[guzzler]

good day all...

right, i finally finished wiring up my META, power supply and all and........ nothing happened... here are the symptoms...

24V in, unbalanced +/- rails by around 1V
Opamp gets pretty damn hot (scolding to touch)
no output at all, not even a buzz
LED works fine

here are the other relevants...

AD8620
1x 2002 on each buffer position
CCS with 500R resistor
ALPS Blue pot
4x470uF + 2x6.8uF + .1uF (by the opamp)
No input coupling caps

any ideas, sounds like a short to me somewhere, but any other suggestions??

thanks...

g

 

[Excalibur]

If the opamp is getting very hot something is drawing too much current from it's output pin. Check that node of the circuit for a short. Before you do that though, check the voltage (+-) going to the opamp to see if it's correct.

 

[fiddler]

I had a problem similar to that, and it ended up being a bad opamp. Could be that I damaged it by overheating. $12 opamp down the drain. It got so hot that it would start smoking (!). No matter what amp I plugged it into, it showed the same symptoms...

But hopefully it's not that.

Also, it's not a bad idea to wire up and test the power supply before all else. It helps narrow down the possiblities when you're trying to troubleshoot later on.

 

[ADS]

A hot IC is usually an indication of incorrect polarity. You probably have + and - reversed (and you'll probably need a new opamp, unless the one you have didn't get hot enough to fry

)

 

[tangent]

Pull all the chips, turn it on and see if things improve. Current draw should be very low, rails should be precisely balanced if 2001G is populated or shorted. Then add the chips one by one until it misbehaves.

 

[guzzler]

damn damn damn, should have used sockets... oh well, we'll see what happens...

i sat up for a while last night with my meter checking everything:

no continuity between +/-, good (can't reverse polarity, diode protects)

OPA is biased at 1mA


i'll try desoldering the OPA (put a socket there while i'm at it!), as thats my prime candidate of error!

cheers all, i'll let you know how i get on!

g

 

[guzzler]

ok, there is something seriously wrong with this damn board:

pulled opamp (after desoldering), but it's ripped up a trace and trashed the Browndog (so prolly the AD8620 as well, unless i can desolder that!)

measured GND-SUPPLY IN: still 1V unbalanced
measured GND-OPAMP SUPPLY PINS: .2V unbalanced

how the &^£$ can the two be different?? as it stands:

potential gain: 0
potential loss: ~£170 + time + hair(!)

i can't think of anything else that could possibly be wrong, checked for shorts, dodgy connections and no, it still wants to play rough...

not a good day! any other suggestions (aside from throw the damn thing away, i'm suggesting that just now) would be helpful! sorry bout this, just i've been waiting to finish this for so long, and when it goes so badly wrong it's a total whitey!

g

 

[fiddler]

Quote:

ok, there is something seriously wrong with this damn board:

Heh, yeah it's tempting to blame the board. I've done the very same...

Oh, a tip to remove SOIC chips from its pads... suck up as much as much solder as possible using desoldering braid. Take a small flat head screwdriver and prop it between the chip and the board. Usually by now there's not enough solder there to really hold the chip there firmly, so you can remove the chip with minimal force.

 

[fiddler]

By the way, make sure your rail splitter isn't fried.

Try replacing it with voltage splitting 1K resistors and see what happens.

 

[guzzler]

yeah, will check out the TLE... just been to the range, calmed myself down a bit... never the best to work when the red veil is in the way

g

 

[morsel]

On rare occaisions a fleck of copper shorts out traces on ExpressPCB boards during manufacturing. This is why the board should be candled (held up to a strong light to see the traces through the board) prior to assembly, as such flecks will show up. I had a META42 board with one output shorted to ground by a fleck shaped like Czechoslovakia which I extracted with a razor knife. The odds are very low that this is your problem.

 

[guzzler]

i heard another method of finding shorts, unfortunately can't do it with a populated board:

plug into a car battery and watch the sparks fly (i do not advocate this and have never tried it!)

g

 

[tangent]

It's too late to fix now, but it the future, avoid putting pressure on a solder joint when desoldering until you're sure the solder is flowing or you don't care about the PCB. If the joint doesn't flow, add liquid rosin or fresh solder. If it doesn't move easily, you'll rip up pads and traces for sure.

For desoldering SO-8 chips on Brown Dog adapters, you can simply lay the iron flat along all four pins on one side while you pry it gently up with something sharp underneath. This will bend the pins on the other side, but this method is quick, so the minor problem of bending the pins back is more than made up for by the lower risk of overheating the chip.

As for your op-amp, at this point I would cut the adapter off the board, desolder the op-amp from the adapter and remove the pin stubs from the META42 board, and solder the op-amp to the SO-8 pads on the back of the board if they're still okay. You may have to repair some top-side traces first.

I doubt you've blown 170 pounds. Even if you have to start all over, you can scavenge most of the parts for another try. And, now you've learned the value of stepwise testing, so you should succeed next time.

As for the bad PCB thing, the incidence of that problem has been very low. It happened two times with the first run back in June of last year. The only other time I've seen anything close was recently where some solder had flowed over the solder mask and not quite shorted something. This latter one was easy to see and fix during assembly. I've heard no other reports of bad boards.

 

[guzzler]

sure it will all be fine, got a day off tomorrow so i'll have a days work on it!

thanks

g