Options for getting 3 1/2 inch drives, clock carts, other stuff.

[re-atari]

Who is making the SD card interfaces? I saw the website but nothing about buying one.

You might want to contact A8-user Mega-Hz at www.abbuc.de (German A8 enthousiasts club). He has designed and built a number of hardware extensions, including SIO2SD. Real quality stuff. A built and tested rev. 2 model will set you back around € 60.

 

re-atari

[re-atari]

That's a nice conversion on the top right. I think I might try this with my busted XF551. I was going to replace the broken mech with another 5.25 mech but I had trouble finding one which was suitable. Are 720K 3.5 drives easier to find? What did the ST use ? I have a broken ST kicking around which has become a donor machine. Can I use the drive from there ?

Never had an ST myself, but IIRC the early types came with a 360K drive that had a 80-track single-sided mechanism. That model will basically be compatible with nothing else but itself. Good chance the modded XF-551 firmware will not work with such a drive. Caveat emptor.

 

re-atari

[pepax]

Who is making the SD card interfaces? I saw the website but nothing about buying one.

You might want to contact A8-user Mega-Hz at www.abbuc.de (German A8 enthousiasts club). He has designed and built a number of hardware extensions, including SIO2SD. Real quality stuff. A built and tested rev. 2 model will set you back around € 60.

 

re-atari

I won't be surprised if he starts making either bare PCBs or even whole SDrives soon.

[+bf2k+]

Actually yes. I have a stock X551 somewhere... I'll try it... thanks!

 

Sounds broke to me... Do you have another XF551 that you can swap parts with? Put the original ROM and drive in? You may just have bad SIO connectors.

 

Bob

 

 

The 3.5 should read and write as fast as the old 5 1/4, at least. What upgrade did you use?

 

Bob

 

I didn't do it - don't know who did. I bought it on eBay I think. It has U6 socketed and a 28C64E prom in there - but no other markings.

 

[oracle_jedi]

It is pretty much a matter of changing the ROM and connecting a 3.5 drive. I hooked mine up externally so I can switch select either 3.5 or 5.25 as the boot drive. You should use a 720K drive, although a 1.44 works if you are careful. (don't mount any 1.44 disks)

 

Bob, where can I find this replacement ROM?

 

Thanks

[spookt]

It is pretty much a matter of changing the ROM and connecting a 3.5 drive. I hooked mine up externally so I can switch select either 3.5 or 5.25 as the boot drive. You should use a 720K drive, although a 1.44 works if you are careful. (don't mount any 1.44 disks)

 

Bob, where can I find this replacement ROM?

 

Thanks

 

Not sure if this is the right one:

 

XF551.zip

 

I've not tried it (yet).

[+Larry]

There are several options for using a real 3-1/2" drive. All involve using an XF551. None of the other options are still being sold -- maybe once in a bue moon on eBay. I would argue that the only good reason to get an XF551 is to convert it to a 3-1/2" drive. And as Bob1200XL said, 720K diskettes have been far more reliable than 5-1/4" disks in Double Density.

 

Bob's 3-1/2" conversion for the XF551 is widely available on the internet. There was an update published several years ago. This was my first mod for an XF551, and it works very well. It runs essentially at 2X SIO (same as an XF551 in DS/DD), assuming that you have the appropriate high speed driver and have a diskette formatted with the proper skew/interleve. The worst combination is to have have a fast interleve disk with a standard SIO driver -- this will be slower than stock. (And incidently, DD always *sounds* slower, but twice as many bytes are being transferred per sector.)

 

Another option is Stephen Dorndorf's HyperXF which allows UltraSpeed. This mod had some really interesting features. I have one of these and it is a very nice drive.

 

Still another option is to get a CSS 3-1/2 upgrade. Last I checked these kits were still available, but now? The CSS upgrade can be either one drive or with their dual drive kit, you can slave a second drive. For many years, I used a dual drive kit with a 5-1/4 as the master drive and the 3-1/2 as the "slave." True UltraSpeed SIO.

