• Please review our updated Terms and Rules here

TU58 repair and tolerances

czunit

Experienced Member
Joined
Aug 7, 2015
Messages
499
So in addition to everything else I'm working on fixing a TU58 drive. It's one of those "Got to do one thing, then another then...." kinds of projects. So far I have:

  • Fixed the power supply so I have solid +5 and +12 volts.
  • Gotten it cabled to a pdp11/23+ CPU (10 pin ribbon cable, right orientation)
  • Replaced the crumbled puck rubber with new .5 inch tubing
  • Gotten the unit to pass internal diagnostics (LED comes on for 1/2 second, off for 1/2 second, then solid on)
  • Verified baud rate and interface type (9600/9600, RS232)
I was going to work on it with RT11, but my version is 5.05 which doesn't support TU58 and half the code is missing so I can't build the driver. I tried loading 5.01 from some RX50 floppies but they are bad in the way of "you can read them once, glad I copied them to .DSK files" bad. Oh well.

Anyway I'm trying the 11/23+ bootstrap. When I type Boot dd0: I get:
TRYING UNIT DD0

ERROR UNIT DD0
ERR 14 DRIVE ERROR

Same issue with DD1:. The motors do start and seem to move forwards and backwards for a second then it throws a fault.

So.... First I'm wondering if the puck size is wrong. Looking at the technical specifications I see the drive motor puck has a radius of .273. Mine now is probably about .5 I'll check with a caliper but how critical is that measurement? Is there a specific tube type I should be using (and if so can someone with a lifetime supply send me some?)

Second, where would be a good place to start looking for issues? Any thoughts?
 
"some RX50 floppies but they are bad in the way of "you can read them once, glad I copied them to .DSK files"

Expect the same with the Tu58 tapes. I have a Tu58 drive somewhere in the basement. But never spent a minute on it because of the known tape problems. Maybe others have a more positive view on these?
 
Yep,

I had to restore some software from a 35 year old TU58 tape. They just wouldn't budge in a refurbished drive that I knew worked.

I had to send the tapes away for the data to be extracted. Fortunately, I had two tapes with the same software on them, so I could stitch together one complete image from the two extracted tapes.

Unfortunately, TU58 tape data recovery is now a dying art (literally)!

This was for work, so we'll worth the money...

Expect tapes not to work. You won't be disappointed then...

Dave
 
I haven't used a TU58 in at least 5 years, but last I did almost of the tapes worked fine.
I'm near, but not too near the coast, so my collection lives in moderate temperatures and humidity, so I expect that helps.
 
You are sure that those TU58 has a proper bootstrap on them, right? The ones you got from me probably didn’t.

I would recommend booting of some other medium. I usually boot of a SCSI card with a SCSI2SD. Then I try to read the tapes from within RT11. That was my standard procedure when dumping a big heap of TU58 tapes. Mostly RT11 and VAX diagnostics. Dumps of DECtapes (TU58 and TU56) There is a full TU58 RT11 V4 kit there.

I don’t think that they are very sensitive to the size of the capstan wheel. I have used silicone tubing which I glued on the the aluminium hub. Then sanded it down to the proper size.

Probably restored 6 or 7 TU58 drives without big problems.
 
Fortunately none of the tapes contain anything worth spit, so this is more of a "gee I'd love to run RT11 slower than a garbage truck". Now that I know I can at least communicate with it I'll focus more time and effort on getting 5.01 running.
 
You are sure that those TU58 has a proper bootstrap on them, right? The ones you got from me probably didn’t.

I would recommend booting of some other medium. I usually boot of a SCSI card with a SCSI2SD. Then I try to read the tapes from within RT11. That was my standard procedure when dumping a big heap of TU58 tapes. Mostly RT11 and VAX diagnostics. Dumps of DECtapes (TU58 and TU56) There is a full TU58 RT11 V4 kit there.

I don’t think that they are very sensitive to the size of the capstan wheel. I have used silicone tubing which I glued on the the aluminium hub. Then sanded it down to the proper size.

Probably restored 6 or 7 TU58 drives without big problems.
Well, I have an old xxdp tape and one of your tapes does the same thing: Quick action on the motors, then the failure. I'd expect it to go to beginning of tape and read block 1 before flunking out. What happens on your system when you try to boot a tape without a bootstrap?

Good to know on the capstain size, I'll measure it with calipers and see what it is. What size do you sand down to, if I might ask?

C
 
Fortunately none of the tapes contain anything worth spit, so this is more of a "gee I'd love to run RT11 slower than a garbage truck". Now that I know I can at least communicate with it I'll focus more time and effort on getting 5.01 running.
 
I checked the capstains and with the silicon tubing they are .52 inches in diameter. Should be good enough to work, any thoughts?

