?
Exactly Sir!
Godzil wrote:Could you measure the time needed to read the track? It may help to understand if there is some read errors.
Not before this weekend, but I'd say between 5 to 8 seconds, from memory.
Godzil wrote:The FDC Read Track is not 100% reliable, I suspect that nibble will read the same track multiple times, and do some check before "validate" it.
For my culture, the FDC doc talks about an "AM" the chip is waiting for... What's an "AM"?
Dbug wrote:He probably want to write a program that can do a raw copy of the floppy, independently of the number of sectors, tracks, special gaps, protections, etc.
Exactly. To answer to Chema, the idea was to modify my old Savedisk/Cloaddsk tool (that created a DSK file from a Sedoric disk, by transferring it thru the tape port), so it could transfer any disk (Jasmin, Oricdos...), including the gaps that sometimes hide a protection (like for XL-Dos). The existing tool only reads Sedoric and only transfers sectors, hence losing some information.
I'm not sure it could have been achieved without checking the disk format anyway, but if the track reading is not reliable, it trashes the simple and "universal" solution.
User avatar
Symoon
Archivist
Posts: 2301
Joined: Sat Jan 14, 2006 12:44 am
Location: Paris, France

Re: Track reading (Sedoric / FDC)

Post by Symoon »

Symoon wrote:For my culture, the FDC doc talks about an "AM" the chip is waiting for... What's an "AM"?
Found it in Sedoric à Nu: Address Mark. Not sure what it is exactly yet, but at least I know what it means ;)
Godzil
Squad Leader
Posts: 774
Joined: Sat May 21, 2011 7:21 pm
Location: Between UK and France
Contact:

Re: Track reading (Sedoric / FDC)

Post by Godzil »

It is "not" reliable, true, but it your code that will make it reliable, you will need to read the same track multiple times, and using other techniques to make this "unreliable" command into a reliable one :)
christian
Pilot Officer
Posts: 96
Joined: Sun Nov 24, 2013 9:58 pm

Post by christian »

ID AM is "ID Address Mark", it is just before the beginning of a sector.

A track is composed by:
[GAP1][sector 1][sector 2]...[GAP5]

A sector is:
[GAP2][Index][GAP3][Data][GAP4]

Here is Track 0 / sector 1 / Side 0 (from a Sedoric disk image):

Code: Select all

Start of track
00000100  4e 4e 4e 4e 4e 4e 4e 4e  4e 4e 4e 4e 4e 4e 4e 4e  |NNNNNNNNNNNNNNNN|
00000110  4e 4e 4e 4e 4e 4e 4e 4e  4e 4e 4e 4e 4e 4e 4e 4e  |NNNNNNNNNNNNNNNN|
00000120  4e 4e 4e 4e 4e 4e 4e 4e  00 00 00 00 00 00 00 00  |NNNNNNNN........|
00000130  00 00 00 00 c2 c2 c2 fc  4e 4e 4e 4e 4e 4e 4e 4e  |........NNNNNNNN|
00000140  4e 4e 4e 4e 4e 4e 4e 4e  4e 4e 4e 4e 4e 4e 4e 4e  |NNNNNNNNNNNNNNNN|
00000150  4e 4e 4e 4e 4e 4e 4e 4e  4e 4e 4e 4e 4e 4e 4e 4e  |NNNNNNNNNNNNNNNN|

Start of sector
00000160  00 00 00 00 00 00 00 00  00 00 00 00 a1 a1 a1 fe  |................|
00000170  00 00 01 01 fa 0c 4e 4e  4e 4e 4e 4e 4e 4e 4e 4e  |......NNNNNNNNNN|
00000180  4e 4e 4e 4e 4e 4e 4e 4e  4e 4e 4e 4e 00 00 00 00  |NNNNNNNNNNNN....|
00000190  00 00 00 00 00 00 00 00  a1 a1 a1 fb 01 00 00 00  |................|

...Datas...