 

But aside from collecting, IMO, APE makes all these obsolete, plus it is significantly faster -- especially with APE USB interface.

 

-Larry

[+bob1200xl]

This XF551.zip file is the original hack for a 3.5 drive. It works, but does not allow for density switching. I use a lot of 3.5 disks in SD (92K) format, as well as DD. So, you may be reading directories after booting in DD on a SD disk that you have inserted in the drive. (or, the other way around) This will not work because the XF551 will only switch density when reading the first 3 sectors. The newer XF551 code patches the ROM so that the drive alternates density on every retry, regardless of which sector it is. This allows you to read the directory (or, whatever) of either density on any sector.

 

I would have to dump the ROMs in my drives to get the code, I'm afraid.

 

Bob

 

 

 

It is pretty much a matter of changing the ROM and connecting a 3.5 drive. I hooked mine up externally so I can switch select either 3.5 or 5.25 as the boot drive. You should use a 720K drive, although a 1.44 works if you are careful. (don't mount any 1.44 disks)

 

Bob, where can I find this replacement ROM?

 

Thanks

 

Not sure if this is the right one:

 

XF551.zip

 

I've not tried it (yet).

[ijor]

As Bob is saying, a 3.5 drive should be as fast/slow as a 5.25 drive.

 

You don't need a ROM patch to use a 3.5 drive, it should work out of the box (except for cabling issues). A 3.5 drive looks exactly the same as a 5.25 for the FDC and the drive's CPU. A custom ROM is needed only for the purpose of getting the extra capacity (720 K vs. 360 K). A custom ROM might also use the fast step rate for 3.5 drives (no idea if they actually do or not).

 

You can also use a standard HD PC drive. The only problem is that modern drives are jumperless and factory configured for drive #1, while the XF551 expects it to be as #0 (or #2 instead of #1, if you prefer). The jumpers are usually still there, but they need soldering or a cutting a trace. You can workardound crossing the wires on the cable as well. You can even use HD disks if you don't have DD ones. It would be slightly less reliable, but it should work.

 

Yes, an ST drive should work fine. Even a single sided one should still work, except of course that you would be limited to single sided densities only. It should be possible, if somebody wants, to make a custom ROM to still use a SS drive with 360K formats (by remapping the back side sectors to the higher tracks).

 

(This is mostly from theory, I really don't have much actual experience with XF-551 drives).

[+Larry]

(snip)

I would have to dump the ROMs in my drives to get the code, I'm afraid.

Bob

 

Here is the latest text file for Bob's XF mod that I have:

 

 

 

-Larry

 

Edit: BTW, one can use 3-1/2" (the type that uses pins) IDC ribbon connectors male + female to extend the original cable, if required.

Edited December 6, 2008 by Larry

[+Allan]

My issue with using the XF551s to convert to a 3 1/2 inch drive is that they are rare and expensive. I would love to see a way to use modern PC USB floppy drives. They are cheap and plentiful and would make a nice quick way of getting files back and forth from different computers.

 

Allan

[pepax]

My issue with using the XF551s to convert to a 3 1/2 inch drive is that they are rare and expensive. I would love to see a way to use modern PC USB floppy drives. They are cheap and plentiful and would make a nice quick way of getting files back and forth from different computers.

 

Allan

There is a USB floppy drive on this list but no details AFAIK.

But aren't SD cards cheap and and plentiful, too?

[+bob1200xl]

Yep - that's it!

 

Good job, Larry.

 

Just dump your XF551 ROM and edit the file where indicated. (you can do this with your programmer, usually) Then, burn a new ROM and plug in your 3.5 drive.

 

Bob

 

 

 

(snip)

I would have to dump the ROMs in my drives to get the code, I'm afraid.

Bob

 

Here is the latest text file for Bob's XF mod that I have:

 

 

 

-Larry

 

Edit: BTW, one can use 3-1/2" (the type that uses pins) IDC ribbon connectors male + female to extend the original cable, if required.

