3
\$\begingroup\$

I'm using IXGH45N120 to discharge some capacitors. The gate voltage is derived from the capacitor voltage, through some resistors and a zener. The circuit has worked fine for a long time, many hundreds of discharge cycles. We've recently had cause to look closer at it, and it appears the gate voltage is, at most, 23V when discharge starts. It drops to 20V within 150 mS.

The data sheet says the max continuous gate voltage is 20V. Max transient gate voltage is 30V.

enter image description here

I don't see any definition of what qualifies as "transient" gate voltage. I haven't heard any response from IXYS on the matter. Is there any way I can know if I'm operating within spec? Is 150 mS short enough to be "transient"?

\$\endgroup\$
  • \$\begingroup\$ They seem to state this on several of their parts that I've seen. \$\endgroup\$ – Andy aka Mar 5 '14 at 14:14
  • \$\begingroup\$ Sometimes a full databook (if they still exist!) contains explanations of terms and additional information missing from an individual datasheet. Failing that, search the manuf. website for white papers on test methodology that might answer the question. (Personal opinion : it's better to avoid running a part at "Absolute Maximum" ratings in any respect if you have an alternative). \$\endgroup\$ – Brian Drummond Mar 5 '14 at 14:32
3
\$\begingroup\$

Nooo... I would definitely not allow the gate voltage to approach that limit with 150 milliseconds duration.

Here is an ON Semiconductor application note where they include a data sheet spec of the conditions.

enter image description here

5 microseconds and duty cycle (0.1%?).

Note also the failure mechanism, which is partly thermal, so (as the lawyers say) time is of the essence.

The voltage across the dielectric, between the gate and emitter can cause tunneling of carriers through the dielectric if it exceeds the leakage limit, especially where imperfections (traps) exist. This tunneling creates heat, which if allowed to continue for a sufficient period of time, can cause damage to the oxide which in turn creates more traps. This process can quickly increase to the point where significant damage occurs in the gate oxide. It should be apparent that at higher gate voltages it is necessary to limit the duration of the event to assure that the temperature rise due to the tunneling electrons does not exceed a safe level.

If the oxide becomes hot enough to cause damage, that damage will be cumulative. This is why both the time and duty ratio of the transient event are included on the data sheet. Over a period of time this damage can cause the threshold to shift lower which may cause improper operation of the circuit or in extreme cases can cause a failure of the gate oxide.

ON Semiconductor IGBTs are tested in qualification testing, at levels well above the transient voltage rating to assure that this is a safe transient level for the gate-to-emitter voltage.

The DC rating is a very conservative gate voltage level and no tunneling or other degradation will occur at or below that operational level.

| improve this answer | |
\$\endgroup\$
  • 1
    \$\begingroup\$ 0.10 likely means 10% rather than 0.1% unless the notes say otherwise. Which would be consistent with switching at 20kHz. That doesn't invalidate your point though! \$\endgroup\$ – Brian Drummond Mar 5 '14 at 16:10
  • \$\begingroup\$ @BrianDrummond I agree 100%, that's the likely meaning. \$\endgroup\$ – Spehro Pefhany Mar 5 '14 at 16:14
0
\$\begingroup\$

I am not sure if they have a different definition, however, transient voltage usually refers to 'transient' (short term) spikes in voltage and I would consider 150 ms to be transient. So if the transient voltage rating is 30 volts, then you are probably operating within specs at 23 volts transient... Though, if you can find a definition provided by them as to what time span is transient, that would be better. However, as stated, you should avoid using components at their maximum rating 100% of the time anyways. This allows you more headway with spikes and in general, especially if you are unsure of transient time, and will lead to a longer life of the component.

| improve this answer | |
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.