Discrete Opamp Stress Test - What's your procedure?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I haven't used a DOA for a long time but obvious test is to look at current draw to confirm internal operating bias currents, then look for THD driving a decent load to full scale.

If you are building these for a specific application, perhaps test them in that application.

JR

PS: Back in the 70s when I still performed 100% testing on IC opamps I would put all four sections of a quad op amps in series and run them at high noise gain (unity gain inverting, with like r/100 from - input to ground). Looking at noise and distortion I would cull out low single digits from a couple hundred at a time. Eventually after getting zero rejects I stopped 100% testing.  I used a zero insertion force socket when dealing with that many pieces. 

 
when I worked for an audio manufacturing company we would burn in every discrete opamp we made for a period before evening testing and measuring them.  if there was going to be a fail in what we did, it would go from the start. From there after burn in, we had a test jig where we could run them, it had test points to measure  current draw and an audio circuit to  measure it with an AP machine and listen to it if we desired.  Between the burn in and measuring, we were covered so any listening would be when testing a unit with those discrete pampas.
 
pucho812 said:
when I worked for an audio manufacturing company we would burn in every discrete opamp we made for a period before evening testing and measuring them.  if there was going to be a fail in what we did, it would go from the start. From there after burn in, we had a test jig where we could run them, it had test points to measure  current draw and an audio circuit to  measure it with an AP machine and listen to it if we desired.  Between the burn in and measuring, we were covered so any listening would be when testing a unit with those discrete pampas.
Yup weeding out infant failures. Failure curves for most electronic products follow a bathtub curve where lots of (infant) failures occur right away then level off to a low rate, until end of life when they increase again.

Back in the 80s before I went to work at a real manufacturing company. I would run (LOFT/Phoenix Audio) rack mount products in a burn-in rack overnight before bench testing. I had the burn in racks on a timer so they would cycle on and off every few hours to experience multiple thermal cycles.

At Peavey we used burn in racks for power amps/modules, but not over night (pretty much just confirmed that thermal protection and short circuit protection worked).

JR

PS I wanted to try using infra red cameras to help check power amp modules by looking at component temperature using an IR image but this was last century and the technology was not cheap and easy enough for me to get it happening. If this worked we would not only have simple go-no go test, but troubleshooting advice about what to check (i.e. this is too hot, or this is too cold).
 
JohnRoberts said:
Yup weeding out infant failures. Failure curves for most electronic products follow a bathtub curve where lots of (infant) failures occur right away then level off to a low rate, until end of life when they increase again.

Back in the 80s before I went to work at a real manufacturing company. I would run (LOFT/Phoenix Audio) rack mount products in a burn-in rack overnight before bench testing. I had the burn in racks on a timer so they would cycle on and off every few hours to experience multiple thermal cycles.

At Peavey we used burn in racks for power amps/modules, but not over night (pretty much just confirmed that thermal protection and short circuit protection worked).

JR

PS I wanted to try using infra red cameras to help check power amp modules by looking at component temperature using an IR image but this was last century and the technology was not cheap and easy enough for me to get it happening. If this worked we would not only have simple go-no go test, but troubleshooting advice about what to check (i.e. this is too hot, or this is too cold).
so it's a log curve?
Yes that has been my experience as well.  We ran our discretes on the power jig for several hours and we could power upwards of 100 at a time.  that jig took up a nice size of wall.
 
pucho812 said:
so it's a log curve?
Yes that has been my experience as well.  We ran our discretes on the power jig for several hours and we could power upwards of 100 at a time.  that jig took up a nice size of wall.
Reliability-Bathtub-Curve.png


Back while I was at Peavey we extended our warranty from 3 years to 5 years. I researched the service repair history and it was a no brainer... most field failures happen in the short term.

JR
 
Thanks guys for the response... and sorry for the late reply ;)

Did you run the DOA at their maximum supply voltage when you were testing them?

One thing I noted testing multiple DOA is the variation in DC offset from one to another. I guess better the input transistors are matched, smaller the DC offset?
 
elskardio said:
Did you run the DOA at their maximum supply voltage when you were testing them?
Seems like a reasonable thing to do, yes.
One thing I noted testing multiple DOA is the variation in DC offset from one to another. I guess better the input transistors are matched, smaller the DC offset?
Not necessarily, that depends on the circuit topology.
 
I run them a couple of volts less than max rated supply in my 'lounge room'  mix sum amp (diy build) .. 

- if they make it thru the 'winter' (6mths) of all day usage , then I believe they're good to go, in a 'lasting' sense :)
 
elskardio said:
Thanks guys for the response... and sorry for the late reply ;)

Did you run the DOA at their maximum supply voltage when you were testing them?

One thing I noted testing multiple DOA is the variation in DC offset from one to another. I guess better the input transistors are matched, smaller the DC offset?

well for us, it was a custom discrete opamp design that when in use runs off of +/-36VDC...  We often tested them at that.... For us, the design was sound, we knew what we were going to get  when built as it was all SMT, so we didn't deviate from what is in a working unit.
 
JohnRoberts said:
Reliability-Bathtub-Curve.png


Back while I was at Peavey we extended our warranty from 3 years to 5 years. I researched the service repair history and it was a no brainer... most field failures happen in the short term.

JR

Ha interesting! Thanks for sharing
 
Back
Top