My Cursed Quadra 840AV

[Mac84]

A few of you here may be familiar with my cursed Quadra 840AV desktop. If not, this is a machine I purchased for $5 back in the mid 2000's at the Trenton Computer Festival here in New Jersey. Their flea market was 90% PC stuff, but the few people who had Apple stuff had excellent prices. A lot of the beige machines in the photo below were available for under $20-30.

My Quadra 840AV spent it's life being plugged into my TV (to use as a monitor, taking advantage of it's AV out). Even with a 40MHz 68k CPU it wasn't the best gaming machine, but I did have fun with what I could get working with it. I was quite proud of the $5 price tag, so I left that on there. Of course a few months later the machine sat unplugged. And almost a decade later I would find it in my parent's basement while helping them move. Around 2016 I started recapping my Macs, but I was very green and didn't have much of an idea of what I was doing.

I first attempted to recap my Macintosh IIcx, and although now it turned on and seemingly booted, the NuBus cards still didn't work properly... so I got no video out. Instead of giving up, I decided to double-down on the Quadra 840AV. The poor Quadra had stopped working and I thought I could simply recap it. I used through-hole capacitors (because that's all I had), and needless to say, it didn't fix anything.

I even put the poor thing through the dishwasher, hoping to clean it up... which likely did more harm than good! Again, I had no clue what I was doing. Since then it's been a regular machine on my live streams and has been through the ultrasonic cleaner so many times the poor AT&T chip has almost lost it's markings. Almost every time I bring out the board, I fix a trace or notice a wire that isn't connecting, etc.

So why am I posting about this? Well, because I'm stubborn and I want to fix it! I've spent countless hours (probably more like days at this point) probing and fussing with this logic board. What makes it difficult is the board is brown with dark brown traces, making it difficult (even sometimes with a microscope) to see what is going on.

So... where does this leave us now?

Currently when powered on the Quadra's power LED will turn on and the PSU's fan will spin... and that's it. No chime, no sound at all (even via using headphones). I've tried with and without RAM, VRAM, and a PRAM battery. I've also individually tested the 840AV's components in my Quadra 800. Thankfully the 840AV's 40MHz CPU and power supply function fine with the Quadra 800 logic board.

@Kai Robinson and @Branchus have been very helpful in giving me tips, comparing my board to theirs, and walking me through schematics and things to try. However, something is still wrong. There may be a short on the 5v rail or something that is not causing the shielding of the ports to actually make it to ground. Or maybe this board is just weird and that's how it is.

In my latest attempt I purchased a second-hand thermal camera for another project, but I was curious what was getting warm on the board. The CPU was getting hot enough to warrant installing the factory heatsink back on, so I wondered what else may be happening.

The hottest pieces on the board are the voltage regulators, at about 120° F (about 49° C). This is where the red marker is on the photo below. Although when measured with a multimeter, they seem to be doing their job, taking in 12v, and spitting out 5v.

However... something is painfully obvious. The RAM and VRAM (even with slots populated) stay ice cold. At least a few of the ROM chips look "on", and our multimeter probing proved them to be connected to where they needed to go.

Next Steps?

Since it's so easy to forget what has been tried, when, and how... I am writing this post to act as a log of what I've tried. Anyone is welcome to shout out suggestions of what to try, and I'll be posting more photos and close-ups of the board as I continue. But maybe... just maybe, one day... it'll live again. And at that point, I'd probably not even know what to do with it. And no, I won't be throwing it in the bin!

-Steve

 

[Elemenoh]

FWIW the symptom you describe is exactly what mine had before recap. Beyond replacing the caps themselves, some nearby ICs were quite crusty. Rather than just clean, I removed them, scrubbed the pads and cleaned up both sides of all of those chips. I know you've ultrasonically cleaned it, but have you removed and cleaned the SOIC chips near the caps you replaced?

In particular, U48 seems to handle reset, so that may be especially important to focus on.

 

[Certificate of Excellence]

Great story. It will be a youtube special one day starring Rick Moranis.

I have zero idea how to fix it beyond the obvious clean/check ics, caps, traces etc. but truthfully love the potential story that this thread will unfold - a man and his computer.. That thermal reader is a great idea. I definitely need one of those.

 

[Zane Kaminski]

You mentioned a short in the power. What's the voltage on the +5V rail when the 840AV is on? Also check that with the machine on, the /RESET pin is high (above 2.0V). If not then the system is not coming out of reset and you should check the power-on reset IC as @Elemenoh said.

Have an oscilloscope? Probe /RESET and +5V. /RESET should be low at power-on. Several tens or hundreds of milliseconds after power reaches 4.5V, /RESET will have a slowish (~100 nanosecond) ramp up to high. This is the power-on reset chip holding the system in reset until power has stabilized.

Then probe /RESET and the 40 MHz clock. Clock should start toggling before /RESET goes high so that the 68040 actually runs once released from reset.

If that looks good, probe /RESET and /TS. /TS means "transfer start" and is connected to the '040, MCA memory controller, PSC I/O bridge and MUNI NuBus bridge. /TS should be high until after /RESET reaches 2V or so. At that time, MC68040 will begin to fetch instructions from ROM, as indicated by /TS going low and then high repeatedly. You'll have to zoom in horizontally (time) on the scope to see this activity on /AS, like 100ns/div scale or so, as /TS can be low for as short as 25 nanoseconds. If you see /TS toggling then that means that MC68040 is trying to fetch its initial vectors.

You can refer to the Bomarc schematics to find all the relevant pins. Oh, and also do all of the testing with a good PRAM battery and a single known-good 72-pin SIMM but no VRAM since it's not required. Actually you might even consider not installing any RAM at this point since I think the 840AV will bong but then death chime if it has no RAM and that would be a good sign compared to where the machine is at now.

 

[demik]

Hope this helps: My 840av board is paired to an ATX or Quadra 650 PSU on a lab bench.
It will not boot like yours if the -12V rail isn't there.

Can't find where the -12V go in bomarc schematics however

 

[Zane Kaminski]

Can't find where the -12V go in bomarc schematics however

At the very least, the -12V goes to power the -5V regulator and to the NuBus slots. I think the -5V regulator is what Steve was referring to when he was talking about a part taking in 12V and making 5V. That must be for the negative supply. The main positive 5V supply is too high current to make with a linear regulator on the board, though maybe there’s some 5V analog/aux supply generated from the 12V that I’m forgetting.

Edit: Just watched a bit of your stream from the other day, Steve. Okay now I see the voltage regulator you mean. It’s right next to all the Philips chips. The Philips chips are for composite video input/output so I bet it’s a separate 5V analog supply. The main 5V supply comes from the PSU.

Anyway, a linear regulator generating 5V from a 12V supply oughta get pretty hot. That little regulator has to dissipate 7/12 of the +12V power spent on running the +5V analog line, compared to everything it’s feeding only having to dissipate 5/12 of the power, and of course the little chip is a much worse “heatsink” than one of those big Philips chips. So maybe the temperature is abnormal but I would say probably not.

Edit 2: If you are wondering whether the VRAMs are getting power, just check their Vcc lines. Here’s a pinout:

VRAM runs from the main 5V supply so if its power was shorted, the ‘040 wouldn’t get hot either. So there’s no 5V short. The reason the VRAM isn’t cold is probably that it’s not being actively used by the CIVIC video chip. In the absence of access, refresh, and serial data input/output a VRAM like that consumes only 10-20 milliwatts and shouldn’t get too hot.