Why exactly did the 1050 end up with "enhanced density"?

[zzip]

I don't think I ever heard the definitive story of how the 1050 ended up being 127K despite being advertised as "double density".   I imagine they were using DD mechanisms?  Was it some kind of bug?  Firmware restriction?  Manufacturing defect?

[+Stephen]

Cheapness.  It would have needed 256 bytes of RAM at least, to hold the sector.  The only RAM in the drive was 128 bytes contained in the RIOT.

[zzip]

Just now, Stephen said:

Cheapness.  It would have needed 256 bytes of RAM at least, to hold the sector.  The only RAM in the drive was 128 bytes contained in the RIOT.

How much would an extra 128 bytes have cost back then?

[+David_P]

There is an upgrade that makes the 1050 DD without adding RAM - the Turbo 1050. More info at: https://atariage.com/forums/topic/277204-some-new-not-really-hardware-made-by-tfhh/

 

[+Stephen]

7 hours ago, zzip said:

How much would an extra 128 bytes have cost back then?

I have no idea - the RIOT served other purposes so was essentialy "free RAM".  I can't imagine when the drive came out that a measly 128 bytes would have added much.

[+Nezgar]

12 hours ago, zzip said:

I don't think I ever heard the definitive story of how the 1050 ended up being 127K despite being advertised as "double density".

"Double Density" because it used MFM encoding, instead of FM encoding. It can put double the bits in the same physical space, but with sectors at 128 bytes each instead of 256, all that inter-sector overhead yielded the lackluster increase in overall storage. Atari marketed the drive as "Dual Density" - suggesting it could work it two density modes, but not directly naming the 2nd density as "Double."  Well - the FSM calls it double, but everyone else called it Enhanced or Medium Density, since it wasn't "True" double density with 256 bytes per sector...

 

I think Atari may have began marketing it as double density, but it was changed very early on...

 

12 hours ago, Stephen said:

The only RAM in the drive was 128 bytes contained in the RIOT.

The 1050 (And the 810 too for that matter) actually had 256 bytes of RAM total: 128 bytes in the 6532 used for the sector transfer buffer, and another 128 in a 6810 SRAM chip used for page 0 and the stack. The us doubler mod added another 6810 (384 bytes total now) to buffer full 256 byte sectors without convoluted programming tricks, like the 1050 turbo as @David_P mentioned above.

 

 

 

[zzip]

17 hours ago, Stephen said:

I have no idea - the RIOT served other purposes so was essentialy "free RAM".  I can't imagine when the drive came out that a measly 128 bytes would have added much.

or if it did at launch, release a 1051 a year or two later with true DD..    Instead it took them years to release a proper drive.

[FifthPlayer]

I remember buying the Happy upgrade for my 1050 that gave it true double density.  The Happy was just as significant an upgrade as the 1050 was.

 

[+DrVenkman]

The (not very charitable but likely still accurate) reason was that when the 1050 was conceived, Atari had hundreds of thousands of 6507 and 6532 chips on hand or on order with reliable suppliers for use in 2600 systems, and to a lesser extent 850 interface modules and 810 drives. Cobbling up a new format scheme that allowed greater than standard single density at essentially the same cost in materials made the execs in Warner-era Atari, Inc. pay attention. Too bad they didn’t realize how cheaply and inexpensively they could have made a true double-density drive instead, alas.

 

All this aside, I wonder: have any our our Atari historians (especially @Savetz) every figured out how at Atari came up with the “Enhanced Density” idea for the 1050 and interviewed that person? Using second 6810 chip, or a new board design using even a single DRAM chip for 2K of combined stack and sector space, could not possibly have cost THAT much more than the final design, even back in 1983 when the thing was coming to market.

[Rybags]

Maybe throw in some of the usual paranoia as well.  C= only used 35 tracks on their drives though I think the reason was the mechs as much as the media.  Though by 1983 floppies were pretty good.

The "have to use excess obsolete stock" reason sounds really feasible... it's still widespread today.  I've seen desktops from the brand names that are just built using excess laptop components including charger then housed in ridiculously oversized cases.  The only reasoning I can see is they just wanted to get rid of stuff.

[+CharlieChaplin]

Hehe,

 

maybe they tried to use the double amount of sectors per track (36 instead of 18) and that did not work ? So they tested like this: "36 does not work, 34 does not work, 32 does not work, 30 does not work, 28 does not work, 26 works! We have found a new standard for double density!"

