Commodore PET 4032/8050/3022

[ScottishColin]

Thanks for checking Alan.

I rather hope this one will involve less soldering than the last but we'll see.

Once this is done, there's the disk drives and printer yet.....

Colin.

Originally Posted by ajgriff

Just for reference the unplugged neutral to live resistance on my 2001-8 is 12Ω so 8 or 9Ω for later machines doesn't seem unreasonable. Also the fuse in mine is 750mA slow blow.

Good luck with this one Colin. The fact that a tinkerer has been at work potentially makes it even more of a challenge but you're definitely the man for the job!

Alan

[SiriusHardware]

OK, it wouldn't be wise to try powering the power block up with an overrrated fuse in it, but the next step (with the right fuse fitted) will be to keep the power block disconnected from the PET mainboard and apply mains power to it, keeping a respectable distance.

If it doesn't blow the fuse again - that most likely happened because of that ridiculous bare wire wrapped around the mains switch and dangling into the earthed metalwork - then we'll need to measure the AC voltages coming out on the low voltage side of the transformer (on what I assume is the J8 connector).

If you have some suitable heatshrink you could make the wiring to the mains switch and fuse holder a little bit safer (less easy to touch) but maybe you'd prefer to do that when you hopefully get to the cosmetic restoration part near the end.

You mentioned the ROMs earlier, I think just leave those extra ones in place until there is a specific reason to remove them, because we know the legs on the ROMs / EPROMs have a bad habit of falling off.

I notice that most of the standard ROMs seem to be soldered in place with the exception of the UD8 'edit' ROM, apparently that is quite common and that is why Daver2 elected to write his famous test code to reside in that ROM location rather than in the boot (Kernel) ROM UD9.

[ScottishColin]

Thanks. I'm on hold until the fuses come then.

As it happens we are away on holiday anyway from Friday so everyone can down tools for a week.

Colin.

[ajgriff]

For reference when Colin returns I've attached a copy of the 4032 schematic and board layout.

Alan

[SiriusHardware]

Oh, NO...

One of the dreaded CRTC-controller variations. Let's hope all it needs is a fuse.

[SiriusHardware]

Quoting myself:

I notice that most of the standard ROMs seem to be soldered in place with the exception of the UD8 'edit' ROM.

It seems from the diagram that the familiar placement and order of the Kernel and Edit ROMs changes a bit in these. The smaller (Edit) PROM resident at Exxx on these ones appears to be UD7, and the Boot (Kernel / Kernal) ROM resident at Fxxx is UD6.

[ScottishColin]

Are there more issues with the display on these machines? Just interested to know what I might find.

Colin.

Originally Posted by SiriusHardware

Oh, NO...

One of the dreaded CRTC-controller variations. Let's hope all it needs is a fuse.

[SiriusHardware]

The CRTC controller potentially adds an extra layer of trouble.

In the 'simpler' machines like your first one the video generation circuit is virtually autonomous and will struggle manfully on, even if the main CPU system is not working / not able to execute code at all the video circuit will still generate the Vdrive and Hdrive signals and render the (uninitialised) contents of the video memory onto the screen.

In the ones which have a CRTC controller, the CTRC is a soft-programmed device. Very early during the initialisation process, it receives instructions from the CPU, which are held in the firmware, about how to behave. In one of these systems, if the CPU is not able to execute code due to a fault in the main CPU system it can not send the relevant setup instructions to the CRTC controller. So usually if the CPU is unable to execute code, you don't get anything on the monitor either.

[ScottishColin]

Ah. Just more complexity then. Another opportunity for failure.

I guess there was a reason they put it in though?

Colin.

[SiriusHardware]

One possible reason was to be able to support several different monitor resolutions or numbers of characters across the screen (40 columns? 80 columns?) just by altering the numbers sent to the CRTC at start-up - by altering the firmware instead of the hardware, in other words.

[ScottishColin]

Which begs the question - can my PET be turned into 80 column or is the CRT circuitry only capable of 40 column?

Colin.

[SiriusHardware]

For the answer to that question, I think you might have to go over to the Dark Side and ask the real PET experts but I think someone did try it over on VCFED and they ended up with a narrower than expected display area due to the limitations of the existing monitor - you can probably find that thread faster than I can.

For the time being, focus on making it work as-built. If you ever go to modify anything away from its original configuration, make sure it works perfectly first.

[ortek_service]

Originally Posted by ScottishColin

Are there more issues with the display on these machines? Just interested to know what I might find.

Colin.

Originally Posted by SiriusHardware

The CRTC controller potentially adds an extra layer of trouble.

In the 'simpler' machines like your first one the video generation circuit is virtually autonomous and will struggle manfully on, even if the main CPU system is not working / not able to execute code at all the video circuit will still generate the Vdrive and Hdrive signals and render the (uninitialised) contents of the video memory onto the screen.

