MC33202 op-amp failures

d

=A0

,

After 5 years I would look into the effects of worn-out batteries on the circuit's performance.

-- Joe

On a sunny day (Wed, 9 Sep 2009 11:17:50 +0100) it happened wrote in :

No decouplng caps opamp supply, and anywhere?

d

=A0

,

It might be wrth looking at the 1R resistor. Some wire-wound parts can be inconveniently good inductors.

If your purchasing department has recently found a new - cheaper - supplier, as they have a habit of doing, it might be worth looking into,

-- Bill Sloman, Nijmegen

The opamp is decoupled by C521. There is a common mode choke and another 100n cap next to the DC input connector.

All parts are surface-mount. The current-sense resistor is a 2010 0.5W type, should be a few nH at most.

The PSU is shipped with the unit, and uses a proprietary keyed connector

- this ought to rule out overvoltage/reverse polarity. The only other possibility I could think of was mains-borne surges / transients. The system (with PSU) has been through the EN61000 fast transient and surge tests with no problems.

With such small amounts of capacitance on the input supply along with the inductance of the common-mode choke, it's very possible that any transients on that input supply are generating large voltages at the opamp supply.

Bob

--
== All google group posts are automatically deleted due to spam ==

On a sunny day (Wed, 9 Sep 2009 13:52:08 +0100) it happened wrote in :

Yes, but maybe that output circuit could stil oscillate, when the battery ages?

One thing that doesn't 'spice' very well is turn-on and turn-off behavior. Take another look with/without battery, while fiddling with the enable line, or with intermittent 9V/return, battery/return, logic supply/return. Intermittent grounds can often drive whole sections of circuitry to look unexpectedly negative.

You also might review fab dates for correlation - recent fab processes for potential soldering/handling damage, storage for MSL maintenance etc etc.

RL

=A0a DC

either

Good points. Those PSUs may be starting to act up after 5 years.

-- Joe

The DC-in connector is a 6-way circular DIN - we deliberately steered clear of the common 2.1mm/2.5mm barrel-type as it's too easy for the user to plug in a different PSU with wrong voltage / polarity. The battery pack is internal and the user has no access. It connects to the PCB using a JST wire-to-board connector. I'd have thought the chances of these connectors getting swapped was fairly minimal...

The DC-in plug *can* be forced in upside down by a really ham-fisted user - but the only effect of this is to bend one of the +ve pins into contact with the shield before any of the other pins mate, which shuts down the PSU until the short is removed. We have had a few instances of this, but they don't correlate with the op-amp failure at all.

Monster FET's. Q521's fast switching thru Q524's gate capacitance. Try a series R in the gate of Q524.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
I love to cook with wine.      Sometimes I even put it in the food.

Can you replace the decoupling cap C521 with a larger capacitance of same footprint, say 1uF? 100n isn't much. Besides power supply spikes the other risk is during insertion of the DIN connector when 9V-in mates first and there is some static charge riding on stuff.

A TVS would be another option but iffy in this case because they aren't quite precise enough for the few volts between your supply and abs max of the MC33202. Counterfeit chips on a cheap part like this are unlikely. Do your purchase from reputable suppliers?

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

Now we're getting somewhere... the enable and Hi/Lo inputs are driven by a micro. The 'Lo' setting is only used if the off-load battery voltage measures less than 4V (i.e. severely discharged), to limit dissipation in Q523. This doesn't happen very often in normal use. So, measured Vbatt crosses threshold -> Q521 gate goes high -> Drain goes low ->

current spike from U521 pin 1 thru Q524 Cgs and Q521 Rds. Could be a couple of amps worst-case. Ouch. Am I on the right track? I could test this on a scrap board with some code or a sig gen which toggles the hi/lo pin. Best wear goggles...

Thanks R.

OK, I've just spiced this and there *is* a current spike of nearly an amp, but it lasts all of a few ns - this isn't likely to blow anything at a rep rate of maybe a couple of times a day, surely? Or am I missing something? As I said earlier, this circuit is being binned shortly but I would like to get an understanding of the failure mechanism.

I make it a policy to avoid, whenever possible, connecting any non-ground pins of any semiconductor to the outside world.

Some switching supplies do nasty things if the input is teased. One of our customers was blowing up our VME modules in one particular British-made VME crate. We got a crate and tested it. If you interrupt the power for just the right amount of time, about one second, the +5 supply will spike up to +9 at turnon. The crate had a crappy push-push power button that encouraged finger-fumbling at turnon and turnoff. We told them to replace it with a big toggle switch, but they elected to scrap the crates.

John

The cap is an 0603. I'll see what I can find in this size, I try to avoid the upper end of the capacitance range for a given size because of the awful voltage coefficient.

Regarding the ESD issue, the DIN connector pins were tested to 6kV during pre-compliance work (unit operting on battery, PSU disconnected) with no apparent ill-effects afterwards. I'd expect the input MOSFET U522 to suffer damage first. Aren't old bipolar op-amp designs fairly robust against ESD?

The best (and fastest) fuse in the world is the metalization on a chip... they may be designed for as little as 10's of mA.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
I love to cook with wine.      Sometimes I even put it in the food.

Presumably this means TVS's, clamp diodes, varistors, or at the very least a small RC filter on every line? The other 33202 circuits on our board are a lot better protected in this regard, it's true.

That does sound nasty. The system, with PSU, was put through the IEC

61000-4-11 test which subjects it to short interruptions, brownouts, etc. Presumably the VME PSU was put through this as well, so it obviously doesn't catch everything! I'll try a more thorough 'toggle test' tomorrow.

If it does turn out to be the PSU that'll be another nail in its coffin

- we had to spec a better PSU for our higher-end system as a few more feet of connected cables pushed the radiated emissions over the limit. The specs and standards are almost identical on paper, but the new one is nice and quiet. Virtually no difference in price either...