• Please review our updated Terms and Rules here

DF32/DS32 Platter Emulation

m_thompson

Veteran Member
Joined
Jul 8, 2014
Messages
2,161
Location
Rhode Island, USA
The RICM has an early Negibus DF32 that we would like to connect to the Classic PDP-8. The platter is very corroded and not usable. David Gesswein made a working prototype of a CPLD based platter emulator, but I don't think that it was ever made "production ready". Warren Stearns also worked on a CPLD based platter emulator, but I don't think that he finished the project.

Has anyone else developed a working DF32 platter emulator?

What would it take to get a design "production ready"?
 
Probably not many DF32's remaining in the world, and I'd imagine platter/drum interfaces would vary a lot depending on specific hardware configurations.
You're looking to emulate this part?
1729618075765.png
It might be possible to make this using the main board of the RK05 emulator as the hardware platform and develop different FPGA code to emulate the DF32. The existing emulator has 20 inputs and 20 outputs. There's a 256 Mbit DRAM. DF32 has only 16 tracks of 2048 13-bit words plus a gap, so data to/from the platter could be hugely oversampled and stored in DRAM. That FPGA logic would be completely different than the RK05 emulator, but conceptually not so difficult.
Some details would need to be worked out... saving/restoring data at power off/on, interface, etc.

DF32 logic looks like it's comprised of M-series 5-volt modules, so interfacing might not need negative logic level conversion.

If that's interesting to someone I could help them become familiar with the existing hardware, and possibly support some one-off modifications to PCBs or whatever if it doesn't deviate far from the present path.
 
You're looking to emulate this part?
View attachment 1288416

Yes, 4x platters connected to the controller on the DF32

It might be possible to make this using the main board of the RK05 emulator as the hardware platform and develop different FPGA code to emulate the DF32. The existing emulator has 20 inputs and 20 outputs. There's a 256 Mbit DRAM. DF32 has only 16 tracks of 2048 13-bit words plus a gap, so data to/from the platter could be hugely oversampled and stored in DRAM. That FPGA logic would be completely different than the RK05 emulator, but conceptually not so difficult.
Some details would need to be worked out... saving/restoring data at power off/on, interface, etc.

DF32 logic looks like it's comprised of M-series 5-volt modules, so interfacing might not need negative logic level conversion.

There are two generations of DF32s, The early ones are based on the B/R/S negative logic FlipChips, and the later ones are based on the M positive logic FlipChips. I am not sure if the analog interface to the heads changed with the new generation.

If that's interesting to someone I could help them become familiar with the existing hardware, and possibly support some one-off modifications to PCBs or whatever if it doesn't deviate far from the present path.
 
DF32 logic looks like it's comprised of M-series 5-volt modules, so interfacing might not need negative logic level conversion.
DF32cards.jpg

There's two types of DF32/DS32. The earlier models use 'negi-bus' logic cards (as shown above). The later models used 'posi-bus' logic comprised of M-series standard 5V TTL chips. To create something that Mike wants requires level converters. Vince has info and a pic of his DF32 emulator.
 
The red circle seems to be the platter head interface which is analog signals. Probably could plug in place of the reader/writer cards to have digital interface. To have more useful capacity with a single card I actually emulated 4 DS32 expander drives and plug into where the two cables would go to DS32 drives. That allowed 128k word capacity which is a reasonable amount especially with disk monitor system.

As stated DF32D is TTL, non D is R series logic. My board is for R series. I'm not sure I've seen a D. Seen a number of non D.

My board was done 16 years ago. For it I decided to try toner transfer etching so layout little strange. With modern board could make it proper depth. Biggest problem is obsolete components. The FPGA I used is long obsolete. You can get some large enough battery backed SRAM but very expensive now. FRAM may be easiest replacement.

FPGA I used it was specified for both maximum voltage and maximum current on inputs so you could use the protection diodes as clamps. If a new FPGA doesn't support that you will need some more level shifters. Other option is 3.3V FPGA running at 0 and -3.3V and could get rid of some of the level shifters. Slightly larger FPGA would have been better. Suspect anything modern will be larger.

Then question of if board should be enhanced. Now you need to use my dumprest to write image to the board. Lots of options for ways to directly transfer disk images with some more modern electronics.
 
Both emulators, Vince’s and Dave’s, look very nice. Seems like one of them could be replicated. No need to make a 3rd one.
 
