1050 RAM/Density Question

[+CharlieChaplin]

Maybe a stupid question, but what copyprogram is the best to make copies of copyprotected floppy when using a 1050 with the mini-speedy upgrade ?

 

There are several backup programs for the Speedy 1050 (mini-Speedy) available, all from Germany:

 

- Backup-Copy (on the Speedy system masterdisk), copies only copy-protected 90k disks

 

- Disk-Master (NOT! the disk/sector-editor, but a german program to create, analyze and duplicate copy protections for Happy + Speedy drives; AMC-Software used it for all their copy protected disks, especially for the copy-protected 130k disks; the author of this program also wrote a short series of articles about Happy/Speedy floppy enhancements and copy-protection in the german Atari Magazin; there are also several adverts for this program in the old Atari Magazin and in the new Atari Magazin)

 

- Ultra-Copier by Liberal Dreams Software (author Bernd Dongus ?), afair on side A a program to copy copy-protected 90k disks, on side B a program to copy copy-protected 130k disks

 

- Turbo 1050-Emulator, when executed on a Speedy 1050 this turns it into a Turbo 1050, boot without a diskette in the drive and when the menu appears press Option to load the Turbo 1050 solid-state utilities, there press Select to choose the Backup machine which will copy most copy-protected 90k disks

 

- MS-Copy to copy copy-protected 90k and 130k disks (Author M.Schubert, there exist two versions: one for the Speedy 1050 and one for the Speedy-XF), both versions can be found here: http://www.atarionline.pl/v01/index.php?ct=utils&sub=6.%20Stacja%20dyskietek&tg=MS-Copy&PHPSESSID=d1bf6855f71b1dd3549a62fc9b22d608#MS-Copy

(yes, the ATR and the XEX are different versions! one is the 1050 version, the other is the XF version)

 

- MS-Formatter will create special formats to defeat various copy-protections and can afaik also copy many copy-protected disks

 

(There is also a sector-copy program, named Backup Master and a program named Ultraspeed Backup, both are more or less standard sectorcopy programs and cannot copy any copy-protected disks.)

 

Which one is the best ? Hard to say, since I am no expert and do not own many copy-protected disks anymore (I never owned any originals from Synapse, Electronic Arts, etc.). Even most of my original diskettes are not copy-protected anymore but do contain a pirate version that simply looks like the original version (no hacker/cracker intro, etc.). The Backup-Copy program from the Speedy systemdisk is the easiest one to operate, press Start, wait until reading is done, then insert a new disk and press Start to write (you may have to switch disks several times). But it cannot copy many disks successfully. The Backup Machine from the Turbo 1050 should be better, you can simply press Start to copy or press Start+? (Start+Select? Start+Option?) to use "XALIGN" to copy a disk, so there you have two options. But both programs cannot copy any copy-protected 130k disks. Therefore you have to use Ultra-Copier or MS-Copy and these programs have several configurable options and I was always too stupid to configure or operate these programs and thus could never use them... so you have to decide yourself. Besides, the Disk-Master I mentioned above is hard to find nowadays...

 

All these Speedy programs do not create PDB files or anything similar, they (try to) create the original copy-protection on the backup diskette... (the manual for the Turbo 1050 also explained several copy-protections).

[Binarygeek]

Stupid question about the 1050 Drive.

 

 

... Questions are -- is the 256 byte limitation the primary reason? And if yes, is the sector count limited to 128 only and not something like 192 bytes for the same or another reason? (i.e. 26 sectors x 192 bytes/sector = not possible?)

It is not possible for a 192 byte sector, as the sector length is set during the format. If I remember correctly, the WD2793 floppy disk controller will format with either 128, 256, 512 or 1024 bytes per sector.

[+Nezgar]

How did you do that? I have dissembled the USD rom, 1st thing the rom does is to check for the ram module. No module = no run.

There is an DD upgrade that doesn't add extra ram. Rather unknown one from eastern Europe. Cannot remember name tho.

