Sonnet Crescendo NuBus Revision A

[GeekDot]

Hey Gang,

I had some extra hours at hand so I thought I'm doing something good to my PM6100 and replace its Sonnet Crescendo 250/256k for a 400/1M I recently 'found' on ePay.
Well, no dice ... the extension loads but is crossed-out, i.e. did not find anything... or it's the wrong version of extension. I'm currently using version 2.3.1 which works fine with the 250/256k which is a "REV E".

The 400/1M is a REV A which doesn't have the usual XILINX PLCC CPLD to the left of the heatsink. Searching Google for this Rev, it seems it's pretty rare/uncommon.
Given it seems to be a much earlier revision than the usual "D", maybe it needs to be used with an older extension? (V1.4.x)

Before I try them all one by one I thought I'd ask you, the nice bunch of 68kmla if somebody already had this behavior before and/or knows if a special version of the extension is required.

Thanks in advance,
Axel

 

[Byrd]

Not sure revision but I’ve the G4 nubus adapter in a 7100 using same drivers as a G3, try the usual CUDA and PRAM reset

 

[macuserman]

Driver should work, scrub those contacts with an eraser, and then spray with contact cleaner and try again. Also power off the computer and give the PDS slot a good soak with contact cleaner and let it evaporate and try again. I'd bet money you bring it back to life, I can see the dirty contacts in your photos even.

 

[GeekDot]

Driver should work, scrub those contacts with an eraser, and then spray with contact cleaner and try again. Also power off the computer and give the PDS slot a good soak with contact cleaner and let it evaporate and try again. I'd bet money you bring it back to life, I can see the dirty contacts in your photos even.

Did so in the meantime. Cleaned with Isoprop, scrubbed with glas-pen... they shine like new now.
PDS slot got a nice contact cleaner spaying... still nothing.

So I pulled PM6100 #2... which is running fine with its A/V card, so that slot is physically and electronically ok (while #1 also ran fine with an A/V).
Clean System 8.5 installed.

• Installed Crescendo drivers 2.3.1. System halts as soon the extensions should be loaded (no icon displayed)
• Removed 2.3.1 and installed 1.4.5, Systems boots, but the extension is crossed-out.
• Pulled the 400/1M, Plugged in the 250/256k -> Finds it just fine. Metronome reports everything correctly.

So is the 400/1M a dud? Well let's see what the Tricorder says while the system is hanging with the 400/1M...


40°C (104°F) is not incredible hot, but it does get some juice.


The inductors in the back... the bottom one is definitely warmer (~ same 104°F) than the one above it (darker), while the 4410 SOP Mosfets are even 5° warmer.
So it seems to be "online" but nobody turned on the ignition yet.

Let's see how the working 250/256k looks like...



Oh yeah, nearly 140°F... that's a difference...


and the Mosfet at 153°F.

It looks to me that the 1.4.x drivers can't identify/find the 400/1M and just drops the mike, while the 2.3.1 driver finds something it can't handle and goes into some sort of meditation.

Ah, for those interested... this is how the REV A looks "underneath the skirt" - using a QFP CPLD is even more advanced as the later PLCC ones.

 

[macuserman]

Maybe the card is defective, but I have a hard time believing its the driver, I've sold and owned numerous sonnet cards and I always use the latest driver no matter the system. Never had an issue, Sonnet is really good about backward compatibility with their drivers. Some of the other brands not so much. I just went through my photos and found Rev A, B, C, D, and even F on some that I have owned. Maybe you'll be lucky and find that an older driver is the ticket but I don't know how sure I am about that.

 

[GeekDot]

Maybe the card is defective, but I have a hard time believing its the driver, I've sold and owned numerous sonnet cards and I always use the latest driver no matter the system. Never had an issue [...]

That's a very helpful information... Thanks a ton!
So I guess there's something wrong with either powering or the firmware of the CPLD went the way of the Dodo.
Thinking of powering the card externally, I learned that the pinout of the PMac PDS is not easy to find. Any hints on that? [Never mind, Bolle provided me the PM8100 schematics which has it]

 

[GeekDot]

Just for the roundup, I think I found the culprit.
My Crescendo is warped. Just slightly, but it's doing the "banana". Looking veeeery closely, you can see the PCB bends up a tiny bit in the middle:



This will result in the card not sitting totally straight in the PM6100 slot and thus some pins don't have steady contact.
It is also viewable at the distance-holders of the loop-through AV slot of the card. Left and right sides are flush to the PCB:



...while the middle one has nearly 1mm distance to the PCB (nope, it's not a shadow):


So I drilled two of the 4 screw-holes of the heatsink a bit wider to release at least some tension there (loosing the threads, so I needed to replace the default screws by new ones plus a nut) and gave it a try...



Ok, we're getting somewhere. But still, the card is not straight of course as the loop-trough socket is soldered in holding "the banana" in shape.
But (for testing) I can carefully bend the slot-connector edge by pushing the ever-so-useful guitar picks as shims between the heatsink and the PCB.

This, plus slightly manually pulling the Crescendo to the left side of the case resulted in...


[Yes, my bus is overclocked to 40MHz. Rock solid since 1995 ]

...as long as I kept pulling the card to the side. Letting it go immediately lead to a system halt.

And this my friends is the end of this debugging venture. Thanks to all that helped with their input and ideas!

 

[cheesestraws]

yeahhhhhh that'll do it. Well debugged.

 

[macuserman]

Ah! Nice work I had a hunch your board was knackered. Now to engineer a board splint to keep it straight.

 

[GeekDot]

Ah! Nice work I had a hunch your board was knackered. Now to engineer a board splint to keep it straight.

Yeah, I'm currently thinking how to "iron out" that curve.
Heating up the loop-through-slot soldering might remove some more of the tension.
But it might also well be the case of a general warping created during population of the board or 'unlucky' ordering of layers inside the PCB. There must be a reason for so many revisions, right?