Tidying up a Falcon & CT60 combo (power supply & drive)

[PacificState]

Aha, yeah - it came with an 80Mhz crystal. I’ll dump
sysinfo in this state, but otherwise I’ll fit a 50Mhz part and go from there.

[exxos]

Aha, yeah - it came with an 80Mhz crystal. I’ll dump
sysinfo in this state, but otherwise I’ll fit a 50Mhz part and go from there.

When I did my tests originally the maximum I ever got out of a rev5 was 75MHz. But in the end I think I concluded that just running it at 64MHz would be about as high as I would go. 50MHz would indeed help alleviate that problem.

[PacificState]

Hmm, so - putting in a 50Mhz oscillator may have actually made the problem worse; it's now happening more frequently!

I pulled the card and re-examined the soldering job - there's a picture below. FWIW, I had just re-touched some of the DIMM joints on the right after removing the card which is why there's obvious flux (subsequently cleaned off).

I don't think I made a mess of fitting the new DIMM socket, but I do wonder about some of the rework wires I reattached on the left. Is there anything else I can look at? Thanks!

IMG_2414.jpg

[PacificState]

Also - after I get the 'this computer isn't...' error, launching SysInfo hangs.

[exxos]

I think the original CT60 was very temperamental on what OSC was used. I designed my own RSO because I gave up trying to get normal OSC's to work. Sadly I don't sell those anymore. But you could try soldering a 470R resistor from the OSC output to GND to see if that helps.

[PacificState]

While I generally find that posting a question about my problem leads to me immediately figuring it out after hitting 'Submit', in this case the system seems more unstable than ever before - if I reset after hitting the cookie jar issue, SDRAM isn't detected at all. This is true after reflowing the joints on the DIMM.

It'll sometimes come back after multiple resets - sometimes after reseating the DIMM, and sometimes not. I'm a little stumped, except that I see that there's an entirely missing pin from the IC next to the oscillator, which appears to be present in the other images of the CT60 I see on the web. If I've got the right data sheet, this is(?) buffering the clock signal (https://4donline.ihs.com/images/VipMast ... 353EA87419) - and the missing pin is a clock select that would be tied to ground if it was present. Is there a chance it's floating?

IMG_2418 2@0.5x.jpg

IMG_2417 2@0.5x.jpg

[foft]

That pin is explicitly cut on my ct60, guess it’s normal.

[PacificState]

Thanks for the confirmation - the search for stability goes on, I guess.

[Steve]

Thanks for the confirmation - the search for stability goes on, I guess.

I don't know if it helps, but when my ct60e stopped detecting the sdram, it was dry solder joins on the CPLD (and other surface mount chips) But the ct60 isn't the same as the ct60e, so I dunno if this might be your issue.

[PacificState]

Thanks! I might head there next. When I got the card, there looked to be two oddly bridged pins on the right side of the top CPLD; however, the bridge was flimsy and came off when I was brushing the board to remove dust (and yes, I probably should have been more careful - it was so fine that it wasn't clear if it was intentional, though). Is this another known mod? Otherwise, I might need to get the flux out.