Blown opamps in chaos circuit

[George Herold]

So I’ve got this Chaos circuit.
https://www.dropbox.com/s/afy4kb63ftdo1ke/chaos2.pdf
(I just installed dropbox.. I hope that works)
Basically two multipliers and three integrators (with gain and reset)

I’ve tested about 20 of them so far and in 3 or 4 of the 20 one or
more of the opa2277 opamps has been fried. Now all the production
people wear grounding straps when stuffing a PCB. And we use this
opamp all over the place and I’ve never* seen any problem. So I’m
trying to figure out what’s going wrong.

(Finding the blown opamp in this circuit is a bit challenging.
There’s no real input and output, it’s just a multi-connected
loop.)

Now the opa2277 data sheet warns about too much input current. (must
be less than 20mA) I’m wondering if the stuffing order of the board
could be an issue. Could there be some charge on the integrating caps
that gets pushed through the inputs as the pcb is stuffed?

For the moment I’m having them put sockets in for the opamps.. at
least that way it’s easy to change. And the opamps will go in last.

Any thoughts/ ideas would be most welcome,
George H.





*that’s not quite true I had a similar issue with an opa2277 used in
the PID section of a temperature control board.. these were made by a
board stuffing house and in two (out of ~100) the opamp was also
fried.

 

[George Herold]

Maybe the reset button kills the amp. It might discharge one of the integrating
caps back into the front end.

You could also get some (appropriately random) voltage multiplication effects
from playing with SW1.

No, it arrives on my desk for testing already blown. I can't make a
working circuit fry the opamp. (I tried all sorts of switch flipping,
power on /off cycles.. no problem.)

Why do you have switches S1 and SW1?

Well S1 resets the 'thing' You can then watch how the signal changes
with time from a given starting point.
SW1 just changes the time scale for the circuit. It's easier to see
on the 'scope with a fast time scale, but slower has less noise and
it's easier to find the period doubling points.

One thing that does stink about digital 'scopes is the X-Y mode.

S is a switch

U is an IC (not IC3, not IC3POW)

Q is a transistor (not T, not TR) (T is transformer)

L is an inductor (not IND, not CHO)

R is a resistor (not RV, not POT, not RN)

D is a diode (not Z or anything) (CR if you're old enough)

If I was named IC3POW, I might blow up too.

Grin... It was easier just to dump it out of eagle than to try and
redraw it.

George H.

 

[George Herold]

That looks like Eagle libraries, which are quite inconsistent on
nomenclature, except for using "IC" instead of "U".

--
Tim Wescott
Control system and signal processing consultingwww.wescottdesign.com- Hide quoted text -

- Show quoted text -

Yup Eagle, Different people add different parts to the libs and end
up calling things whatever they want. I doesn't bother me... but heck
I use i for sqrt(-1).

"Damn it Tim, I'm a physicist not an engineer!" (Channeling Deforest
Kelly :^)

George H.

 

[Spehro Pefhany]

So I’ve got this Chaos circuit.
https://www.dropbox.com/s/afy4kb63ftdo1ke/chaos2.pdf
(I just installed dropbox.. I hope that works)
Basically two multipliers and three integrators (with gain and reset)

I’ve tested about 20 of them so far and in 3 or 4 of the 20 one or
more of the opa2277 opamps has been fried. Now all the production
people wear grounding straps when stuffing a PCB. And we use this
opamp all over the place and I’ve never* seen any problem. So I’m
trying to figure out what’s going wrong.

(Finding the blown opamp in this circuit is a bit challenging.
There’s no real input and output, it’s just a multi-connected
loop.)

Now the opa2277 data sheet warns about too much input current. (must
be less than 20mA)

It's claimed to be able to withstand +/-30V differential input voltage
(1K internal series resistors), but nothing else leaps out.

(I assume you've got a fixture or something to guarantee the +/-15
isn't mixed up.. there are six different ways to connect 3 power
supply wires, and not all of them are benign- you're also close to the
36V abs max supply limit)

If nothing else comes up, maybe consider back-to-back diodes (eg. one
BAV99 each) from the inverting inputs to ground.

I’m wondering if the stuffing order of the board
could be an issue. Could there be some charge on the integrating caps
that gets pushed through the inputs as the pcb is stuffed?

Seems unlikely.

 

[Spehro Pefhany]

Any thoughts/ ideas would be most welcome,
George H.

P.S. Also suggest you check where the op-amps came from and make sure
that they were not returns or otherwise in question.

 

[George Herold]

It's claimed to be able to withstand +/-30V differential input voltage
(1K internal series resistors), but nothing else leaps out.

(I assume you've got a fixture or something to guarantee the +/-15
isn't mixed up.. there are six different ways to connect 3 power
supply wires, and not all of them are benign- you're also close to the
36V abs max supply limit)

Grin... I guess I'm the fixture. (potted into my chair perhaps.) I'm
the first person to supply power to the boards and I always give them
a visual inspection first... power correct, and check that any tant's
are in the right way.

