To add a hard disk to the PCjr, I've always been a fan of the IBM Fixed Disk Adapter PCjr modification.
Although SCSI is nice and the XTIDE projects out there are very cool, I like the BIOS reprogram of the IBM controller card because it is as close to an XT as you can get. This allowed me to run XT disk utilities without lower risk of compatibility problems and also boot from the hard disk after POST.
There are a number of disadvantages of the solution - limited to MFM hard disks, the disk table was limited to 4 drives defined in the controller EPROM - but the biggest disadvantage to me was the physical size of the card. It's a full length (13 inch!) ISA board, which meant that I simply could find a suitable enclosure for a 5.25" drive, power supply, and the controller board.
Well, I've finally worked up the courage to cut one of those controllers in half and wire the cut signals between the halves. That will finally allow me to get a drive into a small disk enclosure that sits next to the jr. But, I don't want to sacrifice my working controller.
There's an auction of compatible controllers going on right now: http://cgi.ebay.com/ws/eBayISAPI.dll?Vi ... 0753690389
At $47 for 7 controllers, it's a little pricey for me. Is anyone also interested in going down this path who would be interested in splitting the lot?
Anyone interested in an XT MFM controller
Re: Anyone interested in an XT MFM controller
Given their normal pricing, getting 7 original XT controllers for $47 seems kind of cheap. But the problem is what do you do with 7 controllers .. I have about 7 extra that I don't need already!
I'm looking forward to XT-IDE for the Jr. It's just a matter of fixing the BIOS ROM.
You are really going to mutilate a controller to fit it in a case? I shouldn't show you my monstrous Ethernet card. ;-0
I'm looking forward to XT-IDE for the Jr. It's just a matter of fixing the BIOS ROM.
You are really going to mutilate a controller to fit it in a case? I shouldn't show you my monstrous Ethernet card. ;-0
Re: Anyone interested in an XT MFM controller
I wondered if you had tested the XT-IDE on the jr! I tried a quick scan of the thread on vintage-computer, but at 130+ pages, I didn't look too hard. BIOS issue? Interesting. It's certainly solvable.
You know, the crazy thing about the Fixed Disk Adapter is that I have at least a couple spares, too! For some reason, I just don't want to cut into any of the ones I have. Buying another seven, well, now that's insane!
You know, the crazy thing about the Fixed Disk Adapter is that I have at least a couple spares, too! For some reason, I just don't want to cut into any of the ones I have. Buying another seven, well, now that's insane!
Re: Anyone interested in an XT MFM controller
At a minimum we have to change the checksum in the BIOS extension. The PCjr uses a different algorithm than the PC.
But after that, it should work fine. It only uses port I/O - there are not IRQs or DMA. That kept it simple and compatible across a side range of machines.
Like the XT BIOS change, I'd like to do a NEC V20 version of the XT-IDE BIOS too. The V20 can do repeated I/O faster, and that would be a big speedup on a machine that sorely needs it.
But after that, it should work fine. It only uses port I/O - there are not IRQs or DMA. That kept it simple and compatible across a side range of machines.
Like the XT BIOS change, I'd like to do a NEC V20 version of the XT-IDE BIOS too. The V20 can do repeated I/O faster, and that would be a big speedup on a machine that sorely needs it.
Re: Anyone interested in an XT MFM controller
I'd definitely be interested in an XT-IDE for the PCjr. I have an 8-bit MFM controller, but it uses DMA so I wouldn't be able to adapt it to the PCjr.
Re: Anyone interested in an XT MFM controller
Well, that's where the BIOS change comes in. Just because the card has DMA circuitry doesn't mean you have to use it ..
The XT controllers use DMA because the 8088 is horrible at doing I/O in loops. The 80286 does I/O in loops much faster, so they went to using repeated I/Os driven by the processor instead of the DMA engine in 80286 class controllers.
The modification that Matt is talking about takes an original XT controller and changes the BIOS to do I/O in loops. While slower than the DMA approach, it works on the Jr. And if you have a V20 there is a version of the BIOS that can perform much better. (The V20 has the same new instructions that make the 80286 faster in the I/O loops than DMA.)
The XT controllers use DMA because the 8088 is horrible at doing I/O in loops. The 80286 does I/O in loops much faster, so they went to using repeated I/Os driven by the processor instead of the DMA engine in 80286 class controllers.
The modification that Matt is talking about takes an original XT controller and changes the BIOS to do I/O in loops. While slower than the DMA approach, it works on the Jr. And if you have a V20 there is a version of the BIOS that can perform much better. (The V20 has the same new instructions that make the 80286 faster in the I/O loops than DMA.)
