Faulty Radius PC 24X - looking for repair tips

[croissantking]

I've acquired this non-working Radius PrecisionColor 24X (quite a nice card as I understand), which I'm interested in repairing.

Running it in a Macintosh IIci, the current symptoms are that there's no output, and it's not listed in TattleTech either – it's invisible to the system. I guess that if the card is not even detected then the Mac is not able to read the declaration ROM. But... the previous owner @mg.man says that he has already tried swapping in another ROM chip from a known good card, so it's unlikely to be that.

I would be interested in flashing a V2.0 ROM to an EEPROM chip just to double check, if anyone has a dump they could share with me.

Are there other common faults with these cards that could cause this behavior?

As the pics show, the card looks immaculate.

 

[Phipli]

I'd start by checking for lose pins on the ASICs, or if the ROM socket needs cleaning / is damaged.

Have you checked for bent Nubus pins?

mg.man shared the V2 ROM. Confirmed working by me.

Help for a nubus Radius PrecisionColor Pro 24X ROM/firmware upgrade

Oh, BTW, I checked the circuit and they're compatible with 512 byte chips. Interesting. Found a guy on eBay selling a pair of UV erasable '512s for a little over a fiver, so ordered some. What about EEPROMs? Aren't there pin-compatible ones of those that might work?

68kmla.org

 

[Phipli]

I'd check these two. Hard to see, but they could be solder bridges.

 

[mg.man]

mg.man shared the V2 ROM

@croissantking - g'luck - hope you get it going!

 

[Phipli]

They're nice cards, I have the 24xk but have a big ROM set up so I can switch between the 24x and 24AC ROMs.

 

[olePigeon]

I never tried using the 24AC ROM. Is there any advantage to using the Apple one over the Radius?

 

[Phipli]

I never tried using the 24AC ROM. Is there any advantage to using the Apple one over the Radius?

Performance is equivalent. Truth is I prefer the Apple driver, it integrates with the Monitors control panel.

It offers fewer options... But let's be honest I never fiddle with the settings anyway.

 

[croissantking]

No luck so far. I’ve checked:

- all Nubus pins are straight
- ROM socket is clean
- no solder bridges on the ASICs

 

[Phipli]

No luck so far. I’ve checked:

- all Nubus pins are straight
- ROM socket is clean
- no solder bridges on the ASICs

Poke all the pins with a pin or scalpel blade gently to see if any aren't stuck down

You just want to see if any move.

 

[Unknown_K]

I don't see any fuses, or 0 Ohm resisters used as cheap fuses.

Pop the ROM out and see if any connectors are mangled (since it has been removed at least once). That ROM socket is from the thru-hole era so every connector in the socket has a pin on the back of the card to check for continuity.

If the card is not detected (and the ROM and ROM socket are good) you have either something shorted or something open. With no fuses and nothing visibly getting warm or blown (are the main chips getting warm when installed in a turned-on machine?), I would get a meter and start poking around the capacitors checking to see if there is any voltage present. Find a section with no voltage and then start looking for a component that is open (blown).

 

[croissantking]

I've done some more investigation on this card today - I had it powered up on the bench and checked all the caps and major chips for voltage. There doesn't seem to be any area that's dead. It's a very hard card to diagnose, as there isn't an obvious starting point to understand the problem.

With Nubus, if the declaration ROM can be read by the system, does an otherwise non-working card still appear in the Slot Manager? Or is it one of those things that depends?

 

[Arbee]

I believe declaration ROM code can remove the card from Slot Manager entirely if it decides something is wrong, but that's rare.

If you have MacsBug installed, try dumping memory in the area of $FsFFFFF0 and see if any of the expected bytes appear. $5A $93 $2B $C7 would be the valid declaration ROM signature, but depending how the card works they won't be at contiguous bytes.

 

[zigzagjoe]

I believe declaration ROM code can remove the card from Slot Manager entirely if it decides something is wrong, but that's rare.

If you have MacsBug installed, try dumping memory in the area of $FsFFFFF0 and see if any of the expected bytes appear. $5A $93 $2B $C7 would be the valid declaration ROM signature, but depending how the card works they won't be at contiguous bytes.

Some older cards wire the detect function to the ROM select logic and simply won't select the ROM if nothing is connected. I want to say apple even recommended this in an earlier version of DC&D, but i would need to check. I've a radius pivot card with this behavior.

I can't recall if returning a failure in primaryinit can get the card entirely struck out of slot manager. I'm not sure, as I'm pretty sure it takes steps to avoid trying to initialize the card again unless specifically requested via a certain magic return code.

 

[croissantking]

So a fairly big update here. I got the card partially working after replacing the original 74FCT245T bus transceivers. Initially, I mistakenly used 245AT parts, not realizing they had significantly faster switching characteristics than the original chips. I’ve since corrected this and installed the proper parts.







The card now functions intermittently but still has serious issues:

- Stuck in 640×480, 1-bit mode

Attempting to change the bit depth results in an immediate crash and garbled output:



- Artifacting

I did get it to run in 16-color mode at one point, but with visible corruption:



- Sync Loss

The first time I got it running, it worked for a couple of hours before the monitor began dropping sync. Eventually, the card stopped outputting video entirely. I then realised my mistake with the wrong bus transceivers, and while waiting for the correct parts to arrive I let the card sit for a couple of weeks.

After fitting the new transceivers today, it came back to life - during which time I realised I was still stuck in 1-bit mode - then permanently lost sync again after about 20 minutes.

I hooked a logic analyzer to VCLK (the output from the Chrontel CH9203CL clock generator), and saw a wildly unstable signal: fluctuating from 0 Hz up to 8 MHz and back.

I’ve replaced the 14.31818 MHz reference crystal with a new part and confirmed it’s stable. I’m wondering if it’s now worth sourcing a replacement clock generator given these symptoms.

To me, the sync loss and artifacting look like separate issues, anyone else have any thoughts?