TL431 voltage reference on more than 36V (to generate 2.5V reference)?

[Lupin III.]

I have a circuit that's powered by approx. 50V DC (step down DC-DC converter module). I this circuit I'd like to add a TL431 voltage reference for some shenanigans with the current feedback to add external current control.
Assuming the supply resistor is sized properly to keep the TL431 within its allowed current range even in the worst case of >50V, the 431 should be fine once it is regulating, because the resistor is dropping all the voltage. But is the 431 fine at switch-on in the few µs before it starts properly regulating? There the full voltage could be at the cathode pin (assuming there's almost no current flowing and therfore almost no voltage drop on the supply resistor). The obvious solution would be to put a zener diode in parallel, but I'm just wondering if it is necessary.

[StillTrying]

Assuming the >50v takes a millisecond or more to ramp up to >50V I don't think it would be problem for the TL431.

[Zero999]

I agree, the transient shouldn't be a problem. Most semiconductors are designed to withstand these kind of transients without damage.

For your information, another transistor can be added to the TL431 so it can be used at higher voltages.

https://www.eevblog.com/forum/projects/tl431-common-base-amplifier-stability/msg599580/#msg599580

[Doctorandus_P]

Also:
When you put a small capacitor parallel to the TL431 you create a RC constant with a ramp up time, which ensures the TL431 has enough time to get into regulation, but it's probably not needed.

[Whales]

Also:
When you put a small capacitor parallel to the TL431 you create a RC constant with a ramp up time, which ensures the TL431 has enough time to get into regulation, but it's probably not needed.

This could be an interesting solution.   A tiny cap + R across the TL431 might guarantee that it will be at 0V through turn-on.  High-freq transients might be another issue, not sure.  You have to make sure the capacitance value you use doesn't make the TL431 unstable (check the datasheet - by memory there is some variation between TL431 vendors).

I like the option of an approximate voltage regulator: use an NPN in an emitter follower configuration, base controlled by a voltage divider.  4 parts: two resistors, a capacitor and one NPN.

[David Hess]

Going by the datasheet, there will not be a problem because the TL431's internal compensation causes it to slowly rise to its set output voltage.

[Zero999]

Going by the datasheet, there will not be a problem because the TL431's internal compensation causes it to slowly rise to its set output voltage.

Yes, that's true, when power is first applied, the compensation capacitor will charge through the base of the Darlington output stage, immediately turning it on, clamping the output voltage to a couple of volts at most.

[Lupin III.]

Thanks a lot for all the input! Good to know that the 431 is pretty much safe at any voltage. At least given a properly selected supply resistor, that can handle the power dissipation. Anyone daring to create a 5V supply from rectified mains? Non-touchable of course

[Whales]

Thanks a lot for all the input! Good to know that the 431 is pretty much safe at any voltage.

I'm still a bit suspicious.  This would mean the max voltage figure can be ignored.  But then why would they put it in?

[magic]

In TL431 the cathode is not tied to the reference comparator input like in typical "reference diode" ICs. With a suitable divider, one can set it up for any voltage, like 100V. Or it can be used in a myriad other ways altogether. But it won't work very well if the cathode breakdown voltage is exceeded.

As for the power-on transient, I think it's only a matter of rise time of the power supply. If we pushed it down to nanoseconds, the output stage probably wouldn't turn on fast enough to protect the chip More practically, I'm not sure where the safe limit lies but it's probably still much less than 1µs. Also, whether such brief avalanche causes any harm is another question.

[Zero999]

In TL431 the cathode is not tied to the reference comparator input like in typical "reference diode" ICs. With a suitable divider, one can set it up for any voltage, like 100V. Or it can be used in a myriad other ways altogether. But it won't work very well if the cathode breakdown voltage is exceeded.

As for the power-on transient, I think it's only a matter of rise time of the power supply. If we pushed it down to nanoseconds, the output stage probably wouldn't turn on fast enough to protect the chip More practically, I'm not sure where the safe limit lies but it's probably still much less than 1µs. Also, whether such brief avalanche causes any harm is another question.

I think it will be difficult to get such a short rise time, as long as a suitable current limiting resistor is used, because the parasitic capacitance of the TL431 will slow it down.