Warren did not finish his emulator. He thought he was pretty close.. I gave Kyle what I could find of his work but at this point it would be easier to start over since it has been over 5 years.

Like djg I don't think I have seen a DF32D, only the Negibus version.

I have 2 platters in my Straight 8 but I will never spin them up. The platters have degraded too much. I suppose new platters could be fabricated but if an emulator exists then why? Performance would be the same. They don't make enough noise to hear over the cabinet fans (if working correctly).

We ran PS/8 mostly (PS/8 was the precursor to OS/8) and you would start your session by running DDU (Disk Dectape Utility) to load one of 5 images from the DecTape onto the DF32. You would then boot PS/8 from the DF32. There was enough room for the OS, PIP, an editor, and one language compiler or assembler with some room for program storage. When your session was over you would run DDU again and back up the 64k word DF32 image to one of the 5 slots on the DecTape. This was a little extra effort but the DF32 is so much faster than the DecTape especially when compiling that it was worth it. You couldn't leave your work on the DF32 and expect it to be there next time.
 
Warren did not finish his emulator. He thought he was pretty close.. I gave Kyle what I could find of his work but at this point it would be easier to start over since it has been over 5 years.

Like djg I don't think I have seen a DF32D, only the Negibus version.

The RICM has another DF32 that is the version with the M modules instead of the R modules. I didn't check to see if it is a DF32-DP or a DF32-DN.

I have 2 platters in my Straight 8 but I will never spin them up. The platters have degraded too much. I suppose new platters could be fabricated but if an emulator exists then why? Performance would be the same. They don't make enough noise to hear over the cabinet fans (if working correctly).

I haven't seen anything definitive that says what the magnetic material plated on the disk is. It might be possible to get the disks polished and replated, but like you said, why bother.

We ran PS/8 mostly (PS/8 was the precursor to OS/8) and you would start your session by running DDU (Disk Dectape Utility) to load one of 5 images from the DecTape onto the DF32. You would then boot PS/8 from the DF32. There was enough room for the OS, PIP, an editor, and one language compiler or assembler with some room for program storage. When your session was over you would run DDU again and back up the 64k word DF32 image to one of the 5 slots on the DecTape. This was a little extra effort but the DF32 is so much faster than the DecTape especially when compiling that it was worth it. You couldn't leave your work on the DF32 and expect it to be there next time.

You have a DECtape containing PS/8 and the DDU on paper tape?
 
You have a DECtape containing PS/8 and the DDU on paper tape?
I am pretty sure I have a bootable PS/8 DECtape, I am not sure I have a DDU paper tape. I am not sure I have the sources for DDU. I do have the paper tapes for PS/8 so that I can build a PS/8 system for DECtape or DF32. I think DDU was submitted to DECUS. I need to get focused and put all this stuff online.

When I revived the machine after purchasing it I started using OS/8 instead. We had OS/8, but from what I can remember nobody felt any burning need to switch to it.
 
Probably not many DF32's remaining in the world, and I'd imagine platter/drum interfaces would vary a lot depending on specific hardware configurations.
You're looking to emulate this part?
View attachment 1288416
It might be possible to make this using the main board of the RK05 emulator as the hardware platform and develop different FPGA code to emulate the DF32. The existing emulator has 20 inputs and 20 outputs. There's a 256 Mbit DRAM. DF32 has only 16 tracks of 2048 13-bit words plus a gap, so data to/from the platter could be hugely oversampled and stored in DRAM. That FPGA logic would be completely different than the RK05 emulator, but conceptually not so difficult.
Some details would need to be worked out... saving/restoring data at power off/on, interface, etc.

DF32 logic looks like it's comprised of M-series 5-volt modules, so interfacing might not need negative logic level conversion.

If that's interesting to someone I could help them become familiar with the existing hardware, and possibly support some one-off modifications to PCBs or whatever if it doesn't deviate far from the present path.
George,

If you are going to emulate the DF32 with the RK05 Emulator, please consider also emulating the RS08 Drive. I don't know if the controller to drive interface is different between the DF32 and RS08.

I keep thinking of creating a board (with someone's help with the Data Break Logic) to emulate the RF08 Controller and 4 RS08 drives. From what I have been told the only difference between the RF08 and the DF32 controllers is the active level of a couple of bits. If this is true it should not be that hard ti make one board that could emulate both.
 
Back
Top