• Please review our updated Terms and Rules here

Osborne

Gary C

Veteran Member
Joined
May 26, 2018
Messages
2,256
Location
Lancashire, UK
After a bit of a layoff, its time to get back to some of the machines that have piled up

An Osborne 1 arrived the other day. Nice enough condition

Turned on, no display and lots of smoke :)

RIFA's past their best

20210701_145749.jpg

A 'compact unit' which makes it a bit more interesting to strip, but upside down and PSU is easy enough to extract

20210701_144026.jpg20210701_144625.jpg

By luck I had the right replacement caps in a drawer, and once replaced PSU is fine. Reading the manual, the display won't work and isn't even powered up without a 20 pin edge connector plug in the front and its missing on this machine. Made up a temporary by cutting down a larger plug while I wait for a replacement from RS and push in bits of ribbon cable to short top to bottom,
20210702_141047.jpg

And now the display lights up :)

20210702_141038.jpg

At least the CRT is in good condition, but the scrambled display means something is wrong. A couple of times on switch on, I got a "no disk in drive A" message so it looks as if most of the system is actually working.

Looking at the circuit, the screen memory data is read and latched to use as the address for the character rom. Probing this area showed sensible signals, but looking at the schematic, the data from the character ROM is passed to a shift register, clocked with a signal called DOT_CLOCK and this 8mhz signal is missing. This in turn is developed by a 4 bit shift counter and three out of the four signals are present, but the dot clock signal is just a very noisy 5v signal :)

New 74LS161 on order and hopefully that will sort it.

Interestingly, the CPU seems to have a large plastic lump on it filled with a foam like substance with zenith portable computer systems written on it. An early heat sink ? (actually looks like it would insulate it !)

It also has the double density adapter.
 

Attachments

  • 20210702_141047.jpg
    20210702_141047.jpg
    35 KB · Views: 5
Can someone post, send or point me at some pictures of the video shunt ?

I can find some from direct end on so I can recreate the label, but could do with images of how the rest looks.

I have a black 20 pin edge connector on order and I am thinking I need to cut the lugs off, short the appropriate pins, then maybe epoxy over them, file square and stick on the label but I cant quite see how its meant to look from the pictures I can find (all are installed and end on)

Want to get it looking as original as I can.

(unless someone has a genuine plug ?)

Thanks.
 
Confused

The 74LS161 I thought was the problem is used as a divide by 2 to produce four clocks from the 16MHz clock

The first two clocks (1mhz & 2mhz) are fairly nice square waves

wAAACH5BAEKAAAALAAAAAABAAEAAAICRAEAOw==
wAAACH5BAEKAAAALAAAAAABAAEAAAICRAEAOw==


The 4mhz is getting a bit rounded

wAAACH5BAEKAAAALAAAAAABAAEAAAICRAEAOw==


The 8mhz Dot clock that I though was at fault is awful

wAAACH5BAEKAAAALAAAAAABAAEAAAICRAEAOw==


Now, given my scope is a 100mhz scope its not the digital scope itself messing up the image.

I changed the 74LS161 thinking it was a bit slow, but the new one is exactly the same.

Its interesting to note that the clock input to this chip also looks appalling, but it manages to produce the other slower clocks seems to indicate that its 'good enough' to work (!).

Confused and need to do a lot more testing :(
 
You get a similar thing with other computers, you get a poor waveform going in, and it takes multiple divide by two toggles to ‘square it up’.

I don’t know what happened to your trace images though?

A 100 millihertz scope sounds very poor though!

Dave
 
MHz !

;)

Are the images not showing ?

The output from pin 14 of the 74LS161 is a DC 2V with a ~100mV AC 8 MHz signal on it. It then goes to an inverter and comes out upside down but still as bad.

Must be how I am measuring it as the device should manage a 30MHz clock from the datasheet and its only 16MHz

The circuit is very simple.

Osborne clock circuit.jpg - Click image for larger version  Name:	Osborne clock circuit.jpg Views:	0 Size:	191.7 KB ID:	1216167

I wonder if my scope is messing it up as I seem to be barking up the wrong tree. Time to crack out the traditional scope.
 
oooh another Osborne repair :) I picked one up a few months ago locally, along with another donor machine for its monitor and video plug. I replaced those RIFA caps preemptively but mine actually didn't look too bad, there were some fine cracks but epoxy was fully intact. Nice that you have the double density card. I'd like to see your solution for the video plug once you have that sorted out -- great that you have the workaround for now. I've attached two pics of the video plug for reference.
 

Attachments

  • osborne_video_plug.jpg
    osborne_video_plug.jpg
    44.3 KB · Views: 4
  • my shunt.jpg
    my shunt.jpg
    246.4 KB · Views: 4
Cheers, so the label is stuck on a standard black edge connector ?, that should be fairly easy to recreate.

IF i get it working :)
 
Yep looks that way. Please share what you cook up for for a video plug.

Have you tried hitting enter after power on? That’s a good way to see if your motherboard is working. If it spins the A drive after hitting enter, that is a good sign! I didn’t realize this at first when I was working blind with a monitor that didn’t work but fortunately the motherboard did.
 
Ooooh I am stupid.

Put the Hameg on it, and the signal is normal.

