Commodore PET 2001

[ajgriff]

Originally Posted by Slothie

How did you do that?

Sharp craft knife and fine grades of carbide paper. I've lost some of the surface texture and I didn't do too much with the smaller marks for fear of grazing the silver finish to the legend strip. All looks quite acceptable at normal viewing distances.

Alan

[SiriusHardware]

It's now clear what it is that you've been doing for all the years you weren't previously playing with micros.

On the basis of that result, you can now expect all my cosmetically imperfect items... I have a few... to turn up on your doorstep some time soon.

Seriously, good job, I have no idea how you did that.

[julie_m]

Originally Posted by SiriusHardware

If you go down the EPROM emulator route make sure you find one which can specifically emulate the MOS 6540 PROM because they have an unusual (I would go so far as to say - freakish) pinout. It could not be more different to the 2716, 2732, 2764, 27128 types which a typical EPROM emulator will emulate.

Quite sensible from Commodore's point of view, though! You are always going to need a bunch of primitive "glue logic" -- AND / OR / NOT / EOR gates, flip-flops and the like -- to derive the output enable line for each memory IC from the higher-order address lines and the CPU clock, and you have your own chip factory; so why not bake in some general-purpose logic gates to provide multiple active-high and active-low select lines, right on the same die as the PROM? That way, you can easily pick up on various combinations of zeros and ones on the address bus to select the device -- for instance, if your chip has 4K of memory and you want it to respond to addresses 0800-0BFF then you need to connect A15, A14 and A12 to select inputs wanting a zero, A13 to a select input wanting a one, and just tie any redundant select inputs to a suitable power rail. If you wanted 0C00 to 0FFF, then A12 would need to go to an active-high select input instead.

Of course, it risks tying you to one manufacturer; but that's less of an issue when that manufacturer is also you.

(It's interesting to compare this idea of distributing the glue logic around the peripheral devices with the approach taken in the UK; which was to concentrate all the glue logic into a single device which could be placed close to the microprocessor and its bus signals .....)

[Slothie]

I've sent off my 6540 adapter pcb to JLCPCB since they offered me a cheap deal and I will almost certainly want one or 2 for my PET when the time comes. If AJ wants to try one when they arrive I might be able to post a PCB, Although by then he will probably have his ROMulator.

[ajgriff]

Just a couple more before and after datasette shots and then I promise I'll get back to trying to fix the PET itself.

Alan

[Slothie]

You seem to be restoring the texture nicely!

[ajgriff]

Very much 'faux' texture. Since the moulding material (epoxy) inevitably has a smooth finish the first step is to rub down the original texture to a similarly smooth surface. The area is then acrylic sprayed with the final coat being applied at a distance to obtain a slightly rough surface. This would normally be undesirable but in this instance it's a useful way to trick the eye (or camera lens).

By the way, thank you for the offer of a PCB to make a ROM adapter. I'll see how it goes with the emulator and take it from there.

Alan

[ajgriff]

I hope the attached photo illustrates more clearly what I mean about paint spraying. It's only the raised area of the case moulding around where the keys would be that has been sprayed and therefore includes the scar repairs as well as the corner reconstruction. The second shot shows an internal view of the corner repair.

Alan

PS Emulator report to follow.

[ajgriff]

Duly installed the emulator and to begin with didn't have much luck but it did help to point me in the direction of one of the faulty components. I'll provide more detail when I figure out exactly what was going on. For now the PET is up and running again using the emulator's 32k of RAM and one of its ROM images (see photo).

Alan

[Timbucus]

That is good news that it is running again.

[Slothie]

Excellent!

[ajgriff]

A few words of explanation. After installation of the emulator and with it set to use the on board RAM/ROM the PET displayed absolutely nothing. After some experimentation with jumper settings I managed to get a partially corrupted VOSSI diagnostic screen which stated that the IEEE test had failed. This pointed towards a problem with PIA2 which I then removed. Now I got a decidedly non-random character screen. A fair bit of messing around ensued and I eventually removed the four good (supposedly) 6550 system RAM chips from the main board and all was well in the world. It would seem that PIA2 and a RAM chip had failed.

Avid readers may recall that in a previous fault analysis (post #126) I stated that:

Originally Posted by ajgriff

System RAM - substituting four known good 6550s for the 8k card makes no difference.

PIAs - removing these two chips makes no difference.

The reality is of course that there was originally only one fault (PIA2) and I inadvertently introduced a second fault in the form of a RAM chip which had previously tested as working such that removing the PIAs produced no change to the random character screen problem. Lesson learnt.