If nothing else comes up, maybe consider back-to-back diodes (eg. one
BAV99 each) from the inverting inputs to ground.

?? Do you mean back-to-back zeners? Or does back-to-back mean in
parallel?
I could do that on the integrators.

Seems unlikely.

Everything seems unlikely. I thought maybe we got a bad lot of IC's
and tested a bunch from the tube with the same date code... they were
all fine. We use this opamp in ~10 different instruments and I've
never had a problem.

Thanks for the input,

George H.

 

[George Herold]

Looks as if closing reset would try to dump all of the charge built up
in integration cap C6 into the inverting input of IC2A in an instant.
It would also short IC2A's output to ground.   I can imagine the IC
being rather unhappy with either or both of these things.

Hmm, Thanks I hadn't thought about the other end of the cap. (the one
tied to inverting input.) So hitting the reset causes the inverting
input to have a spike (pulse) that's mostly about 1 volt but sometimes
up to 2 volts. (This is with the circuit configured for the biggest
voltage output... again in the 1-2 volt range) Shorting the opamp
output to ground is not that much of an issue.

(The switch notation seems to imply that S1 is normally closed,
and opens on a push... this seems backwards but even if correct the
problem would remain... when the contacts close, the op amp is likely
being asked to do what it doesn't wanna do).

Sorry about the NC/NO switch screw-up. The switch has to be open for
the circuit to run.

George H.

 

[George Herold]

Yes, definitely!  I bought a bunch of AD275 opamps from a VERY
large but non-franchised distributor quite a few years ago.  50%
of them blew up at power on!  I went to AD, and they of course wouldn't
support parts from that source, but they did confirm that it would be
absolutely impossible for parts with different date codes to be on the
piece of tape.  So, that clearly proved the parts had at least been
off the original tape.  So, possibly ESD by re-reeling the parts,
but my guess is they were recycled off boards.  This was right around
the time that AD changed their markings from pin 1 dimples to the
chamfered edge on SOIC parts.  Their triangle logo could easily be
mistaken for a pin 1 mark, and would get you to load the parts
backwards.  So, my guess is these went on boards backwards, the
ones with the lowest voltage parasitic diode on a board would
blow, and the whole production run of something went out as scrap,
and I ended up with them.

(Sorry for the long war story, but that's why I'll NEVER, EVER
buy from a grey-market source again!)

Jon

Grin.. I checked a bunch, they were all fine. Most likely came from
newark. I got some bad 79L12 volt reg from Jameco... a light
gray?
I don't go there any more.

George H.

 

[Jamie]

So I’ve got this Chaos circuit.
https://www.dropbox.com/s/afy4kb63ftdo1ke/chaos2.pdf
(I just installed dropbox.. I hope that works)
Basically two multipliers and three integrators (with gain and reset)

I’ve tested about 20 of them so far and in 3 or 4 of the 20 one or
more of the opa2277 opamps has been fried. Now all the production
people wear grounding straps when stuffing a PCB. And we use this
opamp all over the place and I’ve never* seen any problem. So I’m
trying to figure out what’s going wrong.

(Finding the blown opamp in this circuit is a bit challenging.
There’s no real input and output, it’s just a multi-connected
loop.)

Now the opa2277 data sheet warns about too much input current. (must
be less than 20mA) I’m wondering if the stuffing order of the board
could be an issue. Could there be some charge on the integrating caps
that gets pushed through the inputs as the pcb is stuffed?

For the moment I’m having them put sockets in for the opamps.. at
least that way it’s easy to change. And the opamps will go in last.

Any thoughts/ ideas would be most welcome,
George H.





*that’s not quite true I had a similar issue with an opa2277 used in
the PID section of a temperature control board.. these were made by a
board stuffing house and in two (out of ~100) the opamp was also
fried.

Youi're leaving the caps open, allowing them to charge to what
ever.

It's been a while since I've looked at the specs on that op but
something tells me you could be over doing it on the input, depending
on what the other input is doing...

Personally, I would put high value bleeder R's crossing the switch, at
least this way the caps are more than likely to be close to the existing
value when switched in.

You should be able to select proper values that won't conflict with the
current values selected. 1 MEG. 5 MEgs etc..

Jamie

 

[George Herold]

Have you noted which of the op-amps have blown?  Are the blown
characteristics all the same?

I see the datasheet says the output can be shorted indefinitely so S1
should be fine when closed.

It does say the "inputs can withstand ±30V differential inputs without
damage" and also "The protection diodes will, of course, conduct current
when the inputs are over-driven".  Later I see the inputs are "protected
with 1kΩ series input resistors and diode clamps"

I'm therefore surprised that there is indeed an issue as the device
should be fairly rugged.

