8-bit acorn hardware • Re: RGB to HDMI using a Pi Zero and a small CPLD

Thanks for your help with this.

I've done a bit of experimenting this morning.

I'm using the same RGBtoHDMI Pi and CPLD firmware versions as you are, the same test card, and testing both with and without the analog boards. The cable connecting the Master to the RGBtoHDMI is short - about 30cm.

I have two Masters:
- an Issue 1 (white silk screen and a Hitachi 6845)
- an Issue 2 (yellow wilk screen and a VLSI 6845)

I'll test both...

The Issue 1 Master
With the Analog board, the calibration is not error free:Without the Analog board, it's actually much better and you get an error free plateau of three sample offsets:For reference, this is what the clock looks like:The Issue 2 Master
With the Analog board, the calibration is not error free, and is much worse than the Issue 1 Master.Without the Analog board, it gets a bit better, but is not error free, and again much worse than the Issue 1 Master:For reference, this is what the clock looks like:So I removed C30, replaced it with a socket.

The Issue 2 Master - with C30 removed
With the Analog board, the calibration has improved, but is not error free..Without the Analog board, it's now error free.For reference, this is what the clock looks like:My summary is:
- using the analog board always seems to result in worse results
- in my case at least, reducing C30 helped. I got a noticable improvement reducing it from 100pF to 47pF. Removing it entirely did really yield any further improvement.

I noticed a couple of other things that might be worth following up on.

1. I tried playing with the DAC thresholds (e.g. DAC C (RB Hi)), and that made surprisingly little difference, even over a very wide range (0.5v to 3v) So little difference, it made me wonder if the adjustment was working at all.

2. I measured the 6MHz clock jitter on the two machines: the issue 1 has about 4ns of jitter and the issue 2 has nearly twice this at 8ns. I wonder if that's the main difference between the two machines. I don't really know why this difference exists. It could be the crystal is getting old. It could also just be more power supply noise.

Anyway, I hope this helps, and gives you some ideas to try.

Dave

Statistics: Posted by hoglet — Tue Feb 27, 2024 12:20 pm

---