troed wrote:I'll ponder about how to build such a WS cycling circuit.
EDIT: Scrap the below. It WORKS. It does seem as if I need to time it to actual RESET very accurately though. I'm not sure why?
Congratulations Ijor - yet another feather in your hat! On one single power cycle and three RESETs I've gone through WS1, WS3, WS3 and WS4.
Old content, on what I built:
So I've built it now, but either I misunderstood how it would work or it doesn't work as theoreticized.
I have a push button connected to a 74121 which outputs a 170ns pulse (apparently, I was shooting for 125 .. ) onto the GLUE TEST pin. It definitely does _something_, if I push the button while the desktop is running it "jumps" vertically about one scanline up'n'down. However, pushing it either during holding RESET or otherwise does not change the GLUE-MMU wakestate* as detected from software.
For the ones interested: I initially tried disconnecting GLUE PLCC pin 52 (TEST) from GND by cutting the traces on the motherboard. There's a wide trace on the top, and a small one on the bottom, which would leave a nice via to solder to close to the socket. However, after cutting and scraping and cutting and scraping and not succeeding, I extracted the chip from the socket and saw that there was yet another one underneath. I gave up the trace cutting method and instead desoldered the pin from the PLCC socket and soldered a new pin directly to the GLUE, which now sticks up a bit and is easy to solder to. This is on a Mega ST Rev 5.0 motherboard.
*) While I need to do more testing, it seems my WS2 loving Mega ST suddenly likes WS1 a lot more since I rewired the TEST pin. Could be a fluke, but so far I've only seen WS1 afterwards.