Next step might be to get a scope out and see if the tachometer is sending pulses and if anything is visible on the read heads....
 
So. Finally got my Gotek floppy thing from Ebay and sure enough it works so I can finally boot up RT11 5.3 and see what's going on with the TU58.

Basically TU58 fails with directory error and failed format. I think it's a hardware problem so need to trace this thing a bit and see what is going wrong. So far I have used three tapes (one mine from 30 years ago, two others from @MattisLind . All same issues, interestingly enough in either drive so it's not a channel specific problem unless it's affecting both drives.

It's possible the capstain is not spinning the tape at the right speed, but it looks like it's working. I didn't glue it, but it is tight and I don't think it's slipping. Unless it is :)

I tried disconnecting one of the tachometer LED lines on each drive and with them off it doesn't even start to spin, so those are probably working.

Hm.
 
So, in the "It ain't gonna fix itself" category I figured I would work on this drive again. Fired up xxdp+ 2.5 and while I don't know where the TU58 tests are I can do a dir dd1: and have the unit try to load. It fails, but it's consistent.

Big problem is what's going on. I took a look at the technical manual and it looks like there are a number of common components between the tape read heads and the CPU that should be getting bits. Specifically there is:

Heads--> Head select matrix-->Read amplifier-->Peak detector and decoder-->IO port on 8085 CPU.

So it's quite possible the problem is somewhere between the head and the CPU. I think what I will try is hooking my logic analyzer to the output of the read amplifier and the CPU to see if I can see the tape data signals in any form before they get to the processor. If no data then there is a problem in the read chain. If data I have a different problem.

Anyone know if there are schematics of the board? I'd rather not fuzz this out by hand.
 
Ok. Based on that schematic I should check at the output of E29's Op Amp (pin 13) to see the raw signal shaped. That leads to the input to the clocker which outputs at E29. If I see signals there then it's spotting *something* at the heads. If not, will walk back.
 
Ok, interesting. Let me type out my notes.

First I decided to start at the latching flip flop at the end of the circuit chain. This is E4, which is a normal flip flop which takes as a clock the output of the "bit detect" circuit on pin 3, and the "bit value" circuit on pin 5. Sure enough when I connect to 3 I see a nice burst when the tape runs on a boot attempt for a second, but nothing on pin 5. And obviously the output is all zeros so the controller sees no data.

Fine. Let's walk back to the junction between E29 pin 4 (input) and E30 pin 6 (output) (a pair of op amps). No signal on pin E30, pin 6.

Fine. Let's walk back to the junction between E29 (a different op amp) pin 13 (output) and E30 pin 2 (input past a 47k resistor). This is the last spot where the signal from the tape is in "raw" unencoded format, and sure enough it's not only present when the tape runs, but it does appear to be a data packet of sorts. I can see 1,0 transitions, periods of all zeros and periods of all ones. Good. That looks like raw tape data.

Assumption: Looks like the heads, tape, pre-amplifier, gain control, and the peak detection circuitry is working. That's really good, it's a lot of analog stuff and would suck if I had to walk into it. However the portion that is determining if there are 0-1 transitions seems to be broken right at E30.

Now E30 is a standard op-amp, however it's input and output are connected to a quad bilateral switch circuit on E22, which is interesting. It's a 4066 chip, functioning as a sort of integrator to figure out the ratio encoding of what was put on the tape. The 47k resistor between the peak detection thing and the op amp input is there so the results of this integrator can override the signal if it's on a downslope or something.

I'm wondering if either the 47k resistor is bad (doubtful), the 4066, or the op amp. The amp is a good old fashioned lm301, there's also a tiny capacitor used to reduce hysteresis, I'll check that then try replacing both chips next weekend when parts come in. They're pretty simple to remove, but that seeems to be the part where the bits get lost.

Interesting.
 
Well, let's see....

After doing a bit more work with the analyzer I see READ_STROBE and READ_DATA are both sending data from the tape system. So it looks like everything is working in terms of getting data off a tape, but the controller still barfs.

It's *possible* that all of these tapes were degaussed or something in the past, if so it sucks, but it's odd that my XXDP tape from way back when is also a dud. If anyone has a TU58 and wants to try reading a tape pls let me know.

I checked to see if the tape velocity sensors are working at pin 1 of E15, and they are fine. Good solid clock signal. So it's not that either.

Then I took a look to see if it was in the RUN A RUN B timeout circuits. They are supposed to stop the tape if a block header hasn't been seen by the processor in 400ms. Might be that, the CPU sent a pulse but it looked like it didn't send another one and the tape stopped in 500ms. So maybe the tapes are unformatted.

I'm going to put this away for awhile. If anyone with a TU58 can try to read these tapes let me know. I'd at least like to know if there is a baseline format on them.

Thanks!
C
 
Back
Top