So did the SPARCstation-10 work with the NV1 PSU and the homemade cable-harness?
Well, not really.
Perhaps that should be “Hardly at all”.
OK, the game’s up: no it didn’t.
When switching on, the Sun keyboard emitted its’ traditional beep, the HDD spun up, and the diskette drive performed its’ usual twitch. Seemed promising! However, there was no graphics output, and the keyboard LEDs did not flicker after 15-30 seconds. Once the power-on-self-test (POST) completes, the SS10 normally flashes the keyboard LEDs once.
Next I decided to check whether there was anything showing up on the serial-port – the SS10 can be switched in “debug” mode by holding-down the Stop and D keys on the keyboard when switching the power on, and will output POST messages to the serial-port. How can we monitor the serial-port quick-and-dirty? With ye olde favourite, a serial-breakout box (every retro project should have some way to feature one of these *somewhere* along the line!):
Nothing, not a twitch, everything stuck low, no sign of life after several minutes. So we had power, but something was preventing initialisation or POST…
Checking Voltages for Illumination
Firing-up the el-cheapo digital multimeter indicated that the nominal 5V lines at the SS10 motherboard connection were hovering around 4.8V. Also, the Power-OK TTL signal was inactive – which would prevent initialisation, the SS10 is fussy about that.
So far, the sense-lines on the NV1 were not connected, I had figured that they would not be needed, as the original APS-39 PSU does not have any sense line(s).
Time to take measure the live voltages on my other SS10 (with its ailing but working APS-039 PSU): The orginal APS-39 PSU is outputting 5.6V at its’ end, which is dropped by the cable-harness resistance to 5.06V ot the motherboard connector. It has to be pointed-out that the APS-39 cable harness is not really using AWG-18 wires – they are somewhat thicker than that, and they are not pure copper either: they are aluminium stranded wires (but based on how stiff the cables are, one could easily be fooled into thinking they were made of steel!). Volume for volume and length-for-length, aluminium has coinsiderably more resistance than copper, and drops more voltage.
Back to the Future
Many modern PSUs use sense lines (effectively returning the far-end voltage back to the source) to measure the actual voltage at the load end of the cable, and will either boost or reduce the source-provided voltage so that the far (loaded) end of the cable is at its’ nominal 5V – ie: to automatically adjust for the resistance of the cable itself.
So when using the NV1, I am going to have to hook up the sense lines too. Also, the intermediate plug+socket I had put in my original attempt at a cable would also increase the resistance and takes up far too much additional room for the low-profile tight-fit SS10 case. Back to the drawing board…
I decided to remake a new cable from pure-copper wires recovered from a 700W ATX PSU. These are not really the nominal AWG-18 either – they are a bit thinner (which would increase resistance) but they are pure copper (which reduces resistance compared to the original aluminium wire), and are long enough to not need an intermediate plug+socket. This time I *am* going to hook up the sense lines too.
Oh well, dig out the crimp tools again… The extra sense lines would have originally required a couple of three-wires-to-one-pin crimp, an almost impossible job, so I have reworked the wiring to avoid that: