MFM drive from NorthStar Horizon

[Ashlin Inwood]

I have a drive form a NorthStart the system was in very bad shape but I have got the drive out and am trying to read it. The drive did  have some bad chip, I have replaced them, it may have more. The mechanical elements of the drive seem good.

Drive: Tandon TM 602 S (I would like to try swap the PCB if anyone has one)
Controller: unknown, but its and S100 card made by NorthStart (cant get it out, rusted in)

I have been tring to read with this command:
mfm_read --analyze --emulation_file north.emu --extracted_data_file north.iso --transitions_file noth.txt

Output:

Found drive at select 1
Drive RPM 3600.2
Primary transition period 165 ns, should be around 200
Primary transition period 85 ns, should be around 200
Retrying on cylinder 0 head 1
Primary transition period 162 ns, should be around 200

Board revision C detected
Found drive at select 1
Drive RPM 3600.0
Primary transition period 85 ns, should be around 200
Primary transition period 85 ns, should be around 200

[David Gesswein]

The transitions timing data looks like either the heads are out of alignment or
you still have problems in the read electronics.

https://groups.google.com/g/mfm-discuss/c/xWEmnWrFBy0/m/qFKPi8TbBQAJ
First number is number of 5ns clocks between edges.

If you change the 0 to 1 in mfm/analyze.c and run make you will get the
histogram prints shown in the link.
#define GENERATE_HISTOGRAM 0

Try for each head --ana=cyl,head to see if you get a reasonable histogram for
any head.
If you do some of heads are likely out of alignment and you will need to use
this method to recover contents
https://www.pdp8online.com/mfm/microstep/index.shtml

If none of the heads show good data they all could be out of alignment or
it could be a problem with the read electronics. If you have a scope you
can look at the test points in the read path. Its differential so you may need
to do a-b to get a good signal. You should see transitions at around 200ns,
300ns, 400ns like the histogram.

If you want me to take a look create a transitions file with
mfm_read --transitions_file raw_data --drive 1 --heads # --cylinders #
Compress the file and upload to
https://www.pdp8online.com/upload/abuploadab.html

> --
> You received this message because you are subscribed to the Google Groups "MFM Discuss" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to mfm-discuss...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/mfm-discuss/9bfb1aae-3027-42bd-941a-7dde33e6eca4n%40googlegroups.com.

[Ashlin Inwood]

Thank you very much

I did have a quick try as you suggested. I am sure if it makes any difference which cylinder I used, I tried cylinder c0 and one on c60
The histogram can be found here. It doesn't look great, but it is consistent across most of the heads.
https://docs.google.com/spreadsheets/d/1n70is44w2j4LuwaMIvk5EgTqsWlfoHtSUja0any1exI/edit?usp=sharing

This hard drive, when it was in the system, had a resistor pack installed. I am only able to talk to the drive when it's removed. The mfm emulator also has RN2 installed.

Without knowing anything about MFM drives, the consistency in the reading makes me think it's a fault in the PCB. One of the parts I had to replace was a differential line receiver (AM26LS32), quite possible that the line drive has also died (AM26LS31).

Let me know what you think

Thank you

[Ashlin Inwood]

Thanks

I had made a post previously. It looks to have been deleted, I'm not sure why. Anyway, I'll rewrite it, but super briefly.

Histograms look bad (I think). Consistent shape between heads, I'm inclined to think it is a problem with the logic board. I need to get some differential probes to see what is going on with read heads and related circuitry.

I only check the histograms on cylinder 0, not sure if that makes a difference.

The MFM drive has a resistor pack, I can only talk to the MFM emulator while that resistor pack is removed. The  MFM emulator's resistor pack (RN2) is installed. FYI, the chip that I had to replace to get anylife was a line receiver (AM26LS32). Possible other similar chips are dead.

Let me know what you think

Thanks

On Sunday, June 1, 2025 at 1:26:31 AM UTC+12 d...@pdp8online.com wrote:

[d...@pdp8online.com]

Google implemented a poor spam filter a while ago that I have no control over. It also only notifies me once a day that there are pending messages so if it declares it spam it may be a day before I can approve it. Don't remember it showing the messages as deleted before. May be another "improvement".

Correct the histograms look bad. Does look like we are seeing MFM encoded data  for most of the lines.

The resistor pack on the drive should be installed for proper operation. Normally it won't work at all with the resistor removed. The signal drivers are open collector so need pullups to get proper high levels. What was the error you are getting with it installed? Was pin 1 installed in the correct orientation? I would probably figure this out first.

All the northstar disks I have seen have the same format. You can see if it can decoding any data by forcing format. If you captured a transition file of the disk you can use (replace file names as appropriate)

mfm_util --tran cray-northstarb.tran --format NorthStar_Advantage --sectors 16,0 --heads 4 --cylinders 153 --header_crc 0x0,0x0,16,0 --data_crc  0x0,0x0,32,0 --sector_length 512  --ext /tmp/file.ext --quiet 1

That will print out lines like this if it manages to decode a sector header.  You can see if they seem reasonable if any decode.

Got exp 35,3 cyl 35 head 3 sector 3 size 512 bad block 0

If using mfm_read add --unbuffered_seek --drive 1

Saw picture of drive online. Seems the stepper motor and head lever is all exposed. We may be able to set up the read appropriately and try physically shifting the heads and see if that improves the read to tell if its the heads or electronics. Always some chance of something going wrong when doing that.