Changing Diode DC Blocking Voltage

[b2theory]

Has anyone seen the blocking voltage on diodes change (reliably) in the field?

[bob91343]

no

[Gyro]

I suppose avalanching a non avalanche rated diode could change it's reverse breakdown voltage, but not predictably (unless it goes to zero!).

You need to provide some explanation of the reason for your question if you expect a sensible answer.

[b2theory]

I suppose avalanching a non avalanche rated diode could change it's reverse breakdown voltage, but not predictably (unless it goes to zero!).

You need to provide some explanation of the reason for your question if you expect a sensible answer.

It's a 200V schottky diode used in a buck regulator. Without getting too deep into the weeds about the end use, the diode is substantially over spec'd for the circuit. We are noticing anomalies in the regulator output on a few return units. During our investigation we put the diode on a curve tracer and found that it was breaking over at 300V. We also know if you touch the "malfunctioning" diode in circuit the regulator output changes.

I have seen papers showing the over drive of certain power MOSFETs gates can cause the Vth to drift without resulting in a catastrophic device failure. I was wondering if anyone was aware of an equivalent phenomenon in diodes.   

[amyk]

A diode rated for 200V is unlikely to go into reverse breakdown at exactly 200V, unless it is a zener. In fact it is normal for the actual value to be higher to accommodate natural statistical process variation. The results here show that the common 1n400x series may all be the same internally (and the actual breakdown voltage is much higher than the spec):

https://www.eevblog.com/forum/projects/testing-1n4007-with-a-megger/

[TimFox]

In production, no one measures the actual breakdown voltage (specified as voltage required for given reverse current), except for Zeners.  Instead, the ATE verifies that the current is below the specified level at the guaranteed breakdown voltage.  For a given diode, the breakdown voltage is higher than the specified value and can be much higher.

[b2theory]

A diode rated for 200V is unlikely to go into reverse breakdown at exactly 200V, unless it is a zener. In fact it is normal for the actual value to be higher to accommodate natural statistical process variation. The results here show that the common 1n400x series may all be the same internally (and the actual breakdown voltage is much higher than the spec):

In production, no one measures the actual breakdown voltage (specified as voltage required for given reverse current), except for Zeners.  Instead, the ATE verifies that the current is below the specified level at the guaranteed breakdown voltage.  For a given diode, the breakdown voltage is higher than the specified value and can be much higher


I'm aware of all of these things. We are comparing failed/changed diodes against known good of the same part. V_R, V_F, and capacitance at 0V have all changed, but the device(s) are still largely functional. We believe transients my be an issue, but I was hoping that the observed phenomenon might have been seen by someone else before. I'm just pulling on all of the available threads.

[TimFox]

I am confused about the problem you report.  If I read your second post correctly, you removed the diode in question (a 200 V rated Schottky) from a malfunctioning circuit and found that its breakdown voltage measured about 300 V.  We pointed out that was not inconsistent with a manufacturer rating of 200 V.
To see if the circuit caused a change in the breakdown characteristics of the diode, you could carefully measure the actual breakdown of a new part, install it in the circuit, induce the failure, and then re-measure the part.  Comparing against another part will not give a conclusive result.

[thm_w]

OP is saying it has changed but not actually provided the numbers.
b2theory: can you provide Vf, Vr of new parts and failed parts?

What temperature does the diode reach when the supply is run at 100% load?

"We also know if you touch the "malfunctioning" diode in circuit the regulator output changes. "
So a good unit does not change, change by how much then? Bad solder joint? Problem somewhere else in the circuit?

I have seen TVS diodes voltage drop over time, but that was failure scenario due to overload.

[ANTALIFE]

What sort of diode are we talking about OP? Standard, Schottky, Zener, TVS...

Either way for the breakdown voltage to increase (from say 200V to 300V), that means that the semiconductor doping levels have somehow changed. Which would be a very strange thing to happen in the field as the package is sealed fairly tight, plus to change the doping you would also need elevated temperatures (like in the 500-1000°C range)

So I guess the next couple of question are:

• Is the device used in an extreme environment? Radiation/Space comes to mind
• Are you measuring the breakdown voltage while diode is in circuit or removed?
• Have you contacted the diode manufacturer about typical tolerance on breakdown voltage? I know that it varies, but am not sure by how much without digging into some app notes

[bdunham7]

It's a 200V schottky diode used in a buck regulator. Without getting too deep into the weeds about the end use, the diode is substantially over spec'd for the circuit. We are noticing anomalies in the regulator output on a few return units. During our investigation we put the diode on a curve tracer and found that it was breaking over at 300V. We also know if you touch the "malfunctioning" diode in circuit the regulator output changes.

What sort of voltages are you seeing in the circuit?  How about V_f ? Does your circuit actually depend in some way--intentional or not--that the Schottky diode in question exhibit a certain amount of leakage at high reverse voltages?  If 'touching' the diode changes the operation of the device, perhaps capacitance is an issue.  Have you measured the capacitance of a diode from a failed unit vs a new one?

I strongly suspect your issue is elsewhere and that even if a small or random change in the characteristics of the diode have an effect, that is because of instability elsewhere due to either design or component malfunction.  No reasonable design would fail just because a diode wasn't precisely to spec, and certainly not because the reverse breakdown voltage was over spec.  If your device isn't top-secret, feel free to post a lot more details and I'd sure you'll get better advice.  Don't worry about getting deep in the weeds--you are probably already there. 

[T3sl4co1l]

Bogey parts can have ultimate ratings far in excess of their datasheet values, for example early BJTs that fit a 60V spec, say, might actually break down anywhere from 80 to 200V+.  This is not a change thing, it's a characteristic of the part; just that no one might've tested the part before (at least after initial testing, which is a simple go/no-go test, not a characterization).

I think modern parts tend to be tighter grouped, but it wouldn't surprise me if there are schottkys out there rated for say 100V and breaking down at up to 200.

There's also the possibility that it's completely the wrong part, whether by accident or counterfeit.

We also know if you touch the "malfunctioning" diode in circuit the regulator output changes.

Presumably it has a ton of AC applied to one or both sides?  Without further information, this is easily explained as the result of the EMI going up your finger, affecting the control circuit.

Tim