 

Remember, the A8 OS allows only single density boot sectors for SIO devices and maybe that was the reason they wanted to stay with 128 bytes per sector...?!? Was 180k a Percom standard - and did Percom ask for a licensing fee for using it ? (Like Microsoft wants a licensing fee for some of their formats, like NTFS, etc.)

 

But maybe they simply did not wanted to be compatible to any 3rd party standards, 288 RPM (instead of 300 RPM) was a good start, 130k dual/medium/enhanced density was another big point in not being compatible to any other standards. (If you want to be compatible, buy an Atari floppy drive.) And maybe Atari could ask for licensing fees from anyone that wanted to be compatible with 130k format ?!? 

 

[thorfdbg]

13 hours ago, CharlieChaplin said:

maybe they tried to use the double amount of sectors per track (36 instead of 18) and that did not work ? So they tested like this: "36 does not work, 34 does not work, 32 does not work, 30 does not work, 28 does not work, 26 works! We have found a new standard for double density!"

Frankly, I doubt this. The format, and that is the sector size and the number of sectors per track, is up to the floppy controller chip, and its specification were "out of reach" for Atari to change.

 

The fun part is that the floppy controller actually does support 256 byte sectors, but Atari went cheap - without additional RAM on the drive, they could not use them.

[ijor]

I guess the main reason was, cheapness indeed. A big question is this was an internal Atari design or if it was designed by Tandon.

 

5 hours ago, thorfdbg said:

Frankly, I doubt this. The format, and that is the sector size and the number of sectors per track, is up to the floppy controller chip, and its specification were "out of reach" for Atari to change.

 

The FDC doesn't impose any strict limit on the number of sectors per track. Some sector numbers in the $F5-$FF range are "problematic" (you can read and write them, but you can't format them) but otherwise you could have up to 256 sectors per track, if you insist. Of course, you need enough space on the track to fit all those sectors. The FDC also sets the minimum overhead per sector. But still you have some degree of freedom and as we already know, i.e., it is perfectly possible to fit 19 sectors per track on single density and surely more than 26 on enhanced density. It might be considered less reliable, though.

Edited September 8, 2020 by ijor

[drac030]

@ijor What if they clocked FDC faster or slowed down the rotation speed (yet more)? Are there any bad side effects of slow rotation speed, like too great bit density, which would affect the reliability even if the track size theoretically was big enough to fit more sectors with correct overhead areas?

[ijor]

1 hour ago, drac030 said:

What if they clocked FDC faster or slowed down the rotation speed (yet more)? Are there any bad side effects of slow rotation speed, like too great bit density, ...

 

Yes, that would increase the actual flux magnetic linear density and it would certainly affect reliability. Of course, a very small increase in the density probably wouldn't be so bad, but then you won't gain too much either.

 

There is also an issue of backwards compatibility. Altering the angular recorded density (FDC clock and RPM) would make less reliable to access old disks recorded with the original 810 bitrate and also the other way around, accessing "new" disks with an old drive. Again, the bigger the difference between the original and the new density, the less reliable the backwards compatibility would be.

 

Finally, there is an issue of breaking copy protections. As we know from the XF-551, some protections break just as a consequence of the disk rotating at a faster RPM altering the whole timing; even when the effective magnetic density is the same.

 

It is just a guess, but it is possible that maximum backwards compatibility with the 810 was very important for Atari. I understand that one of the things that killed the 815 project, besides perhaps the cost, was that it was completely incompatible with older disks.

 

Edited September 8, 2020 by ijor

[zzip]

21 hours ago, CharlieChaplin said:

But maybe they simply did not wanted to be compatible to any 3rd party standards, 288 RPM (instead of 300 RPM) was a good start, 130k dual/medium/enhanced density was another big point in not being compatible to any other standards. (If you want to be compatible, buy an Atari floppy drive.) And maybe Atari could ask for licensing fees from anyone that wanted to be compatible with 130k format ?!? 

 

If they just wanted to be proprietary, I would think they'd have found a way to be closer to 180K.   Having a DD drive that only does 127K looks bad on their spec sheet.

[+CharlieChaplin]

12 minutes ago, zzip said:

If they just wanted to be proprietary, I would think they'd have found a way to be closer to 180K.   Having a DD drive that only does 127K looks bad on their spec sheet.

 

Hehe, yes - but read the adverts what the drive in a 1450XLD can do: 254k ! A DS/DD (360k) drive that only does 254k (since Atari said that dual/medium/enhanced density is 127k)... looks even more bad on their spec sheet.

