Proper H-Sync Signal from a Square Wave?

[Jidis]

Hi all,

I've got a quick non-arcade-related question, but it involves a GBS 8200 converter board and CGA signals, so I thought someone here might know about it.

I've got a digital sampler which includes an old style video output signal. As far as I know, it's CGA spec at 15.756kHz using TTL signals for everything. The output connector they use on the back (8-pin DIN) I guess is non-standard and sort of sucks. I built a clone of a hard disk controller card they used to make for it, and discovered that the internal connector for it also carries those video signals. However, for some odd reason, the horizontal sync signal on that internal connector is a perfect 50/50 duty cycle square wave (at the correct frequency).

Is there a simple method or chip for converting a square to that 4.5uS pulse that H-sync expects? I tried converting it with a PIC microcontroller and got the signal straight, but even the slight offset of a couple PIC instructions throws it out of sync with the other lines. I thought there might already be something made for it.

Thanks!

[MonMotha]

Look up any one of many "re-triggerable one-shot" circuits.  You should be able to rig up a good ol' 555 timer (I hate those things, but they are quite versatile!) to do this.

Of course, you can do it with a micro, too.  Depending on what micro you use, you may be able to use the output compare/PWM module to do it in hardware.  Otherwise use an interrupt and count your cycles carefully.

You can also do it with discrete logic or a PLD.  Get a sufficiently large counter, clock it off some faster, ideally synchronous (to the hsync) clock, such as the pixel clock.  Either zero or pre-set the counter on one edge of the square hsync input, and compare the output with either your target count or zero, depending on if you're counting up or down, to generate the opposite edge of your output hsync.  Target count (or pre-set count) adjusts your sync pulse width.

[Jidis]

Thanks MonMotha! Glad I checked back in here.

I'm probably not knowledgeable enough to design anything, but I'll Google around for those 555 circuits you're talking about. I was sort of hoping that there's already a chip made to do that.

Regardless of the method used to construct that one line, how would one typically handle the offset it will create in relation to the other four? Is there a maximum delay time under which it will still work, or do you have to delay other lines to match the new one? Sorry if that's a dumb question.

Thanks!