So far I'm very happy with the emulator but will report fully when I've had more of a chance to play with it.

In the meantime I need to source a replacement 6520 PIA. Does anyone know if the Rockwell R6520 or the Synertec SY6520 would work in the PET 2001-8?

Alan

[Slothie]

Its also noteworthy that the 6521 is an equivalent to the 6520, the 6821/6521 was introduced to correct a rare 'bug' in the design. The Western Digital WDC W65C21S claims to be a "High speed/Low power replacement for Motorola / Rockwell / AMI / *MOS Technology / MOSTEK / HITACHI / ST Microelectronics / GTE / CMD 6520, 6521, 6820, 6821 PIA’s" and is in current production.
I can't imagine a Rockwell of Synertec 6520 wouldn't work as the PET doesn't exactly highly stress the PIAs.

[ortek_service]

Originally Posted by ajgriff

>>
>>
In the meantime I need to source a replacement 6520 PIA. Does anyone know if the Rockwell R6520 or the Synertec SY6520 would work in the PET 2001-8?

Alan

I can't see why any type of 6520 shouldn't work, as long as it's max. clock speed is rated high enough (Although I think PET only used lowest 1MHz versions, rather than 'A' / -2 suffix 2MHz etc. parts).

And it's also possible that a 6522 VIA may also work, if these are pin-compatible (will need to check), as the 6522 is really an enhanced 6520 with extra registers / timers etc.
- With the 6520 falling out of use and the 6522 much more widely used (even if the extra timers etc. in it were not actually required)

It seems the PET was the main user of the 6520, with commodore moving onto the 6522 in the VIC-20, before using their own 'custom' 6526 CIA for the C64 as they were rather-keen in keeping the MOS fab busy.

[ajgriff]

Thanks for that Slothie and Ortek. It looks like there might a few possibilities. I'll investigate further.

Alan

[ajgriff]

It looks like the 6522 VIA isn't pin compatible and wouldn't be a drop in replacement. The WDC W65C21N PIA looks as if it might be a good bet and is available here at a reasonable price:

https://www.ebay.co.uk/itm/Western-D...wAAOSwjVhf19ZU

I'm tempted to go for it.

Alan

[Slothie]

Originally Posted by ortek_service

And it's also possible that a 6522 VIA may also work, if these are pin-compatible (will need to check), as the 6522 is really an enhanced 6520 with extra registers / timers etc.
- With the 6520 falling out of use and the 6522 much more widely used (even if the extra timers etc. in it were not actually required)

It seems the PET was the main user of the 6520, with commodore moving onto the 6522 in the VIC-20, before using their own 'custom' 6526 CIA for the C64 as they were rather-keen in keeping the MOS fab busy.

The 6522 is a completely different chip. The 6522 has 16 registers, the 6520/21 only 4. The pinout is different. The 6520 and 6521 are equivelents however.
The PET has 6520's and a 6522 for the user port.

[ortek_service]

Originally Posted by Slothie

The 6522 is a completely different chip. The 6522 has 16 registers, the 6520/21 only 4. The pinout is different. The 6520 and 6521 are equivelents however.
The PET has 6520's and a 6522 for the user port.

Thanks for the more detailed info.
Yes, I did say the 6522 was a kind of 'enhanced' 6520 - and do recall the 16 registers from my Beeb-Programming days. But I wondered if 4 of these were the same as those on the 6520 (being as it seems the INS8154 PIA can be replaced with 6522, as share they same registers, even if H/W needs changing)

Just looking at the pinouts for the 6520 & 6522, it seems most pins are the same, but 6pins have changed function.
With A0-A1 (for 4 registers) becoming RS0-RS3 (for 16 registers) - but on different pin-positions, although could be re-mapped.
However, main issue might be CS0 (if used) is no longer on 6522 and there's only one nIRQ rather than 6520's nIRQA & nIRQB.

6521's don't seem to be that common, but 6821's should be (I'm sure I've got some, and believe I used one on a project at Uni back in the 90's)

[ajgriff]

I've ordered the WDC W65C21N PIA mentioned above. Fingers crossed.

Alan

[Slothie]

Originally Posted by ajgriff

I've ordered the WDC W65C21N PIA mentioned above. Fingers crossed.

Alan

Its a good one to try, much cheaper that an old one! If it doesn't work in the PET for some reason I'd be willing to buy it off you to try out on my MK14....