I'm also very curious if there exists such a ROM only upgrade to give USD ultraspeed, even if it is just for SD and ED. Maybe _The Doctor__ either had a modified USD ROM that disabled the ram check, or it was actually a 1050 Turbo ROM?

[_The Doctor__]

It was a us doubler rom... the drive ran fine..... I didn't know I couldn't do a thing so I did a thing... and it never gave me a problem.... but after reading all the consternation I should dig it up and add in ram pack just in case it would be a problem with something else........it will feel weird to disturb it after all these years but why not be safe than sorry..... some where I had posted about a pin removed on some chip in one of the XF's as a compatibility mod of some sort I wonder if that's the same drive...

[+Nezgar]

Curiosity got the better of me... I cracked open my USD drive this evening, removed the double-stack of 6810's and replaced it with a single 6810 which was tested working in a stock drive.

 

Drive light turns on, but dead otherwise. Put original ram stack back in, away it goes.

 

My U10 is a just a 2732 EPROM from way back when labelled "USD2" so not sure if its different from an authentic original.

[+Nezgar]

It was a us doubler rom... the drive ran fine..... I didn't know I couldn't do a thing so I did a thing... and it never gave me a problem.... but after reading all the consternation I should dig it up and add in ram pack just in case it would be a problem with something else........it will feel weird to disturb it after all these years but why not be safe than sorry..... some where I had posted about a pin removed on some chip in one of the XF's as a compatibility mod of some sort I wonder if that's the same drive...

 

I'm still very curious about your USDoubler that you say is running without any additional RAM upgrade. If you ever get around to digging it up, would be cool to get some photos of the board.

[Jeffrey Worley]

On 1/1/2017 at 8:29 PM, CharlieChaplin said:

 

There are several backup programs for the Speedy 1050 (mini-Speedy) available, all from Germany:

 

- Backup-Copy (on the Speedy system masterdisk), copies only copy-protected 90k disks

 

- Disk-Master (NOT! the disk/sector-editor, but a german program to create, analyze and duplicate copy protections for Happy + Speedy drives; AMC-Software used it for all their copy protected disks, especially for the copy-protected 130k disks; the author of this program also wrote a short series of articles about Happy/Speedy floppy enhancements and copy-protection in the german Atari Magazin; there are also several adverts for this program in the old Atari Magazin and in the new Atari Magazin)

 

- Ultra-Copier by Liberal Dreams Software (author Bernd Dongus ?), afair on side A a program to copy copy-protected 90k disks, on side B a program to copy copy-protected 130k disks

 

- Turbo 1050-Emulator, when executed on a Speedy 1050 this turns it into a Turbo 1050, boot without a diskette in the drive and when the menu appears press Option to load the Turbo 1050 solid-state utilities, there press Select to choose the Backup machine which will copy most copy-protected 90k disks

 

- MS-Copy to copy copy-protected 90k and 130k disks (Author M.Schubert, there exist two versions: one for the Speedy 1050 and one for the Speedy-XF), both versions can be found here: http://www.atarionline.pl/v01/index.php?ct=utils&sub=6.%20Stacja%20dyskietek&tg=MS-Copy&PHPSESSID=d1bf6855f71b1dd3549a62fc9b22d608#MS-Copy

(yes, the ATR and the XEX are different versions! one is the 1050 version, the other is the XF version)

 

- MS-Formatter will create special formats to defeat various copy-protections and can afaik also copy many copy-protected disks

 

(There is also a sector-copy program, named Backup Master and a program named Ultraspeed Backup, both are more or less standard sectorcopy programs and cannot copy any copy-protected disks.)

 

Which one is the best ? Hard to say, since I am no expert and do not own many copy-protected disks anymore (I never owned any originals from Synapse, Electronic Arts, etc.). Even most of my original diskettes are not copy-protected anymore but do contain a pirate version that simply looks like the original version (no hacker/cracker intro, etc.). The Backup-Copy program from the Speedy systemdisk is the easiest one to operate, press Start, wait until reading is done, then insert a new disk and press Start to write (you may have to switch disks several times). But it cannot copy many disks successfully. The Backup Machine from the Turbo 1050 should be better, you can simply press Start to copy or press Start+? (Start+Select? Start+Option?) to use "XALIGN" to copy a disk, so there you have two options. But both programs cannot copy any copy-protected 130k disks. Therefore you have to use Ultra-Copier or MS-Copy and these programs have several configurable options and I was always too stupid to configure or operate these programs and thus could never use them... so you have to decide yourself. Besides, the Disk-Master I mentioned above is hard to find nowadays...

