• Please review our updated Terms and Rules here

Reproduction Osborne 1 DD upgrade card

Thanks for that!

The math there doesn't really check out, but I suppose we are measuring in-circuit, and with a powered-off power supply. Let us know when you get the power supply fixed, and I'll pull out my OCC1 and rig everything up again.
 
Yes, there are other paths in-circuit, so having the resistance in-circuit may be helpful. The first thing it will help provide is a wholistic perspective on that pot. The second it a baseline to measure it while adjusting, and the result suggest measuring from the ground side to the wiper to set it, then make sure all the same resistances are on your circuit.

Also, those measurements are with the PCB disconnected from the motherboard and drives.
 
272KHz according to the cheap aliexpress scope I measured it with- it's not super-accurate, but I imagine it would be within 1% or so - Also I don't know what the frequency is supposed to be, so as my Osborne 1 isn't working at the moment, I can't say with any certainty that this is what it should be. 271.5 is the closest I can measure it by eye, +/- 1.5KHz.

Forgive me for stating the obvious, that the point you mentioned isn't attached to the potentiometer at all in the above schematic. Is it supposed to be?

Anyway, if you need anything more accurate than this, send me a list of pins and I'll stick the logic analyzer on it and record it. Though It may be pertinent to mention that I'm measuring the board without any drives plugged in.

Thanks
David
 
Yep, just desk wired it all. Stays on 270Khz or thereabouts while loading from a Double Density disk, so that seems to be what to expect on U6 Pin 5.

Regards
David
 
Hi NRoach44,

Hoping you're still around and working on this - If you have any space PCBs I'd be happy to populate one and work on it from this end also?

David
 
Sorry about the delays with this - had a ton of stuff going on, some good, some bad.

I was going to say "I would send you a board but it might be more expensive than just ordering another set" but then I noticed your location. I'll shoot you a message direct.

I've just gotten my OCC1's keyboard fixed (hot air gun to the back plate at 400℃for tens of seconds until the stuck key cleared) and made a serial cable, so now the limitation of 90K disks is starting to get to me :^)

I just need to figure out what the frequency part of the board is doing (something like a PLL..?) and then maybe I can figure out why I can't get the frequency high enough.

I've updated the KiCAD repo so it should reflect what I send to JLC: https://gitlab.com/NRoach44/occ1-dd-upgrade
 
cj7hawk said:
OK, so tell me more about this "Hot Air Fix" -?

What is the idea and how does it work? It sounds like something one of my keyboards might need.
Click to expand...
The keyboard is just a membrane (two layers of plastic with conductive traces) that have little air pockets where the keys are. The issues with stuck keys seem to be that the two sheets of plastic get stuck together. The original solution was to remove the key mechanism (which is "melted" in), but Gary found that you just blast heat at the back side of the key and mash it until it stops being stuck.

In my case, I had 7 & 8 stuck on the main block, and it took probably a few minutes before it was fixed. I've since had to do it the day after I first fixed it, but it only took a few seconds of heat.
 
I took mine completely to bits and had bent a corner of the membrane. To ease it back straight I ran my hot air gun over it and noticed that one of the membrane key cells seemed to pop back up. Did it over all the membrane and all the keys started to work.

I didn't discover that you could just apply the heat to the back though, that was inspired and makes it so much easier.

Didn't expect it to last but it has so far.

Not sure if its a 'permanent' fix, but it seems to have had some success so far.
 
Oh, well I only did that because I thought that's what you did! Unfortunately it now means that I can tell Drive B: isn't finding track 0, but maybe that's the nudge I need to Gotek it.
 
Will do, as soon as I get time to myself and a moment with the printer.

Toooooo many projects backing up
 
Okay, quick status update:

Boring life events have slowed down!
I've fixed the keyboard (as above) and figured out why - the original cable, right at the fold, presses on the 6&7 or 7&8 keys when the unit is packed up. It'd work all fine and dandy until I packed it up and moved it. Swapped the cable out with a bare IDC cable (original is folded up in the keyboard) and so far so good.
I have tweaked the floppy adaptor to expose the 12V power line, and pass through the side select (since it's wired to the controller): https://gitlab.com/NRoach44/Osborne1FloppyAdapter
B drive has some issues, and I've gotten very close to calibrating it but not completely. No need now that the adaptors have arrived.
I've also found some more schematics and troubleshooting info in Sam's guide ( https://archive.org/details/sams-computerfacts-osborne-occ1 ) which should help.

The upgrade board still behaves as before, I can see it being toggled on and off when it's booting, so fingers crossed I can progress again soon.
 
