Philips PM5544 electronic testcard

[ppppenguin]

I've biased the comparator inputs and it's looking a lot better. There is still some raggedness and there's also an interlace problem.

[ppppenguin]

There's a field timing adjustment that affects interlace. It's also meant to adjust the top and bottom castellations but it doesn't. The circle looks a bit stepped at the top and bottom because only one of the two fields is captured in the photo.

In this photo I've switched the text to BBC2 and also turned on the chroma test panels.

[Mike Watterson]

I've biased the comparator inputs and it's looking a lot better. There is still some raggedness and there's also an interlace problem.

You are obviously a genius, so no doubt solve the interlace issue.

[ppppenguin]

I think you missed my last post?

[`FRANK.C]

Hi Jeffrey
I was following this thread with interest.
Well done. Not an easy repair.

Frank

[ppppenguin]

Thank you. I'm just glad nothing was wrong with the core store itself. Everything else is fixable, all discretes and very ordinary TTL. There's what looks like a mask programmed ROM, tiny by modern standards, in the text generator. If that had been duff I could have reconstructed the contents and put them in a modern EPROM.

When the SN7525 chips arrive I'll remove my bodge.

[ppppenguin]

I found another subtle fault that caused the top and bottom castellations to be unequal height. One input of a 7402 quad NOR gate was stuck at zero. Managed to replace the wrong chip, an adjacent 7402. No excuse except sheer stupidity. Although they weren't on my stock list I found a few 74LS02 in the drawer. A good substitute.

Makes me wonder if there are any more subtle faults like this.

[Mike Watterson]

Yes, great the cores are fine.

[ppppenguin]

The PM5544 circle memory moved from core to some kind of ROM in early 1970s. That was for about 2k bits. The 1702, the first EPROM, was 256x8. It was introduced in 1971. I think the later PM5544 used a National mask programmable 256x8 ROM (the scanned manual posted online isn't clear enough to be certain) with a number like MM5210. Though this data book says it's only 256x4: https://www.retrotechnology.com/dri/NatS...77_300.pdf

Having crawled all over it for many hours, I'm getting less and less impressed by some of the design. I know it was done before even MSI logic such as four bit counters were available, let alone synchronous counters like the '161, but there is any amount of bad practice. Leaving aside all the RC differentiators and suchlike, the worst offence is that the circle width and position are determined by analogue adjustments. Even back then in 1971, I don't think it would been much harder or more complex to make the circle digitally correct by design.

[Mike Watterson]

Maybe some of it was a student on work experience?