 

All these Speedy programs do not create PDB files or anything similar, they (try to) create the original copy-protection on the backup diskette... (the manual for the Turbo 1050 also explained several copy-protections).

There's also the old method of putting a strip of tape on the floppy jacket leading outside.  Set the program, or your own BASIC program, to write a specific sector and read it back.  Crank it up and tug on the tape to slow the rotation down by adding drag.  The drive will eventually write a bad sector.  This won't foil really sophisticated CP schemes, like fuzzy sectors, long sectors, short sectors, duplicate sectors, or phantom sectors (I get em' all?), but it does satisfy the majority of CP schemes that require an error at a specific sector number.

 

Best,

 

Jeff

[+Nezgar]

On 6/9/2017 at 12:06 AM, _The Doctor__ said:

It was a us doubler rom... the drive ran fine..... I didn't know I couldn't do a thing so I did a thing... and it never gave me a problem.... but after reading all the consternation I should dig it up and add in ram pack just in case it would be a problem with something else........it will feel weird to disturb it after all these years but why not be safe than sorry.....

Hey @_The Doctor__... I'd still really appreciate if you could dig up this drive and confirm if it has a ROM that functions in single or enhanced density only with only 1 6810 RAM chip? (No double density). It would not require any "disturbing" just to test.

 

2 hours ago, Technoid Mutant said:

This won't foil really sophisticated CP schemes, like fuzzy sectors, long sectors, short sectors, duplicate sectors, or phantom sectors (I get em' all?),

Some other schemes that couldn't be defeated by the early "disk pull" method included duplicate/phantom sectors (reading the same sector multiple times would yield different results), deleted sectors, missing sectors, sectors with good data but incorrect CRC and a few more... or just a different sector order/interleave that can exhibit timing differences.

 

[Rybags]

I'm not so sure on the simple bad sector check being very common in later times, like after 1982 or 83.

Though with some games it's surprising how easily defeated the protection is.

 

Great American Cross Country Road Race is one that comes to mind.  Just for the sake of it after investigating the original I found that the bad sectors resided on a certain track and that no data was present after that track.

So I just formatted a never used disk in my 1050 and flipped the lever to stop the operation when it got to the track before the bad sector.

Then copied the game disk from start to the track before the bad sectors, and it worked just fine.

 

But later times, I would suspect that the likes of duplicate ID and fuzzy sectors became more popular.  Duplicate ID sector especially hard to defeat given that simply reading the disk could in theory give a clean run with no evidence of anything non-standard.

Edited September 25, 2019 by Rybags

[+CharlieChaplin]

On 1/8/2017 at 3:35 AM, Binarygeek said:

If I remember correctly, the WD2793 floppy disk controller will format with either 128, 256, 512 or 1024 bytes per sector.

 

So, it would be possible to a) write firmware for the drive and b) e.g. a DOS that formats 1024 bytes per sector (not 8 sectors with 128 bytes each that are treated as 1 block like in DOS 3, but really every sector with 1024 bytes)...?!? And errmm, if so, could we still use 720 sectors per diskside - resulting in 720k (720 x 1k) per diskside ?

 

[BillC]

3 minutes ago, CharlieChaplin said:

 

So, it would be possible to a) write firmware for the drive and b) e.g. a DOS that formats 1024 bytes per sector (not 8 sectors with 128 bytes each that are treated as 1 block like in DOS 3, but really every sector with 1024 bytes)...?!? And errmm, if so, could we still use 720 sectors per diskside - resulting in 720k (720 x 1k) per diskside ?