[+bob1200xl]

How about a 1050 hack that not only allows 3.5 drives, but also 1.44meg? The problem (IMHO) with other types of hacks is the very limited expertise available in the community. By that, I mean 8040 code, USB drivers and such. We are a community of 6502/65816 folks. Something like a 1050 would be built on 6507 based hardware, where, with the commented source code, all sorts of Atari users might be able to contribute. Once we venture into other hardware platforms, who has the tools or the knowledge? I had to write an 8040 disassembler in Atari Basic and then sit there for days with an 8040 hardware manual just to patch the XF551. I would much rather spend that time hacking a 1050.

 

Bob

 

 

 

My issue with using the XF551s to convert to a 3 1/2 inch drive is that they are rare and expensive. I would love to see a way to use modern PC USB floppy drives. They are cheap and plentiful and would make a nice quick way of getting files back and forth from different computers.

 

Allan

[ijor]

How about a 1050 hack that not only allows 3.5 drives, but also 1.44meg? The problem (IMHO) with other types of hacks is the very limited expertise available in the community. By that, I mean 8040 code, USB drivers and such.

 

The main problem, IMHO, is that is not worth. You can't use HD density just by implementing a custom ROM. You would need a high density FDC, or use something like the Mega STe Ajax. Either way you would need to, at least, double the system clock.

 

While this is certainly possible, it is not a simple hack. And then, if you are going to implement a major hack, why not go all the way to MMC, SD, CF cards or similar?

 

You could if you want, implement something to use USB floppy drives. But what's the point? USB drives can't read old "real" Atari floppies. So again, I think flash based cards, or USB thumb drives, are probably a much better solution.

[+bob1200xl]

You would have to double the FDC clock, yes, but the 2797 will run an HD drive of either size. By gating both 2mhz and 4 mhz to the FDC, you can do all three desities, SD, DD, and HD with the one FDC.

 

It looks pretty simple to me... only question is if a 1mhz 6507 can keep up with the data rate of a HD disk. Maybe. For sure, a 4mhz 65816 will.

 

Cf cards and such don't seem very useful unless they are on the PBI. (or, at least have a very high speed path into memory) You have 10 or 20 megabytes on media that you want to move - how long will that take on SIO? How about 200 megabytes? How would you search such a device? You know, "find me all disks that satisfy *.BXL"

 

Actually implementing a CF card on a 1050 would not be too hard, and only a little more difficult if you wanted to run both floppies and CF card(s). At least you could run data between CF cards and floppies at decent speed inside the 1050.

 

Bob

 

 

How about a 1050 hack that not only allows 3.5 drives, but also 1.44meg? The problem (IMHO) with other types of hacks is the very limited expertise available in the community. By that, I mean 8040 code, USB drivers and such.

 

The main problem, IMHO, is that is not worth. You can't use HD density just by implementing a custom ROM. You would need a high density FDC, or use something like the Mega STe Ajax. Either way you would need to, at least, double the system clock.

 

While this is certainly possible, it is not a simple hack. And then, if you are going to implement a major hack, why not go all the way to MMC, SD, CF cards or similar?

 

You could if you want, implement something to use USB floppy drives. But what's the point? USB drives can't read old "real" Atari floppies. So again, I think flash based cards, or USB thumb drives, are probably a much better solution.

[ijor]

Hi Bob,

 

You would have to double the FDC clock, yes, but the 2797 will run an HD drive of either size. By gating both 2mhz and 4 mhz to the FDC, you can do all three desities, SD, DD, and HD with the one FDC.

 

Yes, you can double the clock and gate it. That's what I meant by using "something like the STe Ajax", because it is implemented exactly like that.

 

However I was talking about an XF-551 mod, not about a 1050 one. Implementing such a mod on a 1050 sounds more complicated. You would need to modify the mechanism interface, which is already "PC drive compatible" in the XF-551 case, but not in the 1050. You would probably need more RAM, or use an already enhanced 1050. And what is perhaps more important, doubling the clock on the 279X would not be enough.

 

The controller on the XF-551 (the 1772) has a fully internal digital data separator, it operates a the chip clock rate. But the 279X uses an analog data separator that requires external components. You would probably need to duplicate the external analog logic, adjust them both, and select between them.