I've cross checked the Sam's schematics against mine, and found one mistake on mine, and one on theirs (compared to the official OCC schematics). Mine was of no functional consequence:
  • RN5's LATE and CLK was swapped (mine)
  • No pull up on EARLY (Sam's)
I also checked the board view to make sure the board was correct, and to the best of my knowledge it's all correct. so the fault should lie with my assembly or parts. Since I made one substitution on the logic, and made the resistor networks manually, it's not unlikely.

Fingers crossed David's able to make a working one with the boards he should have today or tomorrow.
 
I'm not expecting it to be a quick fix - it's already had eyes on it and I'm more hoping that a fresh set of eyes can help. :) It's a nice PCB.

I was examining the circuit today. The Write seems pretty simple and restructures the pulses, but shouldn't affect reading a disk. It gets some sign

The read has two circuits - one that drives for normal density and one that drives for DD. Both are timed through the pulses generated by the LS221, which uses 2% tolerance resistors and seems designed for this application. Selection is by DDEN ( Double Density Enable ) and are then NANDed together to recover the clock synchronisation, so that the timing affects whether the bit is read as a zero or a one.

If anywhere, that's where I think the likely issue will be. Also there's the oscillator circuit which follows from the synchronisation section, and should put out around 270 KHz to recover the final timing for the clock... It's a bit weird and I think the clock is jittered by the edges of the read line to adjust timing - I may need to use the logic analyzer to follow that circuit more closely once I get the waveforms to look correct.

I got through about 50% population of the PCB this evening... I am missing a suitable sized 10K pot which I think is used to find tune the timing between a 1 and a 0 ( eg, the arbitrary 75% cutoff point ).

I may need to generate some signals for 1 and 0 to fine tune it. So I have to find some more data on what kind of MFM format the Osborne took. Or I might just figure it out from the oscilloscope timing since I haven't worked with disk drive signals at this level before.

I'll study the circuit a little more - but I'll build it and have a look at the signals first. As long as EBAY don't screw it up, I have another Osborne main PCB coming that I can check in real time against a working unit in parallel.

David.
 
At the moment I've booted the system with the board plugged in only on the 34-pin main cable, the additional 10 / 8 pin that goes across the board isn't connected. I've pulled DDEN to GND and that seems to do the trick.

The system will boot normal SD disks like this, so I've booted the diagnostics disk and using the drive exerciser (option P) to keep a drive with a DD disk spinning. That ~should emulate enough to allow diagnosis, since everything except the output is hooked up (and it doesn't replace the existing read signal).


So far my analogue rail is a bit low (8.4V vs 8.7V), and the output of U1/11 isn't matching the Sam's guide, but that's as far as I've gotten, so going to check all of the waveforms. I was able to get the trimmer to the right voltage, which is promising.
 
Okay, two main suspects:
I think the low analogue voltage is the cause for my frequency woes, so that'll be the first thing I fix / workaround (lab power supply here I come)
Since the two flip flops in U5 (74LS107AN) appear to super-impose one another, I'm going to assume that the output of one of them being inverted is due to a desync because of the frequency. Either that or it's because I used 74HC107E instead.
 
It works, in read at least. I haven't tested write yet. Getting accurate components is critical, and I substituted a lot. Plenty of weird values in the schematics.

Anyway, the pulses for the incoming read need to be exact, and the READ DATA needs a 400ns pulse - Trimpots on the R2 might be needed. I think the other output on U1 is 1uS to 1.7 uS ( either one or the other, it switches ) - THIS should be also made perfect. I'll recheck these values later - well, the 400ns to the READ line IS critical.

Then the trimpot is adjusted until Pin7 on U9 rubber-bands to 1 MHz exactly. It will have a small range over which it rubber bands - seek the middle of this range. Then it should be calibrated for whatever components you used.

Finally, You will need to make board supports out of something to hold the board in place, just like the original.

But so far, NRoach44's design seems to work OK for read at least, and I'll try write when I have a little more time. And a keyboard attached.

one more note - on the Gotek, it switches into WRITE mode for some reason when you move the pot to one of the extremes. Clockwise I think. It WILL corrupt the disk image, but it creates a good calibration signal at the same time, which gives a clear data source, and is a good signal for adjusting the 1 MHz clock. It also means the disk will keep booting due to the corruption, so you can keep loading the track over and over, which helps while you are calibrating it. Also, Pin 7 on mine shows 962 KHz when the disk is not running. 1 MHz is only when the disk is accessing.

David.
 
You must log in or register to reply here.
Back
Top