Trouble with new Exxos Integrated Falcon RAM Board

[forty2]

Hello -

I recently installed the new Exxos integrated 14 MB RAM upgrade in my Falcon, and it worked great for a few days, but yesterday, I got a pattern of Xs on RAM boot, and confirmed a block of bad bits with the diagnostic cartridge (see attached screenshots). The listing for the new integrated RAM upgrade clearly states you should either use a new PS, or perform a capacitor upgrade. While I didn't use the Exxos kit, I did perform a capacitor upgrade on my own, and the board shipped after February, so it appears to have a voltage suppressor fitted.

IMG_5791.jpg

IMG_5783.jpg

I went to touch up the solder joints on the board, just in case that was an issue, and when I went to touch the second from right pin in the top row (see image), it was as if there was no pin there, and the solder ball just collapsed.

IMG_5793.jpg

Has anyone else had an issue with this board? Should I just order another one, or will the same thing occur? Would the older 72 pin SIMM adapter be a more reliable choice? (annoying, since it ends up being 2x the cost...).

thanks,
Paul

[stephen_usher]

Looks like the connector in the socket has become detached and when you melted the solder it dropped down further into the socket. So, defective socket.

[exxos]

That pad is mentioned in the store I think and the short across 2 pins. If you "fixed it" then you might cause more problems. The pin is intentionally missing.

You could have a bad connection on the connector itself or iffy joints on the falcon.

All the boards are tested before being shipped out, but I don't tests them for days on end. One guy managed to kill 2 boards but he didn't service the PSU and I could only assume that was the reason he managed to kill them. It's why I fitted voltage suppressors on the last batch. But something killed the boards but I don't know what. I'm still running my original board in my test falcon and it's not failed.

If your second board fails then please let me know, if something is causing them to fail, I will just remove them from my store as I have no idea what could cause the failures other than PSU problems as they are already known.

[forty2]

That pad is mentioned in the store I think and the short across 2 pins. If you "fixed it" then you might cause more problems. The pin is intentionally missing.

There is still a short across two pins, I didn't alter that - it's across the top and bottom rows of the upper-right (the same ) connector, about half way down. both of those pins are in place. The pin I originally mentioned is indeed missing, so that seems to be as designed.
I got the same bad bits from the diagnostic cart before and after I resoldered, so I'm confident I didn't alter anything.

Can anyone tell from the error report how many chips or lines are affected? To me, this seems like too few # bad bytes to indicate an issue with a data or address line - Does anyone else concur with that assessment?

Even though it's more initially, I'll probably just replace it with the DIMM socket adapter, as the replacement sticks will be easier / cheaper to procure here in the states in case I have another issue.

thanks,
Paul

[exxos]

The cartridge says bit 13. If you look at the schematics section for the falcon you can see which pin bit 13 is. Then you could check it's connected back to a RAM chip. I'm not at home until Monday so can't tell you which chip it goes back to. But there are 4 bits on each chip.

You could try pushing down on each RAM chip while doing ram test on the diagnostic cart to see if anything changes. But I suspect the soldering is good and a chip has died. I know a lot of original Atari ram boards failed for some reason as well. It's kinda why I designed the simm adapter board as the RAM could be easily changed. But I wouldn't bother spending much time on it. I'll just send you a new board if you PM your address.

EDIT:
If you have a scope it would be useful if you could post a image of what D13 is like plus 5v on the board.

[forty2]

Thanks for the offer, I really appreciate it!

I just so happen to have an oscilloscope (or 30...). I soldered some fine magnet wire to pins on the memory board so I could get pics on the scope, and I used a VDD line nearby for ground.

IMG_5800.jpg

IMG_5808.jpg

VCC was clean from power up. I have a differential plug-in for the scope, so I as able to measure ripple voltage as well. It was around 30mV, which seems totally reasonable.

Here's 5V per division

IMG_5803.jpg

Here's 20mV per division w/ 5V offset

IMG_5802.jpg

D13 (pin 19 according to this page http://the.protos.free.fr/yaro/falcmemc.html) looked normal - here it is during a few phases of the memory test

IMG_5806.jpg

IMG_5805.jpg

Let me know if there's anything else suspicious, or if there's anything else you want me to measure?
thanks!
Paul

[exxos]

Thanks...

I'm not sure your scope is showing the waveforms correctly.To have a perfect 0v level is near impossible for starters. Are you using x10 probe ? Though I think those old scopes don't capture all the flaws in the signals. You should see some under and overshoot of like half a volt.

The 5v is probably correct being low ripple. The RAM board has lots of caps so I think we could rule out a PSU issue. But there could still be spikes there which we are not seeing.

It's almost midnight here now but will try and hook my test falcon up tomorrow and measure what the signals are like there.

[forty2]

Yep, it's a 10x probe, and the readout on the scope compensates for it (so it's actually set to .2v instead of 2v)
Here's a slightly better shot showing under/overshoot - it looks like I'd expect it to.

IMG_5809.jpg

And here's the same wave as a one-shot on a digital scope

IMG_5810.jpg

[dml]

Nice Tek scope :) I have two 7xxx series sitting in the garage waiting for repairs (someday).

[exxos]

I have taken a reading off D13 (others look similar).

Basically they are spiking up to and even slightly over 7 volts!!

IMG_0119.JPG

IMG_0118.JPG

IMG_0117.JPG

I tried one of the address lines and it looks ok spike wise.. Aside from crazy type of crap is why I generally recommend changing the resistors on the bus... Oh let's not go down that road again.. I'm getting too old for this sh*t :lol: :roll:

IMG_0120.JPG

I think this is why a lot of original Falcon RAM board died as well :roll:

So indeed I assume the RAM has failed but not down to a actual fault on the RAM board but the system it has been plugged into has blown it up :roll: I would imagine if I left my Falcon turned on continuously for several days it will probably fail in the same way.

I wonder if this is in part why I had such trouble is finding chips which worked in the Falcon in the first place. I went for the Siemens brand because they seem to be the most reliable in general. Other brands would continuously come up random RAM errors or not work at all. That was a total nightmare in itself.

The only real solution to this would be to do a re-spin of the board and introduce a lot of series resistors to reduced spikes. Maybe even diode clamps. But with a limited number of these boards selling now I don't think it is really worth all the effort these days.

In any case @forty2 I think there is little point in sending you a replacement board is the same thing will likely happen again anyway. I am prepared to offer you a refund for the board in any case as I was unaware of the actual cause for these failures until now.

EDIT:

I just looked at the Falcon schematic and I cannot even see any series resistors on there at all :roll: I will see if I can hack in a resistor on D14 as it looks a easy track to cut to bodge a resistor in to try.