• Please review our updated Terms and Rules here

What did I do to my PDP-8 today.

Now that the PC04 is working I can load FOCAL and the 4USER Patch and run it with 2 Teletype ASR33. It's fun to do so. But the main Teletype did not load any short FOCAL program.
I noticed, the Reader Run relais is closed all the time. So it loads much too fast.
I opend the relais and inserted a new reed glass into the coil. So this is now working again.
Then I also noticed, the READER is not reliable enought. About one or two Bytes of 100 Bytes are wrong. So I began adjusting.... Long story, no real progress.

There are so many adjustment points:
Trip magnet 2x
Reader Magnet 1x
Reader 4x (or even more)

Is there a recipe how to find the correct adjustment for an ASR-33?
 
gnupublic said:
Is there a recipe how to find the correct adjustment for an ASR-33?
Click to expand...
I maintain 3 33's. They all act up in unique ways. I go with figure out what the exact symptom is such as is a single bit error, multiple etc and then try to figure out what could be causing it. Last problem reading tapes was really the contact block in the keyboard broke so was moving around and contacts were occasionally closing while tape was reading.

gnupublic said:
Now that the PC04 is working I can load FOCAL and the 4USER Patch and run it with 2 Teletype ASR33. It's fun to do so. But the main Teletype did not load any short FOCAL program.
Click to expand...
Early FOCAL's have a design problem. They run the teletype off interrupt and don't drop reader run when the small buffer fills they just generate the overflow error code. Since each character is echoed if the teletype sends data faster than the serial echos back to the teletype you will get the overrun after a while loading a program. With the earlier 8's where baud is controlled by I pot I just adjusted it to make 8 send faster than Teletype. I assume your using later machine with crystal clock so don't have that option. Can't remember if baud rate I adjusted was slow or teletype was faster than exact 110 baud when I adjusted.
 
djg said:
Early FOCAL's have a design problem. They run the teletype off interrupt and don't drop reader run when the small buffer fills they just generate the overflow error code.
Click to expand...
Turns out that's kind of all you can do on the early machines. If you try to return after a KRS, the flag is still set and the interrupt is re-triggered. If you KRB, then reader run is set. There isn't a KIE, to allow you to mask the interrupt. So input continues until the buffer overflows.

(I actually do know a way to work around the problem, but if would take some explaining, and there's no version of FOCAL that does it, to my knowledge.)
 
Last edited:
djg said:
Last problem reading tapes was really the contact block in the keyboard broke so was moving around and contacts were occasionally closing while tape was reading.
Click to expand...
Oh, this is very unexpected.
djg said:
Can't remember if baud rate I adjusted was slow or teletype was faster than exact 110 baud when I adjusted.
Click to expand...
Maybe it's worth to think about an own "Reader Run generator" inside the ASR-33 to slow down the reader to a reasonable frequence.

For testing the READER with FOCAL I made a endless papertape with "@". So when I read this in there ist mostly one M instead of an @ in a line. I will do this with other patterns next weekend.
 
I repaired a M8300. There were two DEC 8266 and one 74151 broken. Repairing these modules helps understanding the 8.
 
Two days ago I was able to fix the TC08 of one of my pdp8/e! Very happy about this, and finally the M228 card was the reason. That's not too easy to find, because it could be the M8350/60, TC08 or the drive itself. It took me some times. I changed all FlipChips of a second TC08, but nothing changed. I knew, I only had one M228 working. So I took a bad one and got out all IC's. With 21 Sockets I put in only tested IC's, but had no Sn7453. Found one old equivalent from TESLA and the TC08 works again!
(So now I'm driving a TESLA....)

In the last weeks I also was able to bring three TU-56 back to work. They had different defects, and my sacrificial drive loses a head and many Flip Chips. I still have some drives to repair, this will never end....
I made some observations regarding the drives using 110V or 220V.

this drive is clearly a 220V drive. You also see the 1.5A fuse on the right side, a sign for 220V. Also the plugs are 220V Types. I got this working today and I'm happy about that.

1717268388890.jpeg

The next drive ia also 220V but has the wrong power plugs. The fuse ans sticker are correct.