 

HD mechanisms have 80 tracks/side, which requires smaller stepping. I believe the highest density 5-1/4" mechanism is only 1.2MB(600kB/side) while 1.44MB is most common for the 3-1/2".

[+CharlieChaplin]

Okay,

but considering we have 40 tracks and we could use 1024bytes (1k) per sector on the 1050 drive (with some enhancement) - what would be the maximum number of sectors and thus the max. number of kbytes we could use per diskside ?!? Would it still be 180kb per diskside then or more ?!?

 

Edited September 25, 2019 by CharlieChaplin

[+kheller2]

Just because you can write a bigger sector, doesn't mean you have the physical room for the same amount of sectors with 10x more data.

[Jeffrey Worley]

3 hours ago, CharlieChaplin said:

Okay,

but considering we have 40 tracks and we could use 1024bytes (1k) per sector on the 1050 drive (with some enhancement) - what would be the maximum number of sectors and thus the max. number of kbytes we could use per diskside ?!? Would it still be 180kb per diskside then or more ?!?

 

I'm not sure a 1k sector size would be a good fit for the 1050, though 512bytes would be good.  The reason is that DD is 4.5kbytes per track.  Say you put 5 1k sectors per track, you'd get a bump in storage of, um, 15%-ish?  There would be less overhead for sector information, so that might actually make room for the extra 512bytes per track...  Assuming you could do it within the constraints of MFM, you would get more storage both actual and logical.  The downside of 1k sectors is that there would be a minimum allocation of 1k per file, and that would mean that your whole disk could store no more than like 100 files, no matter how small they were, say 100 files of 2 bytes each, would eat the disk for lunch.  There's something to be said for a small allocation unit.

 

Jeff

[phaeron]

No matter the sector count or sector size, a fundamental limit is the number of raw data bits that can be fit on the track. The math is as follows, for MFM:

 

ideal rotation time = 1000000us / (288 RPM / 60) = 208333us

raw bit cell time = 2us

raw data bit time = 2 * raw bit cell time = 4us

raw data byte time = 8 * raw data bit time = 32us

max ideal raw data bytes = ideal rotation time / raw data byte time = 208333us / 64us = 6510 bytes

 

...and out of that budget comes some bytes for address fields, sync marks, required gaps, and a bit of slack for motor speed variation. Generally the per-sector overhead is around 80-100 bytes per sector, so if you exclusively used 1K sectors, you could only reliably fit five sectors unless you try to cheat on the gaps and use a drive with precise speed and clocking to write the track. The sectors don't have to all be the same size, however, it is possible to mix 5 x 1024b + 1 x 512b. This also doesn't have to be exposed to the computer, it can be translated by the drive like modern 512e hard disks.

 

[Jeffrey Worley]

20 hours ago, CharlieChaplin said:

 

So, it would be possible to a) write firmware for the drive and b) e.g. a DOS that formats 1024 bytes per sector (not 8 sectors with 128 bytes each that are treated as 1 block like in DOS 3, but really every sector with 1024 bytes)...?!? And errmm, if so, could we still use 720 sectors per diskside - resulting in 720k (720 x 1k) per diskside ?

 

I gather

 

On 6/9/2017 at 4:38 AM, Nezgar said:

Curiosity got the better of me... I cracked open my USD drive this evening, removed the double-stack of 6810's and replaced it with a single 6810 which was tested working in a stock drive.

 

Drive light turns on, but dead otherwise. Put original ram stack back in, away it goes.

 

My U10 is a just a 2732 EPROM from way back when labelled "USD2" so not sure if its different from an authentic original.

Oh cool.  Thanks for doing that, because it was bugging me and I reassembled the drive with the RF shield, so I would have eventually just HAD to pull it all apart to do this same experiment.  you saved me a lot of effort to get this information!

 

Jeff

[Jeffrey Worley]

20 hours ago, CharlieChaplin said:

 

So, it would be possible to a) write firmware for the drive and b) e.g. a DOS that formats 1024 bytes per sector (not 8 sectors with 128 bytes each that are treated as 1 block like in DOS 3, but really every sector with 1024 bytes)...?!? And errmm, if so, could we still use 720 sectors per diskside - resulting in 720k (720 x 1k) per diskside ?