While my new digital scope is great, with a nominal input impedance of 1MΩ and pF capacitance, I think the leads are not quite up to it capacitance wise. Put them on x10 and the signal looks fine on this unit as well. Didn't often have to think about stuff like this with the equipment I used in a past life at work.

Ah, well. Trouble shooting continues. The processor is running but I note the reset PB operates through NMI rather than pulling a real CPU reset. With the keyboard in, I get no response from the drives though I do sometimes get one drive spinning, and I have even had text on the screen suggesting a bad connection or duff logic chip somewhere.

Time to get methodical.
 
Last edited:
Got the black 20 way edge connector for the video shunt but as its a solder to board version, while it will work, it won't quite look like the original which appears to be an IDC one (and must have the ribbon displaced by one) so I have found an IDC version on the bay which looks more like it.

https://www.ebay.co.uk/itm/292816837...8AAOSwEcxb7A8s

Still not got the system up and running, but some checks on the CPU shows that address lines A8 & A12 are not changing at all. seems odd (especially A8, might be just what its addressing so need to work out the map. Different CPU tried but exactly the same symptoms.

What I really want is one of those 3M or Pomona IC test clips so I can connect the logic analyser but they are £50 for a bit of plastic !!!
 
Last edited:
Nice find on that IDC connector - looks correct to me. You can see on the shunt that the leftmost ribbon wire is “missing” which supports that the ribbon wires are displaced by one.
 
So the diagnosis continues.

The processor appears to be reading the ROM ok but then crashes.

I believe when powered up, the ROM is from x0000 to x1FFF then memory mapped IO (!) up to x3FFF then its 48K of RAM up to xFFFF. The lower 16K of ram is bank switched using the Z80 IO command.

So I have written a bit of noddy code to write x01010101 and check, then write x10101010 and check, if either dont read back ok, I then jump to a section to write to port xFF seven times so I can see pin 20 IORQ toggle 7 times or if its ok, toggle pin 20 three times.

When I run this from ROM, I get seven IORQ toggles continuously which seems to indicate that I have memory access error.
Interestingly, with this ROM in, the screen consistently displays the same character.

As the CPU can read the program in the ROM, the address and data lines seem ok.

Also, as it seems as if every location is faulty (or my program !) then its probably something in the select logic.

Think the first thing I need to do is take out the error signalling and only leave the pass signalling to see if I ever get a successful read then start looking at the decoders and buffers.

I will also modify the program at some point to use pin 20 to indicate the location of the errors (though I do need to find out which port the osborne uses to bank switch just in case I am interfering with it)

Anyone else diagnosed an Osborne ?


loop1: LD hl,$4000
LD d,$00

loop2: LD d,%01010101
LD (hl),d
LD a,(hl)
cp d
jp nz, error
LD d,%10101010
LD (hl),d
LD a,(hl)
cp d
jp nz, error
jp pass
return: INC hl
ld a,$ff
cp h
JP NZ, loop2
ld a,$ff
cp l
JP nZ, loop2
JP loop1

error: out ($FF),a
out ($FF),a
out ($FF),a
out ($FF),a
out ($FF),a
out ($FF),a
out ($FF),a
jp return
pass: out ($FF),a
out ($FF),a
out ($FF),a
nop
nop
nop
nop
jp return
 
So moving on slowly. I have run a small bit of code to write 11111111 and test each bit then write 00000000 and test each bit and use the IORQ line to signal the bit(s) in error.

I have only run this on one location (F000) and its showing that bit 2 is stuck at 1, which is a start so I will need remove and replace all the ram on line DI3/DO3 which are chips UA25 to UD25 which is only 4 chips and see if that improves the test. Given the EPROM can be read fine, the data lines themselves must be working suggesting its a RAM chip.

But first, I think I will change the memory location I am testing to several random locations and see if I can see a pattern.
 
Last edited:
I appreciate your debugging on this! Can you just piggyback some RAM chips on those 4 suspect chips? Would that be a reasonable test?
 
I have thought about the piggyback, but I am going to bite the bullet and just replace.

I have scanned locations C000 to FFFF (its better doing a row of chips at once, otherwise multiple faults in different banks are hard to see on the scope) and it was showing bit 2 as stuck and I have replaced that, and its now clear, but is also showing an occasional bit 7 problem even though I have replaced that one too. I have socketed the two locations because the only 4116's I had are 200ns and I think the Osborne needs 150ns (thats whats fitted) and my actual replacements are still somewhere between here and Poland.

Just looked at the manual and it seems to state that the access time is 250ns, so I need to think about this. It does look as if I have fixed the first memory fault so need to extend the search.
 
Last edited:
I think your right. The manual states 250ns for RAM and 350ns for ROM.

150ns though its what's fitted. Replacing bit 2's chip in bank 4 has stopped that throwing an error, but I occasionally get a bit 7 error showing, but that might be the program.

I will move on to test the next four rows, but I think I will wait for the equivalent chips to turn up (should have been today)
 
Ah well :)

I am thinking of whipping up a 4116 tester. Have bought a largish batch of chips so I have a bit of stock.

Should be an easy arduino project if it wasn't for the voltages needed, though I could use an old PC supply or repeat the -5 voltage trick from the osborne.
 
Back
Top