1717268556510.jpeg

The next drive is 110V, plugs and fuse are OK for this, even the small sticker:

1717268711810.jpeg

But with this drive the sticker is wrong. Plug and fuse say 110V.
I'm always starting the powersupplys with low voltage and bring it slowly up to 110V, then I measure the voltage of the capacitors. Glad that I did not turned to 220V....

1717268929873.jpeg

So regardless what plugs/fuses there are, I start with 110V.
 
I've been restoring an RX02 drive on my PDP8e, which had quite a few issued when I started. In order to align the drive (track alignment, head lifter adjustment, track 0 sensor, and index sensor), I wrote a pretty useful program, called RX8ALIGN. This program lets you select a drive, seek any track, load or unload the head, and cause continuous seeks back and forth between any two tracks. You can find both the source and the binary here.

I have also included reasonably detailed alignment instructions in the block comments at the beginning of the source code, for both the RX01 and RX02 drives.

<><>

On a related note, I noticed that when I wrote to a diskette with drive 1, and then looked at the data coming off of the drive, it looked perfect. But when I removed the diskette and put it back, the amplitude of the data was no longer consistent across the track. Stranger still, the high- and low-amplitude position on the track changed every time the diskette was removed and reinserted. Finally it dawned on me: the spindle hub was wobbling! The bearing is ok - it is as though either the shaft or the hub itself was bent. But these are big, meaty metal parts, and I can't imagine how they could get bent. I kinda suspect the hole in the hub was not drilled straight in the first place. Obviously, this drive will never align perfectly.

Has anyone ever seen this before? And the big ask: does anyone have a "trash" RX01 or RX02 drive they'd be willing to part with, where I might find a straight spindle hub?
 
mfeberhard said:
You can find both the source and the binary here.
Click to expand...
I love to look at other peoples code. Thanks for sharing! I will probably use this at some point.
mfeberhard said:
Has anyone ever seen this before? And the big ask: does anyone have a "trash" RX01 or RX02 drive they'd be willing to part with, where I might find a straight spindle hub?
Click to expand...
My first thought is that it probably is not centering the diskette when it clamps. So each time the centering is different. I remember seeing diskettes where the hole is no longer round or the edges are deformed. Just some other possibilities.

I don't have a trash drive at this time and I think there were a couple of variations of the drive assembly used. I think the ones I have were Shugart units DEC purchased without electronics. If I am correct you might be able to pick up a Shugart and get the parts you need.
 
mfeberhard said:
On a related note, I noticed that when I wrote to a diskette with drive 1, and then looked at the data coming off of the drive, it looked perfect. But when I removed the diskette and put it back, the amplitude of the data was no longer consistent across the track. Stranger still, the high- and low-amplitude position on the track changed every time the diskette was removed and reinserted. Finally it dawned on me: the spindle hub was wobbling! The bearing is ok - it is as though either the shaft or the hub itself was bent. But these are big, meaty metal parts, and I can't imagine how they could get bent. I kinda suspect the hole in the hub was not drilled straight in the first place. Obviously, this drive will never align perfectly.
Click to expand...
What you're describing is called "non-repetitive runout". On larger drives like the RK07, back in the day you could purchase replacement spindles as spares or FCO kits from DEC. Back then, if it was an RXxx floppy drive, it was field replacement of the entire unit drive. You could DECmailer it or just buy a replacement.

These days, your best bet is indeed getting a drive carcass from someone else and transferring the good parts from your drive.
 
Terri Kennedy said:
What you're describing is called "non-repetitive runout". On larger drives like the RK07, back in the day you could purchase replacement spindles as spares or FCO kits from DEC. Back then, if it was an RXxx floppy drive, it was field replacement of the entire unit drive. You could DECmailer it or just buy a replacement.

These days, your best bet is indeed getting a drive carcass from someone else and transferring the good parts from your drive.
Click to expand...
My first assumption was that the disk was not centering properly, and so I spent some time and effort studying how the cone centers the disk onto the spindle hub. The short answer is that the disk is centered on the hub correctly, even on a wobbling spindle hub.

I used a straightedge as a gauge, and I could see the spindle itself wobbling as it rotates - about maybe 0.01" (half a track width) in the plane of the disk, and a similar amount up and down, exactly as though the shaft was bent, or (maybe more likely) the hole through the spindle hub was drilled a bit crookedly.

