Notifications
Clear all

Ex BBC Scotland PM5544 Colour Pattern Generator Restoration

Page 2 / 2
 
Refugee
(@refugee)
Member Deactivated Account
Posts: 4055

If you look at the part numbers on the chips you will see that many of them have a date code in the form (week number/year) 4 digits.
Look for chips that have solder that looks original for an accurate date of manufacture of the chip.
It is unlikely the chips will have been held in stock for more than a year.

ReplyQuote
Posted : 12/03/2015 5:32 pm
Anonymous
(@anonymous)
Group Deactivated Account
Posts: 16765

i have been told that the BBC always give their modified equipment a four digit number . I have looked at the ex BBC Scotland version but can"t find any .
do any of you know how i could obtain the information ? .
regards .
Gary.

Do you mean the Coded Equipment Number? (e.g. EP5/512, AM4/536, LS5/8, that kind of thing?) I have a friend who used to work in CAR at Television Centre: once upon a time they had PM5544s but they may well have been ditched after the fully-electronic generators for Testcards F, J and W were introduced. The paperwork might have survived though, but I expect the exodus from TC led to a lot getting binned. I'll ask.

The danger is that it could have been a local mod. and thus never properly documented.

This site may help but only if you can find the Coded Equipment Number. http://www.bbceng.info/ti/ti_top.htm

Al

ReplyQuote
Posted : 12/03/2015 5:44 pm
Doz
 Doz
(@doz)
Noble V-Ratter Registered
Posts: 1329

Facinating. What sort of ferrite core?

Surely not the sort here .. http://en.wikipedia.org/wiki/Magnetic-core_memory

Bubble memory?

ReplyQuote
Posted : 12/03/2015 8:37 pm
Anonymous
(@anonymous)
Group Deactivated Account
Posts: 16765

That's a grand first post to VRAT. Welcome!

The PM5544 was a groundbreaking piece of kit so it's good to know that one more is preserved. What did you use for the SPG? SAA1043 or SAA1101 would have been appropriate as they are Philips. And the coder?

Thank you.

I was barely a teenager when colour TV arrived, and for some reason I was intrigued by the idea that a box of electronics could generate a complex pattern like the PM5544. I never thought I would actually own one eventually!

I used some spare bits and pieces I had lying around for the SPG and coder.
The SPG chip was actually an HD44007A, which came out of an old and dead home video camera.
The PAL coder was an MC1377 from an Amiga A520 modulator, that came from my junkbox.
Of course, as you can see, I made a few last minute additions to the circuit.

ReplyQuote
Posted : 13/03/2015 3:42 am
Anonymous
(@anonymous)
Group Deactivated Account
Posts: 16765

Facinating. What sort of ferrite core?

Surely not the sort here .. http://en.wikipedia.org/wiki/Magnetic-core_memory

Bubble memory?

Yes, good old magnetic core. This is the memory card. The cores are under the square PCB in the centre.
You can see the cores in the gap between the two boards in the second photo.

The most recent date I can find on any of the original ICs is 7150, so I expect the unit was made sometime in the first half of 1972.
The nameplate on the back says it's a PM5544G/04. I don't have any documentation with this PM5544, so don't know what the G/04 in the model number means.

ReplyQuote
Posted : 13/03/2015 4:17 am
Anonymous
(@anonymous)
Group Deactivated Account
Posts: 16765

I once looked at that Hitachi SPG chip for a design but didn't use it. The MC1377 was a common PAL/NTSC coder chip.

The core store looks very simple and homespun compared to the high density ones used in 1960s computers. Making core store was very labour intensive.

I did see a video somewhere about people making core memories for computers used in the Apollo moon missions in the 1960s. It really was a slow and laborious process.

I wonder what the capacity of the PM5544 memory is?

ReplyQuote
Posted : 14/03/2015 2:33 am
jjl
 jjl
(@jjl)
Trusted V-Ratter Registered
Posts: 85

It looks like there are 8 rows and 12 columns of pins on the top PCB in the first picture above. Assuming these are connected to the rows and columns of the core array, that would give 96 bits of storage. Not a lot by modern standards.

John

ReplyQuote
Posted : 14/03/2015 12:33 pm
Anonymous
(@anonymous)
Group Deactivated Account
Posts: 16765

It looks like there are 8 rows and 12 columns of pins on the top PCB in the first picture above. Assuming these are connected to the rows and columns of the core array, that would give 96 bits of storage. Not a lot by modern standards.

John

I was able to count the number of ferrite cores by shining a light through the PCB and seeing the shadows cast by the cores through the PCB. I counted 8 rows of 21 cores, making 168 bits. Still not a lot.

I wonder how big the EPROM in the later version of the PM5544 was?

ReplyQuote
Posted : 14/03/2015 2:06 pm
Terry
(@terrykc)
Famed V-Ratter Rest in Peace
Posts: 4018

The basic pattern is generated in hardware. The only thing that the memory is used for is for generating the circle. Otherwise, the central part of the pattern is a square.

I always understood that it contained the information for one quadrant - essentially the offset for the pixel counter from the edge of the square but this would surely need much more than 168 bits? Assuming that both fields use the same information, there are about 130 lines in each quadrant.

