I did it that way so that a normal low cost turned pin socket could be used rather than some expensive extended pin socket but that does mean that the connections have to stick out of the end as there is no room underneath. (The Amiga and Atari ST pickup boards use the same technique)I like the way you've used a thin PCB with larger diameter holes for the socket to fit through, so when plugged into the one of the motherboard sockets, the board adds next to no extra height beyond the height of the additional of the socket!
I suppose it would make sense to use the clock fixer calibration as the default in future as that is likely to be the same for all clock fixers. The current default is just taken from one random machine and each one will be different due to the use of resistors and capacitors in the standard 6Mhz clock circuit.I reset RGB2HDMI to default settings to see if it would work without any calibration. There was a bit of shimmer, so an initial calibration was required.
In my experiments with the NuLA, there is some drift in the 6845 hsync position with respect to the 6845 clock input as it warms up and that amounts to one or two 144Mhz clock cycles (one or two positions using the x12 multiplier). However the clean band of sample phases should be wide enough to keep the image noise free as long as the default value is in the middle of that band. Interestingly this drift only affected the Hitachi 6845 and not the VLSI one which seemed to be a lot more stable (although that was the one that produced glitchy hsyncs on my Master)However, unlike my board, that calibration continues to work after repeated power cycles Image may be NSFW.
Clik here to view.. I'm going to leave the computer powered up overnight to see how it copes with thermal drift. I'm using the recommended Mode 7 calibration screen for all these tests.
Statistics: Posted by IanB — Fri Jun 14, 2024 1:03 am
