LM317T voltage regulator dying

[SERGIO_372]

Hi! I was trying to make an LM317 circuit to light up some LEDs and I noticed some behavior that I didn't understand, so I hope you guys can clarify to me what's happening. The thing is... If I use an input up to 24V everything works, but above that the regulator shorts itself between the input and the output pin. If I straight up put 24V and slowly start to increase the input voltage using my adjustable power supply, it does not short. Why?

I will attach the schematic...

[wraper]

Most likely there are two factors at play. You have very low input capacitance but high output capacitance (the proper place for 220uF caps is input). Which means once voltage is applied, input-output pins of LM317 see a full input voltage between them until output capacitors charge. Which is not good but something like 25V should not kill it as it must survive 40V. Which leads to a second factor of your LM317 most likely being a counterfeit.

[MikeK]

Dropping over 20V through the regulator is quite a bit.  How much output current?...Calculate your power dissipation.  You might be going into thermal overload protection.  And, yeah, it could be counterfeit if not from a reputable source.

[magic]

It is possible that a fairly high surge current (several amps) develops to charge the low ESR 2×220nF caps when you connect the cable. Once they reach 24V, cable inductance keeps pushing the current forward and LM317 will not absorb a lot of current, limiting itself to ~1.5A. So the current will go into charging the input caps above 24V, potentially up to 48V.

Try again with an electrolytic added at the input. If you have a scope, observe the input voltage waveform when power is connected.

[tunk]

Don't know if it influences your shorting, but R2 is a
bit on the high side. The TI datasheet says you need
a "Minimum load current to maintain regulation" of
3.5-10mA (typical/max), so a 120ohm resistor may
be better.

[wasedadoc]

You have the regulator configured to output just over 2 volts.  Why are you powering the input with 24 volts never mind going even higher?

[SERGIO_372]

Most likely there are two factors at play. You have very low input capacitance but high output capacitance (the proper place for 220uF caps is input). Which means once voltage is applied, input-output pins of LM317 see a full input voltage between them until output capacitors charge. Which is not good but something like 25V should not kill it as it must survive 40V. Which leads to a second factor of your LM317 most likely being a counterfeit.

I made the capacitance the same for input and output to see the difference, and it did not short itself. Nice.

It is possible that a fairly high surge current (several amps) develops to charge the low ESR 2×220nF caps when you connect the cable. Once they reach 24V, cable inductance keeps pushing the current forward and LM317 will not absorb a lot of current, limiting itself to ~1.5A. So the current will go into charging the input caps above 24V, potentially up to 48V.

Try again with an electrolytic added at the input. If you have a scope, observe the input voltage waveform when power is connected.

Unfortunately, I don't have an oscilloscope yet, but increasing the input capacitance seems to have worked.

Don't know if it influences your shorting, but R2 is a
bit on the high side. The TI datasheet says you need
a "Minimum load current to maintain regulation" of
3.5-10mA (typical/max), so a 120ohm resistor may
be better.

I didn't see that spec. I will give it a look later.

Dropping over 20V through the regulator is quite a bit.  How much output current?...Calculate your power dissipation.  You might be going into thermal overload protection.  And, yeah, it could be counterfeit if not from a reputable source.

You have the regulator configured to output just over 2 volts.  Why are you powering the input with 24 volts never mind going even higher?

I know it's not the best scenario, in reality I will use less than 10V as input and the output current will be lower than 150mA. I was just poking around and noticed that behavior that and I couldn't understand.

[Siwastaja]

If input capacitors are ceramic, overshooting from fast input voltage rise time because of the LC circuit formed by the wiring inductance + input capacitance can be the issue. See https://www.analog.com/media/en/technical-documentation/application-notes/an88f.pdf . You can try to add any random electrolytic capacitor in parallel with the input capacitors, damping the network by its high ESR. You can also probe the input voltage with oscilloscope first, to see if there is a voltage peak.