 

1450XLD catalog 1

1450XLD catalog 2

1450XLD catalog 3

 

[zzip]

4 minutes ago, CharlieChaplin said:

 

Hehe, yes - but read the adverts what the drive in a 1450XLD can do: 254k ! A DS/DD (360k) drive that only does 254k (since Atari said that dual/medium/enhanced density is 127k)... looks even more bad on their spec. sheet.

 

1450XLD catalog 1

1450XLD catalog 2

1450XLD catalog 3

 

Ouch!   But I suppose in 1983, the average consumer barely knew what a floppy drive was let alone know that DS/DD should handle 360K  not 254K   :)

[1050]

I'll always vote for the incompatibility reason. SD in FM and 26 sector per track MFM were "lifted" from 8 inch discs. Put 8 inch topology on a 5 1/4" and you have instant incompatibility with all other players. This is too perfect of a fit to ignore. 

[drac030]

21 hours ago, ijor said:

a very small increase in the density probably wouldn't be so bad, but then you won't gain too much either

If I calculate correctly, slowing the rotation down to 288 rpm gains 250 bytes per track in MFM. But yes, there is probably a good reason why they (or anyone?) did not use floppies with even less rotations per minute.

 

21 hours ago, ijor said:

Altering the angular recorded density (FDC clock and RPM) would make less reliable to access old disks recorded with the original 810 bitrate and also the other way around, accessing "new" disks with an old drive

This could probably be solvable by slowing the rotation down only in the new density (MFM). I doubt though if such a drive could be cheap.

 

Anyways, the 1450XLD catalogues (attached above) speak of the drives being "connected directly to the computer's processor bus" and therefore being much faster than an ordinary 1050. Other sources (like this one http://www.atarimuseum.com/computers/8BITS/XL/1450xld/1450xld.html ) say something about "100K per second data transfers" which I presume must be 100 kilobits per second considering that MFM is 250 kilobits/s, i.e. 30 KB/s. On the other hand, 100 kbps is not very much for a parallel data transfer considering that Pokey divisor 0 is ~127 kbps.

 

So, has anyone tested how these drives work and is able to quote some precise figures?

[ijor]

3 hours ago, drac030 said:

If I calculate correctly, slowing the rotation down to 288 rpm gains 250 bytes per track in MFM. But yes, there is probably a good reason why they (or anyone?) did not use floppies with even less rotations per minute.

At 288 you are already using a density 4% out of the nominal value. You would need to reach close to 10% higher density to gain just one extra sector. And MFM is more error prone than FM. The same RPM and density that might more or less work at single density might be much less reliable at MFM.

 

3 hours ago, drac030 said:

This could probably be solvable by slowing the rotation down only in the new density (MFM). I doubt though if such a drive could be cheap.

Yes, it could be an acceptable solution to that particular problem. And actually it might be affordable. Several enhanced Atari drives have two, or even more, pots for changing the RPM speed. This was used to copy some copy protections with more than 19 sectors. But then how much cheaper, if at all, that would be than adding 128 bytes of RAM for using "true" double density.

[TGB1718]

Maybe to maintain backwards compatibility the 1050 should have been double sided, it could still

read/write both single sided and double sided disks, wouldn't that have been nice back then

 

And even better when USD and Happy got their hands on it.

[Rybags]

True double sided doesn't help with "flippies" since turning the disk over means it's written in the opposite direction - though it was never a recommended by OEMs method to get more data on a disk.

[+CharlieChaplin]

On 9/23/2020 at 7:57 PM, TGB1718 said:

Maybe to maintain backwards compatibility the 1050 should have been double sided, it could still

read/write both single sided and double sided disks, wouldn't that have been nice back then

 

And even better when USD and Happy got their hands on it.

 

Think double sided was not a necessity, since everyone used flippies back then (not only on the A8). The 1050 simply should have been true DD / 180k right from the start. Maybe even the 810 should have been DD / 180k right from the start, because then no-one would ever have had the idea to make the 1050 do only 130k...

 

Next topic: Why did the 810 end up with "single density" ?     Hmmm, if we can trust Wikipedia, then 5,25" DS/SD disks with 180k (90k per diskside) were available since 1977 and 5,25" DS/DD disks with 360k (180k per diskside) since 1978. Guess the DD format was too new (and too expensive) when the 810 was designed on paper...?!?