3
\$\begingroup\$

I am building a +/- 45V power supply to drive a master-slave amplifier for a sonoluminescence experiment. I am following a stock design, and am having massive sag problems. The circuit powers up fine, but if I engage the LED indicators (each draws about 0.015 A), the supply starts to sag, and the voltage drops to around 10 V. Hooking up a power resistor is much worse - 75R 100W should draw 0.6 A, but the voltage plummets to less than 1 V almost immediately. I tried disconnecting the Varistor network, but that didn't fix the problem. The bridge is rated at 6A 600V, so that shouldn't be the problem. Diodes D1 and D2 are 5 1N4001 diodes in parallel, to allow more current (and because I had them laying around).

I am using the schematic shown below, based off of standard circuits in the data sheet for the LM317HV and the LM337HV (HV = high voltage versions, these can handle +/- 45 V), and Rod Elliott's Bench supply, found at: https://sound-au.com/project44.htm.

So here's what I'm using:

Sono Amp Power supply, 1v0

I plan on adding protection diodes to the output, to prevent external voltages from feeding back into circuit, but I haven't done so yet.

I would be happy to answer any questions or perform voltage/current tests to try and track down the problem.

\$\endgroup\$
  • \$\begingroup\$ What transformer are you using? \$\endgroup\$ – pgvoorhees Jun 24 '16 at 19:38
  • \$\begingroup\$ At what point do these LED indicators connect? \$\endgroup\$ – Sparky256 Jun 24 '16 at 20:08
  • \$\begingroup\$ LED indicators run from ground to appropriate power supply, with a 3k current limiting resistor. Should draw 15 mA, and they turn on normally when I power up, Red+, Blue-. They are both switched, so I can turn them off (need complete darkness for experiment.) When switched off, I get +/- 45 V. Switching them on starts sag after a couple of minutes. \$\endgroup\$ – aeronaut Jun 24 '16 at 20:23
  • \$\begingroup\$ Transformer is a toroidal, with dual inputs and outputs. 50 V (2x 25 V), 2.5 A. Here's a link to the actual item. Probably overkill. mcmelectronics.com/product/287-1090 \$\endgroup\$ – aeronaut Jun 24 '16 at 20:26
  • \$\begingroup\$ Also, it's probably not working because of something stupid that I did wrong. Don't assume I have tons of experience with this. A Ph.D in physics doesn't make one a circuit builder, just a circuit understander. \$\endgroup\$ – aeronaut Jun 24 '16 at 20:29
4
\$\begingroup\$

From the comments, the V-drop is ~35V. This is a huge V drop for those regulators. You're probably over heating them and triggering their thermal shutdown mechanism.

  • Reduce the input voltage as close to +-45V (Plus drop out voltages) as possible.
  • Add heat sinking to the cases suitable to the anticipated power dissipation. (remember to add the sense resistors in, they consume almost 1/3W by themselves)

Edit: Took out my guesses. Added clear suggestions

\$\endgroup\$
  • \$\begingroup\$ I disagree. MCTA250/50 clearly says it has 50V output windings. \$\endgroup\$ – Adam Lawrence Jun 24 '16 at 20:58
  • \$\begingroup\$ Clean ground to positive output from bridge is 71 V DC. Clean ground to negative output from bridge is -71 V DC. This seems reasonable to support a 45V output. That said, the thermal shutdown is a reasonable explanation for the problem, but why do tiny currents cause it. \$\endgroup\$ – aeronaut Jun 24 '16 at 21:06
  • \$\begingroup\$ 148 V from positive to negative output on bridge. But neither regulator sees this difference, just the difference between pos or neg and ground. \$\endgroup\$ – aeronaut Jun 24 '16 at 21:11
  • \$\begingroup\$ @aeronaut Well, then you're dropping 36V (71-45) at the regulators. If you're drawing even as little as 15mA, it makes more than 0.5W dissipated. Depending on the regulator package / environment / ... Even this little current may be too much. You should use a transformer with lower voltage. 71V is definitely not reasonable for a 45V output. \$\endgroup\$ – dim Jun 24 '16 at 21:45
  • \$\begingroup\$ So something like this: mcmelectronics.com/product/287-1085, where the output voltage is 35 V AC which rectifies to 49.5 V DC less a diode drop. Same size body, massive over current. Too bad they don't have a smaller 35 V output one. Is 49.5 V enough to get 45 V out of the regulators without dropout, and with capacitor losses? \$\endgroup\$ – aeronaut Jun 24 '16 at 23:09
0
\$\begingroup\$

How are you connecting your LED's ? Point the position of your LED's in your circuit. Nevertheless, I would suggest using a 3.3K resistor or higher along with the LED to limit current . usually 15 mA of current is typical. Hopefully it will work for you

\$\endgroup\$
  • \$\begingroup\$ LED goes from power output to ground, with a 3K current limiting resistor, and is switched, (DPDT, one side for +, one for -), so I can engage or disengage the LEDs. Both light up when supply is on and they are engaged. \$\endgroup\$ – aeronaut Jun 24 '16 at 20:32
0
\$\begingroup\$

You know that the capacitor values 10m and 6m8 mean 10,000 uF and 6,800 uF (microFarads), right? Using 10 uF and 6.8 uF caps would cause the issue you describe.

Can you measure the input and output voltage of the regulators with an oscilloscope?

\$\endgroup\$
0
\$\begingroup\$

Reducing the transformer voltage helped. It can now source/sink at least 250 mA of current. Higher loads still produce sag. But this is enough for the intended application, I think.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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