Shunt Reg issues (TL431) found in some TRIPP LITE UPS.

Take a good one from that batch and carefully run it on a curve-tracer. There is a possibility that there is a "N" shaped peak before normal regulation voltage, where that peak gets higher as temperature goes up. That peak can be rather large WRT datasheet nominal operating point. If i remember correctly, one of the National Semiconductor references gets bad that way from 165C up.

More likely, given the failure mechanism... short, they're running too much current in the TL431's.

...Jim Thompson

That peak in the "N" curve before the operating point can be 10 times the nominal datasheet spec. So..assume the peak is low or absent at 25C, the supply powers up, and after X hours of use gets warm enough to have a large peak. Continuing, assume the supply is shut off and there is a (relatively) large capacitor across the TL431. As the capacitor discharges, the TL431 will follow this new I-V curve and result in a large current which either degrades or destroys the device. Without knowing if the particular TL431s exhibit this temperature sensitive "N" characteristic and without knowing the temperature that peak starts to grow (assuming there is one) this is all speculation. However, knowing one possible mechanism might be useful in tracking down a possible cause of damaged devices. Be aware that every manufacturer of regulators of this type makes more than one version and that EVERY version has a different characteristic, and that if the process is changed, start over again. Processes may creep over time and will BE DIFFERENT if a new fab is used to make the "same part".

You completely lost me there. What is this "N" curve of which you speak? The TL431 is a SHUNT regulator (bandgap-style).

...Jim Thompson

Yes I'm lost too!

I use this part a lot so the thread got my attention. I've looked through all the 431 datasheets I have and I have several and I can't find any "N" Curve. I've always found them to be pretty robust within reason. I use mainly TI's and Nationals.

Are you talking about stability curves versus load capacitance?

STMicro is OnSemi? AIUI TLxxx is originally a TI part number

Shunt regulators are very sensitive to overcurrent. Unlike 3-terminal voltage pass regulators, which almost universally have foldback shutdown, the shunt regulators do not have that kind of shutdown.

Spikes etc. can cause brief overcurrents that can very effectively take out the TL431's or other shunt regulators while causing little or no damage anywhere else. I think that effectively, the shunt regulators "sacrifice" themselves by taking the brunt of the hit and end up protecting the other stuff.

Tim.

The electrical environment inside a UPS switching from "standby" to "active" can be quite nasty and possibly even worse than what mother nature is introducing. If something goes wrong in the sequencing, BANG, lots of smoke comes out!

TL431 originally a TI part. I think the L nominall stood for "linear" although as time went on TI started preceding their UARTS and other digital parts with TL too.

Second sourced by everyone else under the sun since then. Motorola became On Semi (except for the parts that got sold to other folks.)

Tim.

OK; put it on a curve-tracer; nominal IV curve shows no current until the rated voltage where the current goes up rapidly. Say the rating is 1.2V for discussion; the Esaki-diode like "N" curve i am talking about happens near 0.6V and happens only for a few makes and models and shows near 165C on the particular make and model i was talking about. That peak obviously is absent below the "critical" temperature, but starts to show near that temperature, and rises rapidly as the temperature of the device increases (a small amount). Does this help?

This action i am talking about will NEVER be found in a datasheet. First, if you noticed, i mentioned that for the part that i observed this particular phenomenon, that the "N" peak started near 165C which is a bit above the max temp rating of (almost) any semiconductor device. This has nothing to do with stability curves. See my other response, please.

That is another possible failure mechanism that might have happened in the UPS.

I have posted some scope curve tracer screen shots showing what i mentioned, in ABSE using the same thread title as this one.