Yeah! I've been swapping 'em in and out on my bench (no wrist strap
most of the time.) and hooking up the power backwards every once in a
while. And never any problems.

So all of the time it's IC2 (U2) the Y opamp that has been fried, but
once the Z opamp was also fried. The Y opamp does have the biggest
feed back R.

I'm done testing the latest batch, there'll be more coming in a few
weeks or so, waiting on a part. I'm guessing it won't happen again...
Some one time ESD event that took out a handful? (doesn't explain why
it's IC2?)

George H.

 

[P E Schoen]

"George Herold" wrote in message

So I’ve got this Chaos circuit.
https://www.dropbox.com/s/afy4kb63ftdo1ke/chaos2.pdf
(I just installed dropbox.. I hope that works)
Basically two multipliers and three integrators (with gain and reset)

I’ve tested about 20 of them so far and in 3 or 4 of the 20 one or
more of the opa2277 opamps has been fried. Now all the production
people wear grounding straps when stuffing a PCB. And we use this
opamp all over the place and I’ve never* seen any problem. So I’m
trying to figure out what’s going wrong.

(Finding the blown opamp in this circuit is a bit challenging.
There’s no real input and output, it’s just a multi-connected
loop.)

Now the opa2277 data sheet warns about too much input current.
(must be less than 20mA) I’m wondering if the stuffing order of
the board could be an issue. Could there be some charge on the
integrating caps that gets pushed through the inputs as the pcb is
stuffed?

For the moment I’m having them put sockets in for the opamps.. at
least that way it’s easy to change. And the opamps will go in last.

Any thoughts/ ideas would be most welcome,

It is generally good practice to use a resistor to connect an op-amp input
to ground. The value should be similar to the impedance of the other input
circuit, so for IC1A through IC3A about 5k would be good. This is mostly to
balance the offset voltage due to input current, but it may also avoid
having high current spikes through the input protection diodes to the supply
rails. Another source could be the direct connection from the AD633s to IC2B
and IC3B. Also, as others have pointed out, the reset switch may cause a
current surge from the capacitor discharge. A 10-100 ohm resistor might be
wise.

I use the OPA2277UA (SOIC8) in my Ortmaster product and I have not had any
failures.

The spec sheet indicates that the inputs have internal 1k series resistors
for the diode clamps, and also bias cancellation circuitry which makes the
bias cancellation resistor unnecessary. But I still think it's a good idea.

Paul

 

[George Herold]

   Youi're leaving the caps open, allowing them to charge to what
ever.

  It's been a while since I've looked at the specs on that op but
something tells me you could be over doing it on the input, depending
on what the other input is doing...

  Personally, I would put high value bleeder R's crossing the switch, at
least this way the caps are more than likely to be close to the existing
value when switched in.

  You should be able to select proper values that won't conflict with the
current values selected. 1 MEG. 5 MEgs etc..

Jamie- Hide quoted text -

- Show quoted text -

So the opamps arrive blown. (or blow on the first power up... but
never on any power up after that once they have been replaced.)

Not much room for more parts, and adding those R's would 'screw-up'
the math a bit.

I've done several switched cap. circuits, I've never had a problem.

Say, how big a voltage can a 1uF film cap get with one side open?
(would it be different for a ceramic cap?)

George H.

 

[P E Schoen]

"George Herold" wrote in message

So all of the time it's IC2 (U2) the Y opamp that has been fried, but
once the Z opamp was also fried. The Y opamp does have the biggest
feed back R.

I'm done testing the latest batch, there'll be more coming in a few
weeks or so, waiting on a part. I'm guessing it won't happen again...
Some one time ESD event that took out a handful? (doesn't explain why
it's IC2?)

IC2 is the op-amp that has its output connected to the reset switch to GND.
It may be able to withstand a short to GND, but it seems like poor practice
to short an output for reset.

Also, a high precision low power Op-amp is likely optimized for other
parameters and not so much for ruggedness.

Paul

 

[George Herold]

"George Herold"  wrote in message



IC2 is the op-amp that has its output connected to the reset switch to GND.
It may be able to withstand a short to GND, but it seems like poor practice
to short an output for reset.

Also, a high precision low power Op-amp is likely optimized for other
parameters and not so much for ruggedness.

Paul

Yeah I know... I had a different opamp in there to start with and then
was confronted with this big DC offset. (note the trimmer pots on the
multipliers*) The circuit has a gain of ~200 or so, and a millivolt
of DC offset was a problem. This opamp has more noise than Phil H
would like, (8nV) certainly part of that is the 1 k ohm R's in the
input protection circuit.

George H.


* They were added after the first spin, I never get my pcb's right on
the first try. (I haven't used a trimmer pot in more than ten
years.)

