Building a Standards Converter

[Jac Janssen]

That's very good news Jeffrey.

I'll collect the relevant data for the 567 line standard and let you know this.

It would be very practical to have the hex switch assigned to generate the two original frequencies used during the life of the standard, and have the remaining positions to generate the "standard" frequency + and - steps of 30 kHz or so as a means of fine tuning.
That would be especially useful during the first 40 minutes of operation of the set(s), since the local oscillator drifts approx 60 kHz during warm up. There is no manual tuning (like on the Marconi 702 / HMV 901) on any of the 567 line sets, so a small range of the RF frequencies (well - basically only sound of course) would come in very handy.

Thanks for your invaluable help in this!
Jac
p.s. I am still waiting for parts for my "first" Hedghog.

[ppppenguin]

All of these things should be feasible. The important things to be done are:

1: Specifiying what you really want. You can always change your mind later if you were wrong.
2a: Learn enough about the design and VHDL to make the modifications.
or
2b: Persuade somebody who already knows enough to make the modifications

One of the great things about doing it yourself is that it's very hard to break anything however badly you get the VHDL wrong. If you've got the H timings wrong you find out why and put them right. If the output disappears completely then you hunt around the design until you curse yourself for making a daft mistake.

When I built a 625>405 converter as an exercise I did much of the development with a 625 line output. I've got much better testgear for 625 than 405. All the interpolator stuff worked just as well by shrinking the height of the picture.

[`FRANK.C]

Hi Jac
As Jeffrey has said it is all doable.
Get back with the specs and if you don't want to do it yourself and are not in a hurray I will have a look at changing the code to suit.
I am currently trying it to do 16:9 input so will want get finished with that first.
Best to get a Hedghog working first on 405 then you will know that your hardware is OK.

Hi Jeffrey
The resources could be tight enough to do colour. I would think there is enough logic and multipliers (18 free). The memory may be the problem.
Currently I am using 5 line stores to do the 6 line interpolater and a time store memory. This uses up 63 % of available memory.
What is the minimum required to do colour interpolation and time redistribution?
I would guess that you probably get away with 2 line stores which would do 2 line interpolation one for Y and one for C.
Would 2 time redistribution memories be enough? if so there might just be enough memory to do that but it is too close to call without trying it first.

Frank

[Jac Janssen]

Thanks Frank,

That is a generous and fantastic offer.
Please see PM.

All the best,
Jac

[ppppenguin]

The simplest method would be to ignore colour interpolation. Remember that there is excess vertical colour resolution in all the colour systems. Since there are only half the number of colour samples in a line (as compared to luminance) you only need double the memory to do cour, not triple.

Other approaches include a smaller Y interpolator. 3 lines is enough. BBC research showed that 4 lines gave minimal improvement over 3 and that more than 4 didn't give any improvement at all.

Forgot to say that you need 1 memory to do the sine and cosine tables for the NTSC modulator.

Virtually every FPGA design I have ever done has been limited by the amount of memory available

[ppppenguin]

Jac has supplied this diagram of the 567 line waveform. Working from the figures on the diagram:

Line scan frequency: 14175Hz
Line scan period: 70.55us (approx)
Line sync pulse width: 5.64us
Back porch: 1.41us
Front porch: 0.71us
These are very narrow front and back porches!

Field sync is 3 each of pre-eq pulses, broad pulses and post-eq pulses. Total field blanking length is not clear. I would suggest 22 lines as a first attempt.

Sync amplitude is 25% of total (as against 30% in System A, I etc) though Jac has said that 30% would give better sync. He has also said that raising black level above blanking (as done in NTSC) would help with flyback blanking on his 567 line TVs.

Vision modualtion is negative. Sound is FM with 4.5MHz spacing. This gives a rather narrow vision bandwidth compared with system I or even system B/G. For vision BW equivalent to system B/G the sound spacing should be about 5MHz.

Aspect ratio is 4:3.

Carrier frequencies are:
Vision : 63.25 MHz
Sound: 67.75 MHz.

Jac also said there are drift problems with the receiver whch has fixed tuning. Hence a fine tune on the converter would be useful. In Jac's own words:
"So a "center" frequency of 67.750 (/ 63.250) MHz and a total range in steps of 30 kHz from 67.600 (/63.100) to 67.900 (63.400) would be nice. Or an extra 2 staps below and above that. That could take care of the fine tuning aspect."

Likely pixel clock frequency at output is 13.5MHz * 14175/15625 = 12.2472MHz (exact)

[`FRANK.C]

Hi Jac
I will reply to your PM.

Hi Jeffery
In that case is should fit. I just had a look at the FIFO's. The line stores use 1026 words each and the time redistribution FIFO uses 4096 words. It is currently using 5 line stores. Reducing this to 2 which is enough to do two 2 line interpolaters will free up enough memory to do 3/4 of a time redistribution FIFO. So together with what is free at the moment there should be enough.

Edit:Just seen your last post Jeffrey while I was writing this reply.
One thing that jumps out is FM sound. Is this a 1930's standard?
I think that FM sound will be a problem. Not because the modulators cant do it but because the sound input pins of the modulators aren't used in Hedghog. But maybe this can be got around.

Frank

[Jac Janssen]

Hi Jeffrey and Frank,

Thank you both!
Back and front porches may be increased.
The sets are now working quite well on 625, so combining the 625 spec with the number of lines of 567 should give good results.

Jac

[Jac Janssen]

One thing that jumps out is FM sound. Is this a 1930's standard?
I think that FM sound will be a problem. Not because the modulators cant do it but because the sound input pins of the modulators aren't used in Hedghog. But maybe this can be got around.

Frank

Hi Frank,

The standard was of 1947 and in use till March 1950 when Philips changed over to 625.
Philips took the 525/30 American standard and adopted it to 567/25, and chose a 4.5 MHz sound to vision carrier distance.
Although Philips did mention 567 lines already in a 1937 document. But no specs and it could very well have been AM sound and positive video modulation at the time. The applicable receiver was intended for 405 and used for demos across Europe in the Philips "Television Caravane" and was described as adaptable to 567. So I think that FM sound was not used at the time.
Unfortunately I have no pre-war Philips TV set in my little collection  
(That would be even more fabulous than your magnificent standards converter!)

I suppose I can easily hard-wire the relevant pins of the modulator IC(s) to wherever it is needed, and cut any not-required connections.
Baseband video conversion is the most important of course, but it would be nice to have RF output as well.

Jac

[ppppenguin]

I've just looked at the Aurora SCRF schematic (I'm not allowed to publish this). The video modulator also has audio connected to its audio input. presumably this is for intercarrier audio withFM audio and -ve vision. Audio feeds the video input of the audio modulator.