PM5544 Circle PROM Memory Dump Calculation?

[Chris55000]

Just a quickie one!

How difficult would it be to calculate the memory bit patterns needed to program a spare MM5213 PROM, OR, program a standard EPROM, in order to develop a PM5544 Circle Board, (and for that matter, the Character ROM on the PM5543 Text Character Board)?

(The MM5213 was still available at a reasonable price when I looked it up this morning, but I havent found an equivalent to the FDR116Z1 yet!)

If suitable dumps can be calculated for standard EPROMs, then there is a possibility of "rolling your own" PM5544 becomes feasible - after all, if somebody managed to build a PM5681 from the huge collection of files a former Dutch Philips chap so kindly made available, (and the PM5544 is simpler than the '5681), then the '5544 shouldn't be unachievable either!

Chris Williams

[Chris55000]

Further to the above, following a read of the MM5213 Datasheet, obviously it is useless buying a new device as it was "mask–programmed" by sending special punched cards to N.S., however, is it possible to read one out?

Looking at the Circle Unit Circuit Description, the PROM has to give (I think!) "00000000" (decimal 0) at the top of the quarter–circle, and "11111111" (decimal 255) at the mid–horizontal point of the circle, so would a 512 bit × 4 (2K) UV EPROM with the address locations programmed in sequence from Decimal 0 to Decimal 255, with the addresses incremented by 1 after the end of each line, the data outputs presetting the counter–chain by one less (or one more depending on polarity) per address generate the circle?

I don't have a PM5544 Circle PCB so I can't prove the point practically I'm afraid!

(However I do have a Selectable Pulse Duration/P.R.F. Generator that can be set to give timings equivalent to the vertical and horizontal pulse timings!)

If anybody can suggest whether the circuit is worth trying to make experimentally on breadboard/vero, etc., I'll give it a go when I'm back at work when I have access to the work's EPROM programmers!

Chris Williams

[FRANK.C]

Hi Chris
As regards the values required for the circle.

Remember different values will be required for odd and even fields and that will need to be accounted for some how to give correct interlacing.
I got over this in the Hedghog 405 line test cards by reading out the values from top to bottom for even fields and from bottom to top for odd fields.
But this required the values for the mid–horizontal point of the circle to be in the centre of the array.

The following is the array of values to produce the circle in the Hedghog 405 line test cards.
A value is give to each visible line. 188 per field. 0 = no circle.
By reading top to bottom for one field and bottom to top for the other, gives the interlacing required without having to store all 376 lines

The values in the array would be completely different to the ones you require. I just include them to illustrate the point

0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,29 ,46 ,57 ,67 ,
75 ,82 ,89 ,95 ,100,105,110,115,119,123,127,131,135,138,142,145,1 48,151,154,157,
159,162,164,167,169,171,173,176,178,180,181,183,18 5,187,189,190,192,193,195,196,
197,199,200,201,202,203,204,205,206,207,208,209,21 0,211,211,212,213,213,214,214,
215,215,216,216,217,217,217,217,218,218,218,218,21 8,218,
218,218,218,218,218,217,217,217,217,216,216,216,21 5,215,
214,214,213,212,212,211,210,210,209,208,207,206,20 5,204,203,202,200,199,198,197,
195,194,192,191,189,188,186,184,182,180,179,177,17 5,172,170,168,166,163,161,158, -
155,152,150,146,143,140,137,133,129,126,121,117,11 3,108,103,98 ,92 ,86 ,79 ,71 ,
62 ,52 ,39 ,17 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0 ,0


Frank

[Chris55000]

Evening Frank!

Thanks for that – from what I can gather from Philip's somewhat simplified description, that's how Philips implemented the circle as well, by storing a quarter of the circle in PROM, then counting up on Field 1, down on field 2 and reversing the signal polarity to generate the right–hand half, adding them all together in the "Circle Matrixing Unit", Unit 7 in the PM5544, so it's only necessary to store one quarter of the circle in memory, so all I need to do is to work out the values per line – is there a simple mathematical expression I can use, or a design reference I can refer to?

The PM5544 circuits do have "left–hand start" and "right–hand finish" adjustments so there is a small degree of tolerance in the timings!

Chris Williams

[FRANK.C]

Hi Chris
The circle on the Picgen 405 line test card generator was calculated by drawing a quarter circle on graph paper and counting the squares.
The 405 line circle on the Hedghog was calculated using trigonometry. If I remember correctly this wasn't a very quick was as a number of calculations were needed per picture line.
There is a 625 line circle used in the Hedghog PSC. I cant remember for sure how I calculated it. I may have used trigonometry or I may have wrote a program in C to churn out thee values.
If I Have wrote a program in C it wont be user friendly as I wrote it to get a single job done and some C programming knowledge would be needed to use it.
I will look later to see if I have it.
There are probably better ways of doing it.

