FAQ
TL;DR: Building a 1.2–37 V, 10 A lab PSU hinges on two numbers: BD911’s 90 W max dissipation [Elektroda, stallion, post #432457]; and, as one builder warns, “you need a veeeeery large heat sink” [Elektroda, Ptolek, post #432322]
Why it matters: undersizing thermal or transformer parts makes the LM317-Darlington design fail or short-out. Ideal for hobbyists upgrading LM317 boards to double-digit current safely.
Quick Facts
• LM317 input limit: 40 V (LM317T) / 57 V (LM317HVT) [Texas Inst., DS]
• Safe dropout: 3 V between Vin and Vout at 1 A load [TI, DS]
• Emitter resistors 0.33–0.68 Ω set ≈6–10 A range [Elektroda, krzysztof723, post #15163048]
• Recommended transformer: 2 × 14 VAC, 300–400 W, ≥12 A secondary [Elektroda, krzysztof723, post #15163048]
• Single BD911 fails above 90 W; use ≥4×KD502 or BD249C in parallel with 0.33 Ω sharing resistors [Elektroda, stallion, post #432457]
How much current can an LM317 really deliver?
Alone, the LM317 tops out at 1.5 A [TI, DS]. By adding a Darlington pass stage (e.g., BD911 + 4×KD502) and 0.33–0.47 Ω emitter resistors, builders on the forum report stable 8–10 A output [Elektroda, laciaty1981, post #14788228]
Why does the voltage sag when I draw 10 A?
A single series transistor adds its VBE drop and internal resistance. At 10 A that drop can reach 0.6–0.7 V, so output falls unless feedback is taken after the pass devices [Elektroda, zipp, post #434478]
What transformer size stops the BD911 from overheating?
Keep Vin–Vout ≈ 3–6 V under load. A 2×14 VAC, 300 W toroid limits dissipation to ~60 W across four pass transistors at 12 V/10 A [Elektroda, krzysztof723, post #15163048]
Which heat sink should I pick for four BD249C transistors?
Use at least 0.4 K/W thermal resistance with a 80 mm fan. Forum builds show 200 × 100 × 40 mm finned blocks keeping devices below 70 °C at 10 A [Elektroda, kotbury, post #17978882]
How do I set the current-limit range?
- Start with 0.47 Ω, 5 W resistors in each emitter.
- Load the PSU at 5 A and turn the 470 Ω pot until limiting begins.
- For lower minimum current, raise resistors to 0.56 Ω; for higher maximum, drop to 0.33 Ω. [Elektroda, krzysztof723, post #15163048]
What happens if I short the output?
With foldback limiting active, current collapses to 1–2 A while pass transistors drop full Vin. Without a 10 A fuse, a single BD911 can blow instantly because "at 1.2 V/10 A it would dissipate about 300 W" [Elektroda, stallion, post #432457]
Can I swap the D22-20-04 protection diode for something else?
Yes. Any 20 A, ≥200 V silicon rectifier (e.g., 6A10, P600M) works. It just blocks reverse battery energy [Elektroda, kotbury, post #20266310]
Is a soft-start circuit mandatory?
Recommended but not mandatory. Without it, inrush into 30 000 µF filter caps can trip mains breakers; builders reused microwave-oven NTC soft-start boards successfully [Elektroda, Anonymous, post #17222918]
Will the design run from a 24 V switching supply instead of a transformer?
Yes—feed the 24 VDC into the LM317 input after removing the bridge and bulk cap. Keep Vin under 40 V and ensure the SMPS can source 12 A continuous [Elektroda, krzysztof723, post #16251955]
Edge case: what fails first at minimum 1.2 V, 10 A output?
Thermal stress. With 28 V across pass devices, total dissipation exceeds 280 W. Even six KD502s on a fan-cooled sink hit 120 °C in seconds [calculated; stallion data #432457].
Can I substitute MOSFETs for the Darlington stage?
Not drop-in. MOSFETs need gate drive > Vout and add reverse-polarity issues. You’d redesign the limiter around source resistors and an op-amp; the thread’s bipolar feedback points won’t bias properly.
Why does the limit LED light only above 4 V output?
The BD140 indicator transistor needs about 0.6 V across the 0.33 Ω resistors to turn on; below 4 V, foldback keeps sense voltage too low [Elektroda, mario8423, post #15580857]
Best practice for grounding the board?
Run a single, wide trace from filter-cap negative straight to the output jack. Tie signal grounds (LM317 adjust node, sense resistors) near that star point to avoid 50 mV regulation wobble [Elektroda, krzysztof723, post #14554493]
Comments
And what happens with this system in the event of a short circuit? Is it not necessary to call the PP Stasis? Although I like it myself. MarCiu [Read more]
As you can see, I used a few additional LEDs on the LM317, that's probably enough. There is no output fuse in the diagram. [Read more]
How big a heat sink for it do you envision. :!: :!: [Read more]
You can use LM317 and add a high power transistor to it and you will have 10A. [Read more]
What is this output coil or something else that is shown in this diagram as a spiral? For this system, you need a veeeeery large heat sink, if someone would like to take 10A current from it. Especially... [Read more]
Maybe a little criticism 26AC to 36.77VDC "NO LOAD" 35V 2A - higher current will cause "feeds" and losses at 1.2 10 And you can forget because (greatly simplifying: With a consumption of 10 A, the... [Read more]
Hello Ideal power supply for charging the Shuttle's batteries :D And seriously - in fact, at the smallest range and at high current, the power losses will be huge. As for the correct operation... [Read more]
My power supply cools the radiator with the fan + driver for it. The output coil is wound on the ring core with 1mm wire. The maximum voltage in this version of the LM317 is probably 37V if I want to ... [Read more]
I think 2N5302 or 2N5686 would be a little better. Although 3 x 2N3055 would be enough. A piece of heat sink is necessary. [Read more]
The diagram shows that the stabilization of the output voltage when changing the load current from e.g. 1A to 5A is poor. (The voltage is stabilized on the base of the BD911 and not at the output). [Read more]
The transistor in this power supply acts as an element increasing the current efficiency. The output voltage is the same as that supplied to the base by the LM317. [Read more]
Andrzej is right. The voltage stabilization in this system will be lower than that of the LM317 alone. Unfortunately, transistors are not "perfect" elements. As the base current increases, the base-emitter... [Read more]
Uwy will be lower but this is compensated by the considerable simplicity of the layout. I think that "at the beginning" (talking about the less experienced) is enough. [Read more]
Honestly speaking, I do not recommend this system for little experienced (high currents, high temperatures). As the zipp says - in the application note there are other ways to implement the power supply... [Read more]
This is what my first regulated power supply looks like, working until today but as a battery charger; P Unfortunately, this solution has a significant temperature drift, no protection, and low efficiency. ... [Read more]
The first devices will not always be refined, we forgot something, we didn't know something. [Read more]
I deleted the answer because I realized that a friend of zipp had "protected" me. [Read more]
I used to make a similar power supply, but instead of one I used two transformers (to power 12V / 60W halogen bulbs). To reduce the power loss on the transistor, I used a switch that connected the transformer... [Read more]
WELCOME :) do you know any link with nice power supplies ... I need it to have stabilized 30V for power amplification I have a transformer that gives me 37.7V I also wanted to connect the led to 3.5... [Read more]