 

It looks pretty simple to me

 

It is certainly doable, but it's far and beyond than a "simple" EPROM change.

 

Cf cards and such don't seem very useful unless they are on the PBI. (or, at least have a very high speed path into memory) You have 10 or 20 megabytes on media that you want to move - how long will that take on SIO?

 

That's a good point. But if using a "mass storage device" like this, then you probably would better partition it into multiple disk images and/or drives.

[+bob1200xl]

Well, now you're going to make me get out my book... except I don't know where it is at the moment. I do recall looking over both chips long ago and determining that the 2797 would be the way to go. In fact, the 1772 is not suitable for all three densities for some reason. The 2797 has all the PC drive signals - the 1050 just does not use them. As far as the analog adjustments go - have you tried them? I don't remember seeing anything in the manual about them being a problem at different data rates. You set them up at 600khz and they should be good for 300 and 150. (from what I remember) The external analog components are already there in the 1050.

 

Heck, as long as we don't have to be compatible with existing 5.25s, (because we're 3.5 media) why not just write everything in HD? The Atari thinks 128 byte sectors are SD and 256 byte sectors are DD. They don't have to be. We can write 128 byte sectors in any density we want, right? That's what ED mode does. We can format 52 sectors per track using 128 byte sectors. 160 tracks... 8320 sectors. 1064960 bytes.

 

The whole point of this is to be able to use 1.44 drives and media in a cheap, available 1050 host, actually. (or 1.2meg drives, I suppose) 720K media and drives as well as XF551s, are getting hard to find. I have a hundred 1050s... I even have one with a PC drive mounted in it. (no electronics) I think you just have to drill out the existing threaded mounting holes and sand a little on the front bezel. Looks OK!

 

Bob

 

 

 

Hi Bob,

 

You would have to double the FDC clock, yes, but the 2797 will run an HD drive of either size. By gating both 2mhz and 4 mhz to the FDC, you can do all three desities, SD, DD, and HD with the one FDC.

 

Yes, you can double the clock and gate it. That's what I meant by using "something like the STe Ajax", because it is implemented exactly like that.

 

However I was talking about an XF-551 mod, not about a 1050 one. Implementing such a mod on a 1050 sounds more complicated. You would need to modify the mechanism interface, which is already "PC drive compatible" in the XF-551 case, but not in the 1050. You would probably need more RAM, or use an already enhanced 1050. And what is perhaps more important, doubling the clock on the 279X would not be enough.

 

The controller on the XF-551 (the 1772) has a fully internal digital data separator, it operates a the chip clock rate. But the 279X uses an analog data separator that requires external components. You would probably need to duplicate the external analog logic, adjust them both, and select between them.

 

It looks pretty simple to me

 

It is certainly doable, but it's far and beyond than a "simple" EPROM change.

 

Cf cards and such don't seem very useful unless they are on the PBI. (or, at least have a very high speed path into memory) You have 10 or 20 megabytes on media that you want to move - how long will that take on SIO?

 

That's a good point. But if using a "mass storage device" like this, then you probably would better partition it into multiple disk images and/or drives.

[ijor]

I do recall looking over both chips long ago and determining that the 2797 would be the way to go. In fact, the 1772 is not suitable for all three densities for some reason.

 

The 1772 is a much better controller than he 279X. It can certainly do all the densities. Once again, all the ST HD floppy upgrades do it. Of course that those upgrades don't implement single density (FM), but only because there was no need to at the ST.

 

OTOH, the 279X is already indicated and designed for 500 KHz with a double clock (because of its support for 8'' drives), while the 1772 is not. Most 1772 chips can be overclocked, but that is not guaranteed, that's why Atari made the Ajax, which is just a 1772 factory tested for x2 overclocking.