Frank

[FRANK.C]

Hi Chris
I found the programme that I wrote to calculate the test card circle values.
There are two values that it needs to do the calculations.
1) The number of lines high the circle is in one field.
2) The maximum value the circle is at its mid point.

It might not be suitable for what you want.
But if you wan to send me those values I will input them and send you what it spits out.

Frank

[Chris55000]

Evening Frank!

Philip's circle data isn't very clear or comprehensive, but I can quote:–

Horizontal diameter of circle :– 34.18 μs ;
(Circle radius 17.09 μs)

Distance from top and bottom of circle to top and border castellations :– 20 lines, excluding circle and white grid–line ;

Number of bits used to generate the radius :– 252 bits ;

Height of top and bottom border castellations :–

14 lines top, 13 lines bottom, excluding the white grid–lines ;

Height of circle (vertical diameter) :– 504 lines ;

I have the full PM5544 bookwork if it's any help, which you can download from my Google Drive :–

https://drive.google.com/drive/folde...MpudAlwnjuC6sx

Address Counter Presetting :–

Start of Field 1 :– All bits 1 except A4 which is 0, decimal 247 ;

Start of Field 2 :– All bits 0 except A4 which is 1,
decimal 8.

Chris Williams

[FRANK.C]

Hi Chris
The circle is 504 lines high so for one field it will be 252 lines.
It says "Number of bits used to generate the radius :– 252 bits " I think we can assume that this is in the horizontal.
Those values sound plausible for a 625 line circle.

Using those values I have run the programme.
The output is in the csv file attached.
It can be opened with notepad, liber calc etc.
If you open it with Liber calc. You can use the "Insert Chart" function and from the values it will plot a semicircle.

I have also attached the C code I used as a txt file.
I am sure someone can make a better job of it as my knowledge of C is poor.


Edit: I had to rename the csv file to txt to get it to upload.

Edit 2:

Hadn't seen your edit before I posted.
At first sight the pre-setting of the address appears odd.

Frank


Frank

[FRANK.C]

Hi Chris
From a look at the circle part of the manual. I don't think that those values will work.
I believe now that it uses the range of values from 0 to 255 and not 0 to 252 as I thought.
it uses a 15Mhz oscillator. 255 cycles of 15Mhz is close to 17us which it states is the radius of the circle.

(In this paragraph I am referencing Fig. XI-1 on page 47 of part 2 of the manual)
I also believe that the values stored in memory should be counted from 'A' to 'B' rather than from 'B' to 'C' as I have done on the values that I uploaded.

One thing I don't understand is why the address counters are been preset to 8 and 247 when there are 252 samples.

I could upload new values but I think it would be better if I understood why the address counters are preset to 8 and 247.
In post #4 you referenced unit 7. Can you point me to it in the manual? There might be an explanation there.

Frank

[Electrical]

Hello Frank
Unit 7 is the Cross Gate PCB which contains the Horizontal Dividers and can be found ion Part 3 of the PM5544 manual, please see attached PDF. Part 3 also contains Unit 5 the Vertical Decoder, Unit 6 Horizontal Oscillator and Unit 8 the Circle Matrix.
If you have a look at the block diagram in Part 1 on manual page 29, PDF page 21 you will see Fig VI-1 Pulse Diagram which may help to explain how Philips generated the circle. Please see attached PDF of Part 1.
I hope this information is of some help.
Regards Stan.

[FRANK.C]

Hi Stan
Thanks for point that out. It is was exactly what I needed.
Now I think I understand.
The pulse that Presets the memory address arrives 10 lines before the start of the circle.
It is for that reason that when counting down it is preset to 8 and when counting up it is preset to 247.
This implies that the 252 circle values are stored in the center of the memory. With addresses 0,1, 254 and 255 left blank.

Frank

[FRANK.C]

Hi Chris
For what it is worth the txt file attached contains the list of values that I think would be in the memory.
I have put the value 255 into addresses 0,1, 254 and 255 which I believe are not read.

One thing I don't know is which field the circle begins on. In the file the order of the values are set so the top of the circle begins on field 1. If it begins on field 2 the order of the values would need reversing.
If the values are in the wrong order it will produce a circle with a very ragged edge.

Frank

[Chris55000]

Evening Frank!

Thanks for giving me your time on this!

I'll use your provisional values as a start, I've got a Manor Supplies Mark V as a test bed, and I can begin by building the Circle Memory Unit 2, using the appropriate vertical and horizontal drive pulses from the ZNA134 to control the timing, and feed the circle output into the Manor Supplies modulator!

I should imagine I can get away with a prototype on Vero board or Proto Board to begin with, and will report my results ar a later date when I've got a result, then I'll make a PCB layout once I've proved the circuit works!

Chris Williams

[FRANK.C]

Hi Chris
Looking forward to see how you get on.


Frank