3
\$\begingroup\$

In the following (partial) circuit, the transistor is usually in its regular forward active mode. However, when the supplies collapse abrubtly - e.g. during power-down - the base-emitter junction can be transiently reverse biased with more than the allowed 5 V, due to the slow discharge of C1.

Do I need a protection diode from the base to one of the supply rails ?

I am wondering because C1 is fairly small and e.g. MOSFETs and diodes can handle avalanche breakdowns with some energy without bad consequences. This answer also suggests that only "sustained" reverse breakdown has bad consequences.

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
3
  • \$\begingroup\$ In my opinion, a diode from the base to 14V rail wouldn't hurt. Not to 15V, because the diode to Vccb will actually keep voltage difference between base and Vccb limited to diode voltage drop during power-down. But don't take my word for it without anyone else confirming \$\endgroup\$
    – Ilya
    Aug 30, 2021 at 14:08
  • \$\begingroup\$ Will \$V_{EB0}\$ or \$I_{EB0}\$ be exceeded? \$\endgroup\$
    – Andy aka
    Aug 30, 2021 at 14:12
  • \$\begingroup\$ @Andyaka V_EBO for sure. I couldn't find I_EBO in the datasheet. Only the max base current (200mA) which will be not exceeded. Current depends on the rate of supply collapse. Base current will be C1 * fall rate. so for 10V/ms -> base reverse current will be about 1 mA discharging C1 into the collapsing VCCB rail. \$\endgroup\$
    – tobalt
    Aug 30, 2021 at 16:42

2 Answers 2

5
\$\begingroup\$

You will be exceeding the absolute maximum voltage on the datasheet (5V). Unless you feel like doing a lot of testing and characterization on many samples from different manufacturers under a variety of conditions in order to reduce the risk, a diode may be the least expensive safe solution.

If the sole pull down is that 1M then you can’t stand much beta degradation.

You could perhaps also consider replacing the BJT with something like a BSS84 MOSFET that has no problem with up to +/-20V on the gate- and has a compatible pinout.

\$\endgroup\$
1
  • \$\begingroup\$ good point about the P-MOSFET. might be simpler/cheaper than two BJT or BJT+Diode. \$\endgroup\$
    – tobalt
    Aug 30, 2021 at 18:37
4
\$\begingroup\$

It depends on the conduction "edge effects" and storage energy of capacitance on both sides and including the junction.

Breakdown can be either soft (Zener effect) or hard (Avalanche effect).

The soft effect is the thermal damage from heat rise in the junction. In this example power is limited by MPT at 50% of $V^2/R= 0.5*15^2/100k= ~1 mW which even at 0.25'C/mW for a TO-92 is harmless.

The hard effect caused by conductive filaments of nanoscale particles causing the avalanche effect that now discharge the energy stored across the junction and can short circuit it, unlike a fuse, merge the conductive particles by spreading heat.

Leakage current doubles every 10'C due to the Arrhenius Effect which reduces the effective resistance.

Since there are no specs for the amount of charge energy (E=1/2CV^2) that may be dumped into a reverse BE jcn, it is unwise to assume anything, but verify all suppliers of such parts with stress testing.

So although the 100k provides DC current limiting from the supply, the cap C1 base cap now may cause a high energy fusing of a nanoscale filament particles that spreads heat like a tiny lightning bolt inside the PN junction and fail. However without C1, it "might" not fail as the reverse bias capacitance is quite small and only increases as the voltage collapses towards 0. But don't assume this is generally true.

So yes it would likely fail anywhere above the -5V and the best policy is respecting the datasheet limits with a reverse diode across Vbe (preferred) or a series diode with lower leakage (less likely).

However, transistor Base Collector junctions do tend to fail with the soft effect and so can make do as high voltage Zeners at very low current levels. I did this once for 150V on my 1st colour TV and replaced a high voltage Zener with a 2N5401 (?) 45 yrs ago CE jcn.

Other

I recall reading that freewheeling diodes used to protect IGBT's can fail due to "edge effects" of the PN jcn. The desired rated breakdown voltage may have thermal induced and stray inductive induced VI power oscillations on transitions that cause soft effect failures occur early before the rated breakdown voltage and then are accelerated by nano-sized particles towards the avalanche failure below the rated breakdown voltage. This was validated in a research journal, which I recall reading.

\$\endgroup\$
2
  • \$\begingroup\$ Is there some way to estimate if the hard or soft mechanism will happen. The reverse base current will be limited to about ~mAs due to the voltage fall rate of VCCB. \$\endgroup\$
    – tobalt
    Aug 30, 2021 at 16:45
  • \$\begingroup\$ With C1 , an avalanche effect is more likely, but I don't know how to predict any part without test analysis. Perhaps a chip designer will know. \$\endgroup\$ Aug 30, 2021 at 17:05

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.