The first video ADC chip

[ppppenguin]

This advertisement for the TRW TDC1007 ADC is from the June/July 1980 issue of Wireless World. Before then, if you wanted to convert analogue video to digital it was a very complex and hairy board full of components. To do 8 bit conversion at video speeds (12MHz and higher) wasn't easy back then. The classic dual slope and successive approximation ADCs just can't run that fast. The opposite extreme is a parallel converter, with 255 comparators, a resistor ladder and a load of logic. That's what TRW accomplished in this chip but it's just about impossible with discrete comparators.

The method the BBC and others used in early digital video work was to do a 4 bit coarse conversion with a parallel converter. Then convert back to analogue, subtract from the input to get an error signal. A 2nd parallel converter then measures the error signal to get the low order bits. It's tricky to design but it worked. I think Quantel made a converter that converted just 1 bit per stage. Lots of tricky delay matching. A sort of pipelined successive approximation converter.

This TRW chip was a godsend to designers, even at over £300 each it was  bargain. That's something close to £1500 in today's money. It was also physically huge, about 3" long and over 1" wide, a chip for your nightmares.

I suspect it had its origins in radar or other military work but it won an Emmy award. It was THAT important.

Data sheet:
https://www.datasheet.live/index.php?tit...:PdfViewer&url=https%3A%2F%2Fpdf.datasheet.live%2Fdatasheets-1%2Ffairchild_semiconductor%2FTDC1007J1C.pdf

Now several 8 bit+ 30MHz+ converters are casually integrated into things such as PAL/NTSC decoder chips.

[ppppenguin]

A good techincal note on flash ADCs:
https://www.analog.com/media/en/training...MT-020.pdf

[beamcurrent]

I have a few bits of kit with these chips in them. This board is from a Rank Cintell slide file (an RGB video picture store).

This must have been an expensive board.

Brian

[ppppenguin]

Not just the ADCs but also the brickwall filters from Matthey or BAL. The filters were probably £100 each back then. They had to be flat within 0.1dB or so to over 5MHz and give lots of attenuation at 6.75MHz and above. This meant 5 or 7 pole elliptic filters with group delay compensation so as not to muck up the phase response.

[`FRANK.C]

At that sort of price you could nearly guarantee that it would be super sensitive to static etc.
What a great piece of history.

Frank

[ppppenguin]

I'm pretty sure it used bipolar transistors rather than MOS so probably not too worried about static. But who would take a chance at that price!