 

As far as the analog adjustments go - have you tried them? I don't remember seeing anything in the manual about them being a problem at different data rates. You set them up at 600khz and they should be good for 300 and 150. (from what I remember)

 

I didn't try, but of course that you need a different circuit and adjustment for HD (otherwise, how in earth you could get the VCO to operate at the different frequency). Note that this is not needed for single density (FM) vs. double density (MFM), it is only needed for HD. This is because both SD and DD use the same data rate after encoding (same flux transition rate), the difference between FM and MFM is in the encoding. HD however, use twice the encoded data rate and flux transition rate.

 

The 2797 has all the PC drive signals - the 1050 just does not use them.

 

I was talking in this case, about the XF-551 vs. the 1050, not about the 279X vs the 1772.

 

Heck, as long as we don't have to be compatible with existing 5.25s, (because we're 3.5 media) why not just write everything in HD? The Atari thinks 128 byte sectors are SD and 256 byte sectors are DD. They don't have to be. We can write 128 byte sectors in any density we want, right?

 

Yes, you can write any sector size in any density. And yes, I agree, using HD all the time seems a good idea.

 

I have a hundred 1050s...

 

Wooow. That was literally? not a typo?

[+bob1200xl]

If you're on the 4mhz clock and set the FDC to 8 inch mode, it runs at HD. If you then set it to 5.25 mode, it runs DD, right? The VCO has to change when you switch - same at 2mhz. The WPW and RPW both change, also. How do you change the VCO? A VCO uses an oscillator and dividers to lock onto an incoming signal. You can change the capture frequency by adding/deleting more divider stages.

 

 

 

Ah-ha!! Found my book!

 

 

The 1772 only supports 125khz and 250khz data streams - no 500khz mode. Is that why you're talking about clock doublers?

 

The 2793 (not the 2797) has three control pins for mode - 5/8, DDEN, and ENMF (Double DENsity and ENable Mini-Floppy). Using these pins, you can select 125khz, 250khz and 500khz data streams.

 

My mistake - the clocks are 2mhz and 1mhz, not 4mhz and 2mhz.

 

Best way to find out if this works is to build one. But, I have to do some 65816 stuff, first.

 

Well, I've never counted them but they take up a lot of room in storage. Many of them are defective, waiting to be fixed. I bought a whole truckload of stuff from San Jose Computer when he closed up.

 

Bob

 

 

 

 

 

 

I do recall looking over both chips long ago and determining that the 2797 would be the way to go. In fact, the 1772 is not suitable for all three densities for some reason.

 

The 1772 is a much better controller than he 279X. It can certainly do all the densities. Once again, all the ST HD floppy upgrades do it. Of course that those upgrades don't implement single density (FM), but only because there was no need to at the ST.

 

OTOH, the 279X is already indicated and designed for 500 KHz with a double clock (because of its support for 8'' drives), while the 1772 is not. Most 1772 chips can be overclocked, but that is not guaranteed, that's why Atari made the Ajax, which is just a 1772 factory tested for x2 overclocking.

 

As far as the analog adjustments go - have you tried them? I don't remember seeing anything in the manual about them being a problem at different data rates. You set them up at 600khz and they should be good for 300 and 150. (from what I remember)

 

I didn't try, but of course that you need a different circuit and adjustment for HD (otherwise, how in earth you could get the VCO to operate at the different frequency). Note that this is not needed for single density (FM) vs. double density (MFM), it is only needed for HD. This is because both SD and DD use the same data rate after encoding (same flux transition rate), the difference between FM and MFM is in the encoding. HD however, use twice the encoded data rate and flux transition rate.

 

The 2797 has all the PC drive signals - the 1050 just does not use them.

 

I was talking in this case, about the XF-551 vs. the 1050, not about the 279X vs the 1772.

 

Heck, as long as we don't have to be compatible with existing 5.25s, (because we're 3.5 media) why not just write everything in HD? The Atari thinks 128 byte sectors are SD and 256 byte sectors are DD. They don't have to be. We can write 128 byte sectors in any density we want, right?

 

Yes, you can write any sector size in any density. And yes, I agree, using HD all the time seems a good idea.

 

I have a hundred 1050s...

 

Wooow. That was literally? not a typo?