In the ones which have a CRTC controller, the CTRC is a soft-programmed device. Very early during the initialisation process, it receives instructions from the CPU, which are held in the firmware, about how to behave. In one of these systems, if the CPU is not able to execute code due to a fault in the main CPU system it can not send the relevant setup instructions to the CRTC controller. So usually if the CPU is unable to execute code, you don't get anything on the monitor either.

Although I presume Commodore did at least use a fairly-standard 6845 CRTC, rather than a particular in-house custom - as this pre-dated Commodore's VIC (II), custom VIDC's,
And (in the Beeb at least), you get a flashing-cursor if the computer hasn't booted fully, which can be handy for initial fault-finding.

[ScottishColin]

Just interested really. I want to try to leave it as original as possible I think.

Colin.

Originally Posted by SiriusHardware

For the answer to that question, I think you might have to go over to the Dark Side and ask the real PET experts but I think someone did try it over on VCFED and they ended up with a narrower than expected display area due to the limitations of the existing monitor - you can probably find that thread faster than I can.

For the time being, focus on making it work as-built. If you ever go to modify anything away from its original configuration, make sure it works perfectly first.

[SiriusHardware]

Although I presume Commodore did at least use a fairly-standard 6845 CRTC, rather than a particular in-house custom chip...

I'm not sure what the CRTC IC is, that is one detail which is ironically not given on the circuit diagram linked to above.

The fact that the PET monitors don't have a free running horizontal oscillator / output makes this one stage more complicated, because if the CRTC is not generating the vdrive and hdrive signals with the correct polarity and frequency, the monitor just doesn't work. PET monitors are 'activated' by the vdrive and hdrive signals from the computer mainboard.

[ajgriff]

Colin's machine is probably fitted with a 6545 CRTC (first attachment). The 6845 was widely used in various machines from different manufacturers (second attachment). As I understand it the two chips are similar but not identical in every respect.

Alan

[ortek_service]

Originally Posted by ajgriff

Colin's machine is probably fitted with a 6545 CRTC (first attachment). The 6845 was widely used in various machines from different manufacturers (second attachment). As I understand it the two chips are similar but not identical in every respect.

Alan

Thanks for the info. Yes, I now recall there was a 6545 (Which I presume wasn't unique to Commodore / MOS they later bought) - But it seems to be rarely used, with most 65xx systems using the 6845 instead.
However, even the 6845 had many slightly-different versions (like the 65(S)(C)02) with extra enhancements that could mean some systems would only work with particular ones (I think ChrisOddy has had quite a bit of experience of these with Acorn systems / recall trying to produce a comparison spreadsheet of all the various variants).

[ortek_service]

Originally Posted by SiriusHardware

I'm not sure what the CRTC IC is, that is one detail which is ironically not given on the circuit diagram linked to above.

Yes, I'd also looked at those circuits, and found the type number wasn't marked - Although I'd guessed it was a 6845-ish type, from pinout.

Originally Posted by SiriusHardware

>>
The fact that the PET monitors don't have a free running horizontal oscillator / output makes this one stage more complicated, because if the CRTC is not generating the vdrive and hdrive signals with the correct polarity and frequency, the monitor just doesn't work. PET monitors are 'activated' by the vdrive and hdrive signals from the computer mainboard.

Yes, that is rather unusual - Maybe a bit like Vectrex / arcade machines vector rather than raster-scan systems, but hopefully not requiring the same higher speed horizontal frequencies on the vertical deflection.
And does mean trying to connect an external monitor rather more involved.

I wonder if 80col models used a 'better' CRT? As they are monochrome (Green), then there's no shadowmask 'dot-pitch' of colour CRT's to worry about, so mono ones are inherently higher resolution.

But it might be that the internal monitor circuitry is optimised for drive by the different 40 and 80 column waveforms (although frequencies may be the same?) - Maybe comparing the monitor circuitry between 40 & 80 col models, will show any significant differences that would make changing between these more difficult.

[TonyDuell]

There are mask-programmed 6845-like devices, the mask programming effectively determines the contents of all the control registers. Most, if not all, such devices could have 2 different setups programmed, these were selected by what would have been the light pen input on the 6845. Tandy used one (programmed for 64*16 and 80*24) in the Model 4 computer. Just about the only register remaining to change under software control was the cursor position

I never saw a fusible link version in any of the databooks.

I don't think Commodore ever used that sort of thing though.

As for doing an 80 column conversion, there were PET boards that could be hardware configured for either 40 or 80 columns, I guess Commodore wanted to make one PCB for all models, Firstly you need 2K of video RAM, there were spaces for another pair of 2114s. Then you had to modify the addressing, there were quite a few soldered links to change. And I think it's different ROMs too.

[Mark1960]

6545 is very similar to the 6845, but I think it adds the feature of addressing Video ram so that you don’t need to put multiplexors on the address bus to video ram to allow the processor to write to video ram. Its probably in the application notes, I didn’t see it in a quick scan through the datasheet linked by Alan.