Philips PM5544 electronic testcard

[ppppenguin]

A recent donation to the Broadcast Engineering Museum is a Philips PM5544 electronic testcard. This was the first ever electronic testcard, dating from the early 1970s. https://en.wikipedia.org/wiki/Philips_circle_pattern This is an important item in TV engineering history. We are very pleased to have it.

Our unit is ex BBC and has various modifications that are specific to the BBC. We have what looks like a full manual including details of the BBC mods. The unit looks in decent condition though there is evidence of overheating on the power supply PCB. This may well be historic. The logic is Philips FJ series TTL which is equivalent to 74 series. In fact most of the chips are marked with 74 numbers. They are all very simple chips such as 7474 dual D-type, 7400 quad NAND gate etc. There is a small magnetic core memory that holds data for the circle.

We don't have the extender boards which look essential for any serious faultfinding. They are fairly small boards with what look like 0.156" pitch edge connectors.

I will run it up cautiously with a variac, monitoring the power rails. If there's a blown series pass transistor I don't want to find out about it the hard way.

There are lots of Philips blue electrolytics. While generally reliable these will have been powered 24/7 for many years. I will probably replace the lot in due course.

This BBC memo was with the manual. It's signed by George Hersee, who designed testcard F and whose daughter Carol appears in the picture. I thought George was with Designs Department (he was in DD in 1977 when I met him there) but in 1973 he was in SCPD (Studio Capital Projects Department). The building "H.W.H" is Henry Wood House, which was one of the many BBC premises near Broadcasting house. It stands on the site of the old Queens Hall which was bombed during WW2, causing the Proms to move to the Albert Hall.

[ppppenguin]

i have done an initial inspection. The rectifier PCB and some associated wiring is badly scorched with extensive repair work. The rectifier for the +5V logic supply is odd. There are 3x BY164 bridges in parallel. These are rated at 1.5A (some sources say 1.6A). The manual says the +5V supply is current limited at 1.8A.

This looks like a typically Philips odd design decision. The current sharing will be poor, though by fitting 4.5A total rectifiers for a 1.8A supply this shouldn't matter. I wonder if this was the cause of the original burn up. A scan of a different version of the manual shows they changed their mind and fitted a larger external bridge on the back panel. Our has the three bridges replaced and also has an external bridge.

[ppppenguin]

First light!

No drama, no smoke. The 5V rail came up correctly. Quite a lot of it is working. There are hints of a circle. I hope there isn't a problem with the magnetic core memory. There are several SN7525N sense amplifiers around the core: http://marc.retronik.fr/CoreMemory/Memor...17p%5D.pdf I don't have any and it might be tricky getting hold of some if I need them.

There is some text on the screen so the text board is fitted. I haven't looked to see how the text is programmed in. The board has 6x 74150 16:1 multiplexers and a 24 pin x 0.6" ceramic cased chip that's probably a ROM. It's soldered in! Probably replaceable with a 27xx EPROM and a bit of fiddling. Correction: I've looked up the manual, the ROM contains the character set. Programming is by wire links.

Apologies for the hideous photo. The diagonal bars are camera artefacts and there are loads of reflections from the screen. This is the Y (luminance) output only. I need to hook up the RGB outputs to my bench monitor which means getting it down from its shelf and a fair bit of fiddling around. Maybe I'll look and see if I have another monitor with RGB inputs.

[Mike Watterson]

Likely the sense amps can be replaced by some species of modern op-amp that can run off 5V.

A good start. What exactly is stored in core, which was later eprom?

Umm, some generators deliberately didn't have a circle. But the circle generation should be straight forward to fix if it is meant to have one.

PAL encoder separate box on all the earlier generators?

Amazing really that we bought a ZX Spectrum and I did a basic test card in BASIC for the screens we were aligning in 1982. It was never used for anything else. I left that company I'd founded in Jan 83.

Edit:
As this one was ex-BBC it probably does have a circle.

[ppppenguin]

The PM5544 always had a circle. In the early versions the data for this was in the core. Later versions had a PROM. The data is in the form of pixel counts to the start of the circle on each line. I think it uses 4-fold symmetry to keep the size down.

Many test pattern and testcard generators didn't have a circle as it was too hard or expensive. The PM5519 TV menders' generator has an analogue circle using parabolic waveforms. Obviously not inherently accurate as it has to be adjusted for circularity and centring. I dimly recall a design in CQTV (1980s or 1990s?) that I think used an arrangement of 74LS161 counters to make a circle. I think it was integrating H and V counts to make parabolae. I've done more sophisticated versions of this in FPGAs for some of my SPG/TPG designs.

The PM5544 was designed before most TTL MSI chips such as the 74161 4 bit counter were available. Hence it uses nothing more complex than dual flip-flops and AND/OR/INVERT gates. Except for the text generator which I think was later and uses 74150 16:1 mux chips.

The PAL coder was a separate unit. We don't have the Philips one, but we have plenty of PAL coders so it isn't a problem.

The PM5544 needs mixed sync and mixed blanking. I had to ferret round the back of my apparatus bay to bring the mixed blanking out from my master SPG to a socket on the bench. Not sure I've ever needed a feed of it before. Nor any other "traditional" pulses: burst gate, PAL square wave, line drive, field drive.

[Mike Watterson]

Was the core only written to at factory / service?

[ppppenguin]

I can't be sure yet, but I think the core was designed as ROM, with the threading done accordingly. The Apollo Guidance Computer is an example.
https://en.wikipedia.org/wiki/Apollo_Guidance_Computer
https://en.wikipedia.org/wiki/Core_rope_memory

The first EPROM was introduced in 1971: https://timeline.intel.com/1971/the-worl...:-the-1702 This was about when the PM5544 was first marketed.

[ppppenguin]

Now connected to the component video inputs of my bench monitor.

A number of boards had intermittent edge connectors. I've cleaned the gold-plated fingers with an eraser and sprayed the edge connector sockets with Servisol. Vast improvement.

The text generator card seems to have an internal intermittent. If I rattle it, I can sometimes see BBC1 as the text.

He central horizontal line seems to have reversed interlace. This may be intentional but it looks flickery. My photo has captured an odd effect just left of centre. The circle is still only fragments.

[ppppenguin]

I've replaced a dead chip and it's much better now. Working without extender boards is a PITA. i have to solder wires on to each point I want to probe and put the card back in its slot. 

I was looking at 70ns pulses once per field which is very difficult on any analogue scope except perhaps the Tek 2467 with its microchannel plate CRT. I don't have one, only a 2465B which could do it, but only in a well darkened room. My Rigol digital scope made it easy.

Much patient probing revealed a dead 7474 (dual D-type flipflop). This is part of an 8 bit counter that takes data from the core memory and passes zero at the right hand edge of the circle. I used a 74LS74 as I didn't have any 7474. Replacing standard TTL with LS is usually OK unless an output is driving a very heavy load.

There's at least one more fault but this is a huge improvement.

[Mike Watterson]

At least the 7474 family are still available. I might have some old stock TTL, though most are LS and HC.

Starting to look good!

Yes. I've given up on some old test gear due to lack of an extender. You are obviously very patient and determined.