I built a 3.3/5/12/25V power supply from a Macintosh G4 API1PC36 power supply using Daniel Haertle's project description. I placed a switch between the PwrOn line and GND and a LED between 5V and GND. I used a \$27\Omega\$, 10W wire wound resistor from the 3.3V/sense line to GND so the power supply does not go into standby mode and I wired 3.3V, 5V, 12V and 25V lines to banana plug ports mounted on the body of the power supply. To try to ensure the wires can handle the power used, the 3.3V, 5V and 12V lines are triple wires. See Figure 1. Figure 1

I want to use the power supply to power a 12V 100W halogen microscope light with a PWM dimmer between the two to control the luminosity. The power supply works fine with the halogen lamp directly connected to the 12V line, see Figure 2. Figure 2

With the PWM between the power supply and the lamp, the power supply switches itself off if started with the PWM set to anything less than about 90%. It does not switch off if started with the PWM set high, but does switch off when I start to turn the PWM down, see Figure 3.Figure 3

If I put a \$1.2\Omega\$ incandescent lamp on the 3.3V line then the 12V line is fine with the PWM dimmer through the PWM's full range. See Figure 4.Figure 4

If I put a \$30\Omega\$ resistor on the 3.3V line then the PWM dimmer only works over the top 3/4 of its range, but power supply shuts itself down when the PWM is in the bottom quarter of its range. See Figure 5.Figure 5

I have previously built a similar power supply for a 12V 100W halogen microscope lamp from a Seasonic SS-300TFX PC power supply which works fine with the PWM and a 12V 100W lamp. In that case, following DroneBot Workshop's instructions I put a \$18\Omega\$ wire wound resistor on the 5V line as the constant load to prevent the ATX going into standby mode. On the API1PC36, I did try putting the \$1.2\Omega\$ lamp on the 5V line as a load but this behaved basically the same as Figure 3.

  • What is causing this behaviour? Is it something related to the frequency of the PWM experienced by this Arduino user??
  • What further data can I gather to diagnose the problem?
  • Do I need to change the resistance I put on the 3.3V/sense line to a lower value? I would like to keep the power consumed by the 3.3V/sense line as low as possible.
  • Is there better solution than changing the resistance of the 3.3V/sense line? Is there a circuit I could put between the power and supply and the PWM to make them work more smoothly together? What would the appropriate circuit design be?
  • \$\begingroup\$ I am accepting @user253751 answer because it allows a lower load to be put on the 3.3 sense line and capacitors are pretty easy to get and at 12V they are pretty safe. But I want to note that the higher load on the sense line also does work so I am glad that both answers are here. \$\endgroup\$ Nov 26, 2022 at 17:04
  • \$\begingroup\$ Additionally I should say that bringing the dimmer up from 0 in either solution requires turning it up very slowly until the bulb begins to glow and letting the filament come up to temperature before trying to go higher on the dimmer \$\endgroup\$ Nov 26, 2022 at 17:05

2 Answers 2


They call this an "XY problem". Your problem is "X" (the power supply turns off) but you have a certain solution "Y" in mind (add a resistor to the 3.3V wire) so you ask how to do "Y" even though it's clearly not a very good solution.

Instead of fiddling with resistors, ask yourself: Why does the power supply shut down?

Because it only happens when the PWM is on, I assume that it responds badly to the rapid on and off switching.

If that is the problem, then instead of messing around with resistors on a different voltage rail, try adding some (big) capacitors in parallel with the PWM input so the power supply does not see the switching!

  • \$\begingroup\$ When I had originally posted this question months ago I thought adding capacitors to shield the power-supply was the right answer but knew nothing about how to do it. At the time I was not so great at asking well formed questions so I did not get any instructions on the capacitor solution but got a lot of questions about the load on the sense line (because it is important when using a computer power-supply). I guess that pushed me in this direction. I would love some pointers on the capacitors solution. What capacitance would you suggest and how many capacitors? Thank you. \$\endgroup\$ Oct 10, 2022 at 16:12
  • \$\begingroup\$ @brettstevens I have no experience with using computer power supplies this way. I would use one or two 470uF 25V electrolytic capacitors and connect them in parallel with the dimmer input. I would put something in between them and my face, just in case they go pop. I would then check whether the power supply still works, and the dimmer can go to a lower setting than it used to. If the dimmer does go lower, but not all the way to zero, I'd try adding more capacitors. \$\endgroup\$
    – user253751
    Oct 10, 2022 at 16:19
  • \$\begingroup\$ Thanks for the suggestions. \$\endgroup\$ Oct 10, 2022 at 16:20
  • 1
    \$\begingroup\$ @brettstevens Risks include: the power supply might not turn on due to the inrush (charging) current (risk goes up with more capacitance), capacitors could overheat from ripple current and then going pop (risk goes down with more capacitors in parallel as it spreads out the ripple current). These are things that can be calculated, of course, but not without knowing a bunch of fiddly details and numbers and stuff, and you don't have those so it's trial and error, but you knew that. \$\endgroup\$
    – user253751
    Oct 10, 2022 at 16:21
  • \$\begingroup\$ If the lamp circuit is 12V should I be using 12V capacitors? Or is the 25V just to make sure they are more robust than needed? \$\endgroup\$ Oct 29, 2022 at 13:14

I found with a lot of experimentation that many uses of this power supply only need a small load on the 3.3V/sense line: I can use a resistor as high as 25-30Ω. The PWM is the worst device I have experimented with. I have found that it is possible to use a resistor as high as 12Ω if I let the incandescent bulb after the PWM warm up first. With the bulb cold I would need a very small resistance as described Figure 4 in the question. So with this range in mind I have built a 30Ω 25W potentiometer into a box that can be used with the power supply and be set appropriate to the "noisiness" of the device I am powering. enter image description here I want to thank commenters from my first posting on this issue for their advice on both the electronics and also on providing better information in my posts here.

  • \$\begingroup\$ That sounds like the PWM in interacting with the power supply's internal control. Try another power supply and see if that helps. Also you will fry a lot of LEDs, you need to put some resistance between it and and the power supply, either leg will be fine. \$\endgroup\$
    – Gil
    Oct 10, 2022 at 15:33
  • \$\begingroup\$ I have had great success with a using a Seasonic SS-300TFX power supply for a dedicated 12V 8A dimmable halogen power supply for a microscope. Just a fixed 18Ω resistor on the sense line works fine with the PWM dimmer. \$\endgroup\$ Oct 10, 2022 at 16:18

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.