Currently I’m working on my first electrical project and having a few issues with the negative side of the supply, it is not adjusting the same as my positive side, and I think the LM337 is blown?

I’ve attached a photo of the circuit, which consists of an LM317/337 with 2 PNP/NPN's in parallel for boosted current.

There is no capacitor on the adj pin, and no output capacitors currently so there are no bypass diodes in place as recommended from the 317/337 datasheet. Although I do plan to add them in the future.

EDIT: I’ve tested the npn transistors on the negative side and they are good. R5 and R6 also have a potentiometer following them to limit the minimum voltage to 24 volts with the maximum at 49 volts.

If there are any recommendations to improve this circuit in any way I'd love to hear them. I’ve seen so many different LM317 current boosted circuits that I decided to try and design my own instead.

EDIT 2: After doing a bit of research I was able to find another person referencing issues with a current boosted LM337.
Using a low-esr, 220µ cap at the output solved the problem completely. But their solution was adding a 12uh inductor in series with a 2.7 Ω resistor on the emitter of the NPN.

Apparently ouput/adj pin capacitors are critical in a current boosted lm337 circuit.

lm317/337 dual voltage power supply with current boost

  • \$\begingroup\$ Am I the only one to think that 49V and LM317 / LM337 don't play well together? Paralleling BJTs has its own tricks as well \$\endgroup\$
    – fraxinus
    Apr 1, 2023 at 6:55
  • \$\begingroup\$ (@fraxinus: With no "choke input", no (minimum) load specified and about 1 A, I harbour doubts - about dissipation as well as voltage. The voltage to rate is output - input: if input was 52-53 V, it might work.) \$\endgroup\$
    – greybeard
    Apr 1, 2023 at 7:42
  • 1
    \$\begingroup\$ You want a symmetrical PS adjustable from 24 V to 49 V, 10 A? Even if the input to the regulators never exceeded 54 V, that is almost 30 V across the ICs. Say 7°/C (TO220 + 2°/C mounting+heat sink) and 70° allows 10 W tops - 0.33 A, not 1.5 A. And you will need to design carefully for a transformer+bridge+filter not to exceed 64 V as well as supporting 49 V out. \$\endgroup\$
    – greybeard
    Apr 1, 2023 at 7:55
  • 1
    \$\begingroup\$ @fraxinus The LM317 only "sees" the voltage between IN and OUT, which is 53-24=29V here, less than the maximum of 40V, so it should be good. \$\endgroup\$ Apr 1, 2023 at 7:58
  • 2
    \$\begingroup\$ With 37 Vrms input to the rectifier (simulated or measured?), you will have trouble to get multiple Amperes above about 42 V. What is the point of the exercise? If it isn't gaining knowledge about obsolete PS designs, pick a different circuit altogether. Boosting linear regulator output current with an active bypass was useful for small factors, only. \$\endgroup\$
    – greybeard
    Apr 1, 2023 at 8:25

1 Answer 1


It's not good practice to place transistors in parallel like that. If they are even only slightly mismatched, one will begin to turn on slightly before the other, as base-emitter voltage rises. That means one transistor does all the work while the other does very little. To mitigate that, it's common practice to employ emitter degeneration, with a small resistance at the emitters:


simulate this circuit – Schematic created using CircuitLab

Notice that Q1 and Q2 are different models, with different characteristics, which causes them to share current very unequally. In the bottom circuit, I've just inserted 0.5Ω resistors, R5 and R6, at the their emitters, and you can see that they share the load much more evenly.

With resistors R5 and R6 in place, the base-emitter voltage required to turn on the transistors is more balanced, but slightly greater, so you may need to increase current sense resistor R7 slightly to turn them on sooner.

  • \$\begingroup\$ Thank you for the suggestion, the diagram I was originally using included 0.1Ω resistors but it was for an npn boosted lm317. And I was unsure of where they went for a pnp transistor. \$\endgroup\$
    – Infernoman
    Apr 1, 2023 at 20:21

Your Answer

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

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