 

The raw format of the disk is only 360k-ish?  The MFM format takes about 50% of that, which accounts for the 180k of storage available after format.  So unless you use some other format, which would involve modifying the drive to a considerable extent, up to and including replacing the mechanism, you can't expect to increase its capacity by much.  There are some exploits that leverage the fact that some mechs step to 41 or 43 tracks, so you can use that in some circumstances - possibly a rom modification to do that, but there will be no order-of-magnitude density upgrades to the 1050.  By the time you were done with something like that, it wouldn't be an Atari drive anymore.  Might as well use the shell and stick a hard disk drive in there.

 

Best,

 

Jeff

[+Nezgar]

On 9/26/2019 at 2:36 AM, phaeron said:

max ideal raw data bytes = ideal rotation time / raw data byte time = 208333us / 64us = 6510 bytes

Ah this is the key info... SO based on this, a 1024 byte sectors+100 bytes of overhead each would require 1124 bytes per sector, so 5 sectors would fit would consume 5620 bytes.

 

Just using a consistent 6 1K sectors per track would yield 200KB instead of 180KB... (5K per track x 40 tracks). Adding 1 512byte sector to each track as Phaeron mentioned would add another 20KB (220KB total per side).

 

Take a look at the "XDF" format used on supported by IBM PC DOS 7, and OS/2 Warp 3 and up for almost 1.9MB on a "standard 1.44MB" HD diskette by using exactly this trick. Each track (except track 0) used 1 8KB sector, 1 2KB sector, 1 1KB sector, and 1 512 byte sector - for a total of 11.5KB usable per track per side, so 23KB per track:

https://www.os2world.com/forum/index.php?topic=2004.0

 

It would definitely be interesting to see this format attempted with custom firmware for existing track-buffered speeders like Happy or Speedy to do all the translation work and just present virtual 256 byte sectors to the computer for maximum compatibility. Writes would have to be buffered, and then 'mixed' in a 2 pass read-write operation on the drive.

 

The resulting "Atari XDF" disks would have 220KB of usable space, and appear to have 880 double-density sectors per side. They physical format would have the same raw MFM bit density and be no less reliable as any other double-density formatted disk. The resulting disks would obviously only be usable in such a modified drive though.

Edited September 27, 2019 by Nezgar
math error

[Jeffrey Worley]

33 minutes ago, Nezgar said:

Ah this is the key info... SO based on this, a 1024 byte sectors+100 bytes of overhead each would require 1124 bytes per sector, so 5 sectors would fit would consume 5620 bytes.

 

Just using a consistent 6 1K sectors per track would yield 200KB instead of 180KB... (5K per track x 40 tracks). Adding 1 512byte sector to each track as Phaeron mentioned would add another 20KB (220KB total per side).

 

Take a look at the "XDF" format used on supported by IBM PC DOS 7, and OS/2 Warp 3 and up for almost 1.9MB on a "standard 1.44MB" HD diskette by using exactly this trick. Each track (except track 0) used 1 8KB sector, 1 2KB sector, 1 1KB sector, and 1 512 byte sector - for a total of 11.5KB usable per track per side, so 23KB per track:

https://www.os2world.com/forum/index.php?topic=2004.0

 

It would definitely be interesting to see this format attempted with custom firmware for existing track-buffered speeders like Happy or Speedy to do all the translation work and just present virtual 256 byte sectors to the computer for maximum compatibility. Writes would have to be buffered, and then 'mixed' in a 2 pass read-write operation on the drive.

 

The resulting "Atari XDF" disks would have 220KB of usable space, and appear to have 880 double-density sectors per side. They physical format would have the same raw MFM bit density and be no less reliable as any other double-density formatted disk. The resulting disks would obviously only be usable in such a modified drive though.

At the higher level, you would of course have overhead for the 880 virtual sectors, sector number, link number, crc, etc.  That would pinch off a few bytes per sector.

