In the below circuit, the diodes protect the LM317 against both of the caps discharging into the IC when input or output is shorted. LM317 is itself capable of having it's output shorted without damage. But the input-output voltage difference must not exceed 40V. In my circuit, I'm applying 45V to the input, and getting 30V from the output. In normal conditions this means a 15V difference over the IC and everything is fine, but if the output gets shorted somehow (maybe because of the initial charging of large capacitors connected to the output), I get a full 45V above the IC and this is enough to damage it according to datasheet (it's above the 40V maximum).

What can I do to prevent this, without using another IC like LM317HV?


  • \$\begingroup\$ What's wrong with the LM317HV? (Hmmm I don't see any in the TO-220 pac on DK. I hope these haven't gone obsolete.) \$\endgroup\$ Jan 9, 2015 at 17:37
  • \$\begingroup\$ @GeorgeHerold: The LM317HV is surely available in TO-220 at least from TI (and stocked by RS). I suspect the main issue is/was price (relative to its little brother) and not wanting to buy a less commonly available part just for this project. \$\endgroup\$
    – Fizz
    Aug 29, 2015 at 20:16
  • \$\begingroup\$ What you may have more trouble with is finding the LM337HV if you want a dual rail +/- linear power supply at more than +/- 40V input... TI in their wisdom obsoleted this part. You can get by with two LM317HV and an extra bridge rectifier though... as long as you have twin-secondary (not just center-tapped) transformer. \$\endgroup\$
    – Fizz
    Aug 29, 2015 at 20:57

2 Answers 2


One simple option : replace D1 with a 30-ish volt zener rated for several amps, and a fuse (upstream of Vin) rated to blow before the zener does.

Normal operation : the zener should not normally conduct.

Power-off operation: the zener is forward biassed providing the same protection D1 does.

Short circuit operation : the 317 goes into current limit and the output voltage drops. The zener conducts before I/O voltage exceeds 40V. The fuse blows. This does depend on the rating of the zener diode. It must be chosen carefully, or a more complex arrangement adopted (essentially the zener diode supplies an emitter follower) as in the other answer.

Power-on operation : vith 45V in, the zener will supply current to the load as long as the LM317 is supplying less than 15V. After that, the LM317 takes over.

  • \$\begingroup\$ What about the power-on? Would a zener here prevent the regulator output from rising to intended levels by supplying current to load itself? Also wouldn't it blow up quickly by the short circuit current before the fuse blows up? \$\endgroup\$ Jan 9, 2015 at 14:32
  • \$\begingroup\$ I accepted this, but I'm going to experiment with it a little. I'm not sure if a regular zener will be able to handle these transient currents. May need to use a TVS instead of a zener. But nevertheless it seems to be the simplest solution. \$\endgroup\$ Jan 9, 2015 at 20:29
  • \$\begingroup\$ Oddly enough, this solution appears in the datasheet of the LM317L (ti.com/lit/ds/symlink/lm317l-n.pdf) right on the same page as the other one below, but this simpler one is not in the datasheet of the regular LM317. \$\endgroup\$
    – Fizz
    Aug 29, 2015 at 21:11

When using a voltage regulator like this in an elevated voltage configuration the output capacitor charging issue will always present this momentary issue of over stress of the voltage across the regulator. In your particular case the solution to the 45V across the regulator would take extra components and circuit additions that may very well add up to the cost of using the higher voltage rated regulator in the first place.

If you need to use the lower voltage rated LM317 then you could add a circuit similar to the following. This design keeps the input to output differential of the LM317 to less than the zener voltage.

enter image description here

The above circuit comes from this TI application note which you can access for more information.

  • \$\begingroup\$ This is a nice variant on my suggestion. Instead of blowing a fuse, Q2 and R3 dissipate power. Note that for the zener diode shown, some change to the value of R3 is needed, if output current over 30mA is required. \$\endgroup\$
    – user16324
    Jan 9, 2015 at 13:32
  • \$\begingroup\$ But here Q2 needs to dissipate too much additional power all the time and will need a heatsink, if I understand correctly. \$\endgroup\$ Jan 9, 2015 at 14:36
  • \$\begingroup\$ Actually the power in Q2 may be a benefit to you as it takes some of the power burden off the LM317. They are in series after all. \$\endgroup\$ Jan 9, 2015 at 14:40
  • \$\begingroup\$ Also if the input to output differential is less than the Zener voltage the Darlington transistor pair would be saturated to around a 1V drop across the Q2. Only when the Zener starts to clamp does the voltage across the Q2 start to rise. So in your case of trying to solve the problem of charging the output capacitor you would have have the increased drop on Q2 whilst the cap was charging. \$\endgroup\$ Jan 9, 2015 at 14:47
  • \$\begingroup\$ You're right about that, but I don't see any benefits of using this over the zener solution. \$\endgroup\$ Jan 9, 2015 at 20:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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