-
southernbob
- Posts: 36
- Joined: Mon Jun 23, 2008 10:36 am
Re: Anyone interested in an XT MFM controller
Are you guys talking about the Joel Smith XT diskcard BIOS change? I have/had 2 systems with the
modified card. One system uses the XT expansion unit with the driver/receiver card and the thick cable
going from the JR to the expansion box. The expansion box has 2- 20 meg disks. The other comes
directly from the Jr buss to the card thru a short cable. You need an EPROM burner to burn a new
BIOS EPROM for the card.I still like the Datafab Compact flash unit connected to the printer card. It is fairly
fast but is just for storing data whereas the modified XT card allows you IPL from the hard files
assuming you can find the appropiate hard files and the connecting cables.
Bob
PS: You also have to format the disks using Joel's program {I think because of a different interleave factor)
and that takes along time.
modified card. One system uses the XT expansion unit with the driver/receiver card and the thick cable
going from the JR to the expansion box. The expansion box has 2- 20 meg disks. The other comes
directly from the Jr buss to the card thru a short cable. You need an EPROM burner to burn a new
BIOS EPROM for the card.I still like the Datafab Compact flash unit connected to the printer card. It is fairly
fast but is just for storing data whereas the modified XT card allows you IPL from the hard files
assuming you can find the appropiate hard files and the connecting cables.
Bob
PS: You also have to format the disks using Joel's program {I think because of a different interleave factor)
and that takes along time.
Re: Anyone interested in an XT MFM controller
Hi Bob- I was talking about Joel Smith's BIOS modification. It's too bad that didn't see the light of day, back in the day. I suppose the responsible team didn't want to take the jr that far and no way could it be published externally from within IBM, lest people believe it would be supported.
You can re-interleave with the XT Advanced Diagnostics, although I think I used SpinRite to do it.
Anyway, the batch of controllers on eBay I'm referring to may be the Xebec controller, which was the OEM of the IBM card. If it's not exactly the IBM version, it may be very slightly different than the IBM version. The chip layout appears to be exactly the same, so I think it would work, but without the board being exactly IBM version there's always a chance of something happening.
If the requirement is a hard disk on the PCjr, starting from scratch, I'd go XT-IDE. The problems with the IBM XT Fixed Disk Adapter are the limited drive support and the now increasing difficulty of obtaining MFM hard disks. Both approaches (XT-IDE or IBM FDA) require a PCjr bus - ISA bus adapter.
Regardless, few things are as exciting in PCjr-land than when the jr boots from a hard disk!
You can re-interleave with the XT Advanced Diagnostics, although I think I used SpinRite to do it.
Anyway, the batch of controllers on eBay I'm referring to may be the Xebec controller, which was the OEM of the IBM card. If it's not exactly the IBM version, it may be very slightly different than the IBM version. The chip layout appears to be exactly the same, so I think it would work, but without the board being exactly IBM version there's always a chance of something happening.
If the requirement is a hard disk on the PCjr, starting from scratch, I'd go XT-IDE. The problems with the IBM XT Fixed Disk Adapter are the limited drive support and the now increasing difficulty of obtaining MFM hard disks. Both approaches (XT-IDE or IBM FDA) require a PCjr bus - ISA bus adapter.
Regardless, few things are as exciting in PCjr-land than when the jr boots from a hard disk!
Re: Anyone interested in an XT MFM controller
btw- I'd love to hear the back story on the Joel Smith project - who was Joel Smith, how the modification was born, how it was developed, and any kind of reaction/feedback was received.
Re: Anyone interested in an XT MFM controller
I never met Joel or bumped into him professionally. However, the first line of his forum posting announcing his success was:
"Wheeeeeeeeeeeeeeeeeeeeeeeeeeeeeew! Look at that PCJR go!"
That was dated June 23rd, 1986. Joel then goes on to describe at a high level what he did. It basically involved rewriting the XT BIOS to not use DMA. He had considered writing V20 specific code, but didn't want to desolder is 8088 for it. (V20 versions of the code were available later.)
He first asked about the feasibility of the project on June 5th. (That's pretty short period to go from start to finish.)
And of course, it was well received and it generated a lot of interest. Prior to this people were hacking on 2nd and 3rd floppy drives, and trying to figure out how to support the higher density drives. Joel's project can also be seen as the genesis of the PC bus adapter project since he had to work all of the mappings between the two buses out to make this work.
"Wheeeeeeeeeeeeeeeeeeeeeeeeeeeeeew! Look at that PCJR go!"
That was dated June 23rd, 1986. Joel then goes on to describe at a high level what he did. It basically involved rewriting the XT BIOS to not use DMA. He had considered writing V20 specific code, but didn't want to desolder is 8088 for it. (V20 versions of the code were available later.)
He first asked about the feasibility of the project on June 5th. (That's pretty short period to go from start to finish.)
And of course, it was well received and it generated a lot of interest. Prior to this people were hacking on 2nd and 3rd floppy drives, and trying to figure out how to support the higher density drives. Joel's project can also be seen as the genesis of the PC bus adapter project since he had to work all of the mappings between the two buses out to make this work.