These drives are not the same as Shugart drives. My best guess is they are stripped down CalComp drives (or just a low-cost OEM assembly that they made), though I have not yet found documentation for a CalComp drive that's exactly the same. As I've noted elsewhere, in the process of studying how the disk became centered on the hub, I learned that these spindles work differently than any other drive that I've seen (I've restored quite a few floppy drives, including Shugart, Siemens, PerSci, Pertec, etc.) For every other drive I've seen, the cone is spring-loaded outward, so that when you close the drive door, you are forcing the cone to close up slightly in order to fit into the hole in the hub. This ensures a centered disk. On the DEC drives, the cone is loose in the hub until the door is cloed all the way. Closing the door pushes a plastic disk (a little bigger than a quarter) downward, causing the cone to expand to fill the hole in the hub, and thereby centering the disk in the process. I have never had my hands on a CalComp drive, and I'd be curious if the cones work this way. (I bet they do.)

Also, I know what you mean about DEC's repair philosophy on these RX01 and RX02 drives. their repair procedures say to just replace the actual drive if it's defective.

Related: I can't imagine how this hub could have become bent over time or even from abuse. The shaft is quite thick and pretty short. The hub is a big hunk of metal. I can only assume that this drive has always wobbled. And it mostly works that way, though sometimes I have to take the disk out and put it back in again, rolling the dice for a better alignment wrt the wobbling hub.

I am indeed on the hunt for a drive carcass with a decent spindle and shaft. (Don't need any other parts...) If anyone has a drive they'd like to pass my way, I'd appreciate it.

And finally, if you ever want to align your RX01 or RX02 drives, my program is pretty handy :)

Martin
 
I was able to bring three further TU-56 transports back to work. Now I have eight of them working. Next is to make something from the sacrificial drive rest. There was one bad motor and one bad head and a missing M531. I plan to make a mono drive with the wheels on the right side in contrast to the original half drives from DEC.

Reading the maintenance manual, DEC writes in 1.3:

If more than one transport is connected to a single controller, only the first transport of the "daisy chain" need contain the G888 modules.

What does this mean for TD8E? Is it possible to "daisy chain" two mono drives with only one set of G888 cards with one TD8E controller?
Or is it even possible to do so with normal drives, to be only able to use ID 0 and 1? Is the limitation of two wheels per TD8E in hardware or in software (for OS/8)?

Have fun
Volker
 
Haven't posted here for a while. Visiting Vince again I watched him play with his new 3D printer. And what he printed was the parts for the SBC6120 front panel stuff. The Bambu Labs P1S does a great job.

While parts are printing we worked on the PDP-12 and his MM8I memory expansion. He has a version that uses delay lines for the timing and that works great, but the delay lines are not easy to obtain and expensive. So the delay line board has been replaced by one-shots and at the moment is not working all that well. So I am assisting with figuring out the problem.

We scoped the CMOS ram chip and all those signals looked ok. We then looked at STROBE and MEMDONE. STROBE tells the CPU that the data is stable and can be latched into the memory buffer register. MEMDONE indicates to the CPU that the write portion of the memory cycle is complete. The MEMDONE signal looked great. The STROBE did not. STROBE is generated by a 74HCT123 which was then inverted by a 7401 to drive the open collector line. Turned out the 7401 drive is not sufficient and we substituted a 7439 which cured the problem. Vince has those and the pin compatible S8881 which was what DEC used with their own label. Neither are good choices due to scarcity. Not sure what the final solution will be. Perhaps simply changing that one part to a 7438.

Next on the list is to try my RK8E controller with Georges RK05 emulator. Vince's controller passes the diskless diagnostics but does not seem to select the drive. My controller has never been in a machine since I got it so it might not even pass the diskless test.

More news later!
 
DougIngraham said:
...

Next on the list is to try my RK8E controller with Georges RK05 emulator. Vince's controller passes the diskless diagnostics but does not seem to select the drive. My controller has never been in a machine since I got it so it might not even pass the diskless test.

More news later!
Click to expand...
I am curious how Vince's controller "does not seem to select the drive".
What exact test are you running and what exact symptom are you seeing?
How do you observe the "drive select" no happening?
Have you got the M930GW RK05 Bus Terminator board plugged in?
 
You must log in or register to reply here.
Back
Top