Given the technology available at the time, how would you set about generating the data for each line with so little memory ...?

ReplyQuote
Posted : 14/03/2015 4:41 pm
Refugee
(@refugee)
Member Deactivated Account
Posts: 4055

I am pretty sure that there was another Philips unit that had an analogue circle generator with a matrix of gain stages around an RC parabola generator back in the early 1980s.

ReplyQuote
Posted : 14/03/2015 5:44 pm
Terry
(@terrykc)
Famed V-Ratter Rest in Peace
Posts: 4018

I wonder why they did the circle that way.

For consistent accuracy, I would imagine.

Philips were, after all, making a replacement for the long established visual test card technique - and I only ever witnessed one case of distortion: Test Card C from ITA Dover. They were obviously marketing a device with long term guaranteed stability.

I am pretty sure that there was another Philips unit that had an analogue circle generator with a matrix of gain stages around an RC parabola generator back in the early 1980s.

There was - I've got the manual for it here somewhere (I think it is also in the library) - but it was much cheaper than the PM5544 and in a different league.

I think one problem with the parabolic generator is that if there is drift, it will almost certainly be asymmetrical, leading to the circle becoming oval!

ReplyQuote
Posted : 14/03/2015 7:31 pm
Refugee
(@refugee)
Member Deactivated Account
Posts: 4055

The way I could tell was by the way it appeared and settled when it was selected on a unit we had at work in the early 1980s.
The manual had a description of the circuit that confirmed that it was analogue.

ReplyQuote
Posted : 14/03/2015 7:40 pm
TVJON74
(@tvjon74)
Honorable V-Ratter Registered
Posts: 658

I am pretty sure that there was another Philips unit that had an analogue circle generator with a matrix of gain stages around an RC parabola generator back in the early 1980s.

The PM5509 produces a circle

ReplyQuote
Posted : 14/03/2015 8:26 pm
crustytv
(@crustytv)
Vrat Founder Admin
Posts: 10839

The PM5509 produces a circle

As does my PM5515T, Teletext too.

ReplyQuote
Posted : 14/03/2015 8:49 pm
Anonymous
(@anonymous)
Group Deactivated Account
Posts: 16765

The PM5509 produces a circle

I have a PM5508, which does not produce a circle, but otherwise seems to be fairly similar to the PM5509.

Incidentally the PM5540, which was an earlier design compared to the PM5544, and is B&W, also has a core memory for the circle.

http://frank.pocnet.net/instruments/Phi ... M5540.html

ReplyQuote
Posted : 16/03/2015 2:41 am
Terry
(@terrykc)
Famed V-Ratter Rest in Peace
Posts: 4018

So, ten rows of eight = 80 bits.

I wonder if there is enough information there to suggest how it worked? The frequent use of ~ in signal names suggest that that it is an early way of showing inverted signals and those large red blocks are obviously some type of encapsulated circuitry - possible discrete components assembled to create a predecessor to the integrated circuit?

The Circle Generator is driven by a 17.24MHz clock so they must have been aiming for a very high precision circle! But how?

ReplyQuote
Posted : 16/03/2015 12:19 pm
Terry
(@terrykc)
Famed V-Ratter Rest in Peace
Posts: 4018

... The red blocks look like multi-tapped delay lines to me though I could be wrong. Delay lines were sometimes used to generate mutliple pulse timings, especially at sub-clock period intervals ....

I don't think so!

The photographs show that the identical devices were used all over the place - there is hardly a board that doesn't have at least one or two! Also, there is no sign of a delay line in the detailed block diagram http://frank.pocnet.net/instruments/Phi ... al_022.gif but there are lots of flip-flops which ties in with the FF507 designation on most of them.

17.24MHz give s 58ns clock period, fast for those days. That would give (approx) 1103 clocks per line of 64us and 896 clocks per active line of 52us. Neither seems to have any significance. To get a full resolution image you needed a bare minimum of 12MHz clock.

Yes, I'd already done the calculations and couldn't find correspondence either. Would you really need a 12MHz clock to define a circle to a reasonable degree of accuracy on a monochrome pattern? The 17.24MHz clock isn't used for anything else, as far as I can see ...

ReplyQuote
Posted : 16/03/2015 1:09 pm
Anonymous
(@anonymous)
Group Deactivated Account
Posts: 16765

Talk about coincidence.
This recent discussion on the PM5544 had prompted me to dig mine out and fire it up.
Would you believe the core memory card has developed a fault, precisely the thing we were discussing!
I tracked it down to one of the LM7525 core memory sense amplifier ICs.

Fortunately I found some for sale on the internet, so hopefully a cure won't take too long.

ReplyQuote
Posted : 16/03/2015 2:23 pm
Electrical
(@electrical)
Trusted V-Ratter Registered
Posts: 87

The Operators and Service Manuals for the Philips PM5544 Colour Pattern Generator are now in the Library. There are two copies of the manuals, one for the early type with the core memory and one for the later type with the EPROM for circle generation.
There are also copies of the Philips PM5532 Colour TV Sync Generator and the PM5545 PAL Colour Encoder, which are used with the PM5544 Colour Pattern Generator.
Regards Stan.

ReplyQuote
Posted : 16/03/2015 11:08 pm
Page 2 / 2