[ijor]

The 1772 only supports 125khz and 250khz data streams - no 500khz mode. Is that why you're talking about clock doublers?

 

Yes, but doubling the clock is required in either case. The difference is that the 1772 need overclocking.

 

How do you change the VCO? A VCO uses an oscillator and dividers to lock onto an incoming signal. You can change the capture frequency by adding/deleting more divider stages.

 

Hmm, no. You are describing a digital PLL, not a VCO. A VCO, as implied by its name, is controlled by voltage, not by divider stages. You can further divide the output of the VCO frequency though. But you need the right VCO frequency and filters in the first place.

 

The 2793 (not the 2797) has three control pins for mode - 5/8, DDEN, and ENMF (Double DENsity and ENable Mini-Floppy).

 

You can use the 5/8 mode to select the higher frequency, but the external analog circuit still need to be changed and adjusted accordingly.

[+bob1200xl]

You could be correct. I'll find out, later.

 

Bob

 

 

 

The 1772 only supports 125khz and 250khz data streams - no 500khz mode. Is that why you're talking about clock doublers?

 

Yes, but doubling the clock is required in either case. The difference is that the 1772 need overclocking.

 

How do you change the VCO? A VCO uses an oscillator and dividers to lock onto an incoming signal. You can change the capture frequency by adding/deleting more divider stages.

 

Hmm, no. You are describing a digital PLL, not a VCO. A VCO, as implied by its name, is controlled by voltage, not by divider stages. You can further divide the output of the VCO frequency though. But you need the right VCO frequency and filters in the first place.

 

The 2793 (not the 2797) has three control pins for mode - 5/8, DDEN, and ENMF (Double DENsity and ENable Mini-Floppy).

 

You can use the 5/8 mode to select the higher frequency, but the external analog circuit still need to be changed and adjusted accordingly.

[+Larry]

FWIW-

 

The CSS Floppy Board overclocks a 1772 to do HD disks. This is the "1772-02-02."

It is not terribly reliable with 1.44's, at least in my sample of two BB/FB's.

 

Atari supposedly started its development of the the XF-551 (original 3-1/2"

version) with a modified 1050 in DS/DD -- 720K.

 

I'd never say never, but all of my experience with HD disks says that it is

unreliable, whether it is on a PC or an Atari.

 

It would be neat to have, but seems like a lot of work for very little payoff. But obviously that's just IMO.

 

-Larry

[+bob1200xl]

The XF551 isn't exactly ISO9001 material, either. Using 1050s is a way around the XF551. It doesn't have to use HD drives - any standard drive would work. But, even I don't have truckloads of new DD drives so HD needs to be an option.

 

My HD experience has been that if it formats OK, it will work OK, but I have seen a lot of disks that would not format clean.

 

Anyway, a 6507/65816 platform in a 1050 could be used for more than floppies, you know. There is space for a CF card, too. No HALTs, no NMIs. Easy to read IBM disks...

 

I can't believe that Bob Puff used an overclocked 1772. Were they actually factory certified to run at 16mhz?

 

 

Bob

 

 

 

FWIW-

 

The CSS Floppy Board overclocks a 1772 to do HD disks. This is the "1772-02-02."

It is not terribly reliable with 1.44's, at least in my sample of two BB/FB's.

 

Atari supposedly started its development of the the XF-551 (original 3-1/2"

version) with a modified 1050 in DS/DD -- 720K.

 

I'd never say never, but all of my experience with HD disks says that it is

unreliable, whether it is on a PC or an Atari.

 

It would be neat to have, but seems like a lot of work for very little payoff. But obviously that's just IMO.

 

-Larry

[ijor]

I can't believe that Bob Puff used an overclocked 1772. Were they actually factory certified to run at 16mhz?

 

They were not, only the Ajax (the one used in the Mega STe/Falcon) was.

 

But it was common practice to overclock the "1772-02-02". That was a later version of the plain 1772, and it was "well known" that most of them worked fine at 16 MHz. I have no idea who made the statistics and how true they are, but 90% success was "reported".