End of sector
00000290  20 20 20 20 20 20 20 20  20 20 20 20 be 12 4e 4e  |            ..NN|
000002a0  4e 4e 4e 4e 4e 4e 4e 4e  4e 4e 4e 4e 4e 4e 4e 4e  |NNNNNNNNNNNNNNNN|
000002b0  4e 4e 4e 4e 4e 4e 4e 4e  4e 4e 4e 4e 4e 4e 4e 4e  |NNNNNNNNNNNNNNNN|
000002c0  4e 4e 4e 4e 4e 4e                                 |NNNNNN..........|

GAP1:
$4e x 40
$00 x 12 => PLL lock-up
$c2 x 3 => SYNC
$fc => Data Mark
$4e x 40

GAP2:
$00 x 12 => PLL lock-up (at least 8)
$a1 x 3 => SYNC bytes (clear CRC register)

Index field:
$fe => Index Address Mark followed by 6 bytes:
$00 => Track 0
$00 => Side 0
$01 => Sector 1
$01 => Sector size (0:128 bytes / 1: 256 bytes / 2:512 / 3:1024)
$fa $0c => CRC

GAP3:
$4e x 22
$00 x 12 => PLL lock-up
$a1 x 3 => SYNC bytes

Datas field:
$fb => Data Address Mark
... x 256 => Datas
$be $12 => CRC

GAP4:
$4e x 40 (at least 24)

GAP5:
$4e x 218 (at least 16)
Symoon wrote:Are there any technical docs for FT-DOS, just like l'Oric à Nu for Sedoric?
Not yet, I'm writing one (FTDos à nu?)
User avatar
Dbug
Site Admin
Posts: 4437
Joined: Fri Jan 06, 2006 10:00 pm
Location: Oslo, Norway
Contact:

Re: Track reading (Sedoric / FDC)

Post by Dbug »

It's important to keep in mind that these numbers of bytes are not fixed values. It depends of the format of the disk, and in particular the number of sectors per track.

If you look in the source code of FloppyBuilder, you can find a bunch of commented sections with informations I got from various tools, webpages, articles and datasheets:

http://miniserve.getmyip.com/svn/public ... Floppy.cpp
Godzil
Squad Leader
Posts: 774
Joined: Sat May 21, 2011 7:21 pm
Location: Between UK and France
Contact:

Re: Track reading (Sedoric / FDC)

Post by Godzil »

That's surprising to use 0x4E as a sync byte

0x4E == 0b01001110 which does not seems a good sync byte.. Anyway, the Oric 0x16 is not "good" either :D (0b00010110)

Why they didn't use 0xAA (0b10101010) or 0x55 (0b01010101) or a mix of both for sync? It will be easier to sync with a regular toggling pattern..
User avatar
Symoon
Archivist
Posts: 2301
Joined: Sat Jan 14, 2006 12:44 am
Location: Paris, France

Re: Track reading (Sedoric / FDC)

Post by Symoon »

Godzil wrote:That's surprising to use 0x4E as a sync byte

0x4E == 0b01001110 which does not seems a good sync byte.. Anyway, the Oric 0x16 is not "good" either :D (0b00010110)

Why they didn't use 0xAA (0b10101010) or 0x55 (0b01010101) or a mix of both for sync? It will be easier to sync with a regular toggling pattern..
Well I'm not a specialist but I disagree ;)
The BIG problem I had for slow speed tapes decoding was that the bits were repeated (even the stop/start bits), so I had signals with 0101010101010101010101 at the beginning.
Well, with such information, you just don't know if you began reading a byte right in the middle, or at its beginning, or whatever.
But with 01101000 it's much easier to know "where" you are in the byte - so it's easier to synchronise ;)
Last edited by Symoon on Thu Jun 26, 2014 7:31 pm, edited 1 time in total.
User avatar
Symoon
Archivist
Posts: 2301
Joined: Sat Jan 14, 2006 12:44 am
Location: Paris, France

Re: Track reading (Sedoric / FDC)

Post by Symoon »

Thanks to all for information.
I wish I could try multiple readings to correct errors but it won't be easy:
- the unsynchronised bytes at the beginning and the end will probably differ at each reading
- the $C2 inserted here and there lead to a track size that changes, so comparing byte for byte won't be easy
- finding the right limit to say "ok, this byte is fine" will change with the hardware reliability: do I decide a value is correct after two, three, four similar reads? How can I be sure there were not two unexpected $C2 bytes at the same place?
-...
User avatar
Dbug
Site Admin
Posts: 4437
Joined: Fri Jan 06, 2006 10:00 pm
Location: Oslo, Norway
Contact:

Re: Track reading (Sedoric / FDC)

Post by Dbug »

Basically I guess what you need is to implement the equivalent of the 'diff' tools for your tracks:
- Have 2 or more track buffers
- Perform a complete read track in each of the buffers
- Perform the diff to find the sequences of identical bytes

Not trivial, but not impossible either :)

You can also probably do a combination of read sector and read track: The read sector will give you access to 256/512 know bytes, and you can identify these blocks (or part of them) in the track buffer using the same 'diff' method.