 

[[email protected]]

So I’ve got this Chaos circuit.https://www.dropbox.com/s/afy4kb63ftdo1ke/chaos2.pdf
(I just installed dropbox.. I hope that works)
Basically two multipliers and three integrators (with gain and reset)

I’ve tested about 20 of them so far and in 3 or 4 of the 20 one or
more of the opa2277 opamps has been fried.  Now all the production
people wear grounding straps when stuffing a PCB.  And we use this
opamp all over the place and I’ve never* seen any problem.  So I’m
trying to figure out what’s going wrong.

(Finding the blown opamp in this circuit is a bit challenging.
There’s no real input and output, it’s just a  multi-connected
loop.)

Now the opa2277 data sheet warns about too much input current.  (must
be less than 20mA)  I’m wondering if the stuffing order of the board
could be an issue.  Could there be some charge on the integrating caps
that gets pushed through the inputs as the pcb is stuffed?

For the moment I’m having them put sockets in for the opamps.. at
least that way it’s easy to change.  And the opamps will go in last.

Any thoughts/ ideas would be most welcome,
George H.

*that’s not quite true I had a similar issue with an opa2277 used in
the PID section of a temperature control board.. these were made by a
board stuffing house and in two (out of ~100) the opamp was also
fried.

From the other comments it really sounds like bad parts. But, if not,
is there any possibility of ESD via the body of SW1? Do any of the
"good" units' U2's show increased Vos? And, on the bad units, are the
inputs fused/shorted/open, etc?

The answers to those sorts of questions might narrow the
possibilities.

 

[George Herold]

From the other comments it really sounds like bad parts.  But, if not,
is there any possibility of ESD via the body of SW1?  Do any of the
"good" units' U2's show increased Vos?  And, on the bad units, are the
inputs fused/shorted/open, etc?

The answers to those sorts of questions might narrow the
possibilities.

--
Cheers,
James Arthur- Hide quoted text -

- Show quoted text -

Hi James, I'm afraid I don't have much good data. Each integrator
has a gain of two section in it's other half. And for those I can
look at input and output to see if they agree. In two of the cases
the output was railed. (I don't recall which rail.. sorry) Opamp was
riped out, replaced and no problems. (I'm getting fast at clipping,
pulling leads, solder sucking, cleaning and reinstalling... Through
hole parts are easy that way.)

For the other two (of four) bad ones I just ripped out U2 and put in a
new one, in one case I had to replace the Z opamp also... maybe U2
wasn't blown??? Mostly I've been just trying to get product out the
door.

Thanks for everyone's ideas... (I'm still scratching my head.)

George H.

 

[George Herold]

What's the weather like there? Not to start a weather thread, but are
you getting ESD zapped much lately?

--

John Larkin         Highland Technology, Inc

jlarkin at highlandtechnology dot comhttp://www.highlandtechnology.com

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro   acquisition and simulation- Hide quoted text -

- Show quoted text -

No very hot and humid here, (70's), till today... (freeze warning for
tonight)

George H.

 

[George Herold]

Yeah, this can be quite frustrating.  My experience with the grey market
OP275's was a bunch of boards with 25 op-amps per board.  Power was
applied, and then you ran your finger down the rows of chips, looking for
a hot one.  Replace all the hot ones, then maybe one more that didn't
get hot but didn't perform correctly, and the board worked.  Gave me
the willies, but those boards are still running in the field!

You might rig up a pair of resistors
to the power supply for the next batch, so you can detect amps that are
fully blown without applying full power to the board.  Maybe put an
LED-resistor across the current limiting resistor, so you get an immediate
indication of a shorted op-amp.  This would pretty much prove that the
amps were blown before you got the board.  You are having these made
at a contract manufacturer?  No telling how good, or bad, their
ESD procedures are.  If some of the parts are through-hole, it is possible
they are getting charged up by handling or maybe test procedures before
assembly.  Geez, somebody takes 50 V caps out of the box and checks them
on a 50 V power supply for leakage, then immediately installs them on
the board.  That would do it!

Jon

Well there are a bunch of different circuits sharing the same power
supply. I might be able to tell if there's a lot of current draw.
But maybe not worth it. I'll just test the chaos bit first.

George H.

 

[halong]

No, it arrives on my desk for testing already blown.  I can't make a
working circuit fry the opamp.  (I tried all sorts of switch flipping,
power on /off cycles.. no problem.)




Well S1 resets the 'thing'  You can then watch how the signal changes
with time from a given starting point.
SW1 just changes the time scale for the circuit.  It's easier to see
on the 'scope with a fast time scale, but slower has less noise and
it's easier to find the period doubling points.

One thing that does stink about digital 'scopes is the X-Y mode.










Grin... It was easier just to dump it out of eagle than to try and
redraw it.

George H.

opam blown b 4 installation ? could it be counterfeit ?
how many suppliers you hvae on this part?

so, blame the supplier