[+CharlieChaplin]

So,

 

one more stupid question: Would it be possible to use 1440 sectors per diskside (MFM) on a 40 track 5,25" DSDD disk with 128 bytes per sector to reach a total of 180kbytes per diskside - or is the overhead too large ?!? Would a 1050 drive (or XF551) be able to write 36 sectors per track with 128 bytes per sector or was/is 26 sectors per track the maximum it can do ? (Think I read somewhere, that the original XF551 OS has a bug and formats with more than 26 sectors per track in medium/enhanced density...)

 

[+Nezgar]

In Dual/Enhanced density (MFM mode), each sector would be 128 bytes, plus 100 bytes of overhead, for about 228 bytes of raw bytes. 26x228=5928 raw track bytes of the maximum 6510 bytes.

 

Based on those calculations you might be able to squeeze 28 full sectors on a track, which would use 6384 of the 6510 bytes, but you risk a drive with a slow RPM writing over the beginning of the track.

 

28 sectors x 40 tracks = 1120 sectors, or 140K per disk. It would still need modified firmware to use...

 

However, if the physical format uses the above "Atari XDF" by using 5x1024 byte sectors + 1 512 byte sector per track, each track could fit 5.5KB of user data, and custom drive firmware could emulate 44 "virtual" 128 byte sectors per track, which would be 1760 virtual sectors per side.

[Rybags]

 I wonder about the diag command that allows the (sort of) half-track stepping - could we do some sort of "shingled mode recording" like modern HDDs?

 

The principle there being that write heads have somewhat larger coverage than read heads but you can write the data "shingled" where the previous track get partially overwritten but subsequent read with it's smaller head can segregate just the track you're wanting.

 

I guess though that the 1050 would likely just read both "tracks" at once and get confused.

[Jeffrey Worley]

1 minute ago, Rybags said:

 I wonder about the diag command that allows the (sort of) half-track stepping - could we do some sort of "shingled mode recording" like modern HDDs?

 

The principle there being that write heads have somewhat larger coverage than read heads but you can write the data "shingled" where the previous track get partially overwritten but subsequent read with it's smaller head can segregate just the track you're wanting.

 

I guess though that the 1050 would likely just read both "tracks" at once and get confused.

These modes are used in the newest hdd's.  The 1050, and nearly all floppy mechs, use stepper motor actuators.  These motors move in jumps and are mechanically limited to those jumps, specified in degrees.  I don't see how one can reliably massage that limitation.  Many steppers step an extra track or three at one or both extremes of motion.  This can and has been used to gain some extra space.  Does the tandon mech stepmore than 40 tracks?

 

[+Nezgar]

53 minutes ago, Technoid Mutant said:

Does the tandon mech stepmore than 40 tracks?

Yes, 1050 firmware steps the stepper twice for each track step. Possible not equal sized steps. Diag mode commands can be used to do half track steps with stock 1050 firmware. But as @Rybags mentioned, it's highly likely to clobber neighboring tracks.

 

There was a thread that discussed this in detail in the past, I'll see if I can find it.

Edit: here it is:  (as investigated by @Rybags I see )

[+Nezgar]

On 9/26/2019 at 9:15 AM, Technoid Mutant said:

Oh cool.  Thanks for doing that, because it was bugging me and I reassembled the drive with the RF shield, so I would have eventually just HAD to pull it all apart to do this same experiment.  you saved me a lot of effort to get this information! 

I wonder if the US Doubler's check for the extra RAM at poweron was NOP'd out, if it would function as a single+enhanced density ultraspeed drive with a single 6810... Of course that would be a dangerous state, because it would appear to access double density disks, but DD sectors would read and written with half the sector blank...

 

Actually, I just remembered about a hacked USDoubler ROM that candle did exactly this, to make it work without the shadowed memory locations when he was originally developing the 1050E..

 

 

For fun I tried loading this ROM into Alitrra with full drive emulation in stock 1050 mode. Interestingly the 1050 does do POST head seek, and sometimes will read sectors, but the drive dies soon after, so there must be other occurances using the 'high' 128 bytes of RAM, or shadowed repeats...