Also, normally the bytes in the gaps and crc are not pure random. You know more or less which values some of these bytes should have (sector number, crc can be computed).
User avatar
Symoon
Archivist
Posts: 2301
Joined: Sat Jan 14, 2006 12:44 am
Location: Paris, France

Re: Track reading (Sedoric / FDC)

Post by Symoon »

Dbug wrote:Basically I guess what you need is to implement the equivalent of the 'diff' tools for your tracks:
- Have 2 or more track buffers
- Perform a complete read track in each of the buffers
- Perform the diff to find the sequences of identical bytes

Not trivial, but not impossible either :)
Sure, but really not easy! Especially if many bytes are inserted, taking the room of end of the track in the reading buffer (even thought if they only replace gaps bytes, it will be easy to re-create... But that's exactly what I was NOT trying to do: re-create instead of reading)...
Well I mean, there will be loads of "possible dirty things and combinations" to think of and deal with.
Dbug wrote:You can also probably do a combination of read sector and read track
Aha, well I thought of that but if I begin reading the sectors then I lose the interest of the track-reading, which was to read without bothering too much with the disk format (not having to write a version for each DOS).

Stupid problem ain't it ? ;)

I'll try something this weekend: changing the machine itself. I'm surprised that I only got about 20 $C2 bytes with my old tests (done on Oric-1 IIRC), and the 140 I got with my latest tests (done with what looks like a heavily modified Atmos).
Godzil
Squad Leader
Posts: 774
Joined: Sat May 21, 2011 7:21 pm
Location: Between UK and France
Contact:

Re: Track reading (Sedoric / FDC)

Post by Godzil »

Symoon wrote: Well I'm not a specialist but I disagree ;)
The BIG problem I had for slow speed tapes decoding was that the bits were repeated (even the stop/start bits), so I had signals with 0101010101010101010101 at the beginning.
Well, with such information, you just don't know if you began reading a byte right in the middle, or at its beginning, or whatever.
But with 01101000 it's much easier to know "where" you are in the byte - so it's easier to synchronise ;)
Ho I don't mean to only use a 0101/1010 pattern for an header :) You are right it will be difficult to know where you are in such situation, but the original meaning of a sync patern is first to sync clocks to get the correct speed, after of course you need to have pattern to determine the start of the data flow, we also could name this a syncronisation.

In the case of the floppy, the sync byte (that are follow by a "PLL Lockup" phase are clearly to set a timebase for a clock, that's why I'm surprised, where on the TAP the sync bytes are more to know where is start, because the baudrate is almost fix (not completely true on floppy as the RPM could change a bit between drives and the physical sector size change if you are on the extern or internal circle..)
User avatar
Symoon
Archivist
Posts: 2301
Joined: Sat Jan 14, 2006 12:44 am
Location: Paris, France

Re: Track reading (Sedoric / FDC)

Post by Symoon »

Ok! Some new tests.
Changed the machine: no effect.
Changed the track read on the disk: incredible differences. According to the track I read (tried about 6 different ones), there are between 6 to 144 $C2 bytes. Of course this is probably changing according to the data but I used the disk that, on track 8, had 144 $C2 when read by a simple call to $CFCD routine, and 44 when read by Nibble.

Anyway, maybe this is a way to check the health of a track on disk? Read it under Sedoric using $CFCD, then count the amount of $C2 bytes...

Well that still doesn't explain the differences between Nibble and $CFCD though. I guess I'll have to dig further in Nibble's decompilation - all I hate because of the time it takes ;)
User avatar
Dbug
Site Admin
Posts: 4437
Joined: Fri Jan 06, 2006 10:00 pm
Location: Oslo, Norway
Contact:

Re: Track reading (Sedoric / FDC)

Post by Dbug »

Can that be caused by the geometry?

Tracks the closest to the center of the disk have a faster rotation speed than the ones at the periphery, so you can expect differences I guess.
In that regard, I always wondered why we had a fixed number of sectors per track, it would make sense to me to have more sectors on the outside tracks :)
Godzil
Squad Leader
Posts: 774
Joined: Sat May 21, 2011 7:21 pm
Location: Between UK and France
Contact:

Re: Track reading (Sedoric / FDC)

Post by Godzil »

I'ts just because IBM Floppy format have been badly designed :)
Post Reply