2
\$\begingroup\$

I'm trying to fix an AC-DC line regulator that I have build and doesn't work as expected.
I'm trying to convert AC 220V mains to DC 5V 10mA without a transformer. To do this I'm using an 1uF 250VAC Polypropylene Film cap like in the schematics bellow, followed by a voltage regulator with TL431.

When I simulate the circuit in Proteus, I get 5V at output with/without C3.
In real circuit with C3 I get 5V output (not sure how noisy though; no oscilloscope), but if I remove C3 I get 2.5V, which clearly indicates that the shunt regulator TL431 oscillates.

I tried putting a 22p cap between the cathode pin and ref of TL431 (pin 1-2), but no dice.I have also tried removing D4.

Any thoughts?

WARNING: - The output of this circuit is not "isolated"from AC mains and should be treated as being at mains potential at all times. [RM]

enter image description here

Notes:

Failure of C1 or reversal of Vac connections or other conditions will result in mains live (Phase) voltage appearing at output.

C1 MUST be an "X rated" capacitor designed and certified for use on 230 VAC AC mains by its manufacturer.

[RM]

\$\endgroup\$
  • 1
    \$\begingroup\$ Have you tried putting the cap on the reference pin (2) of the shunt regulator? You want the reference voltage to stop oscillating which is why I would think it would do more there. If you put it at the cathode, it's like filling up a cup with a fire-hose and expecting that you'll be able to accurately get a certain amount of water in that cup. \$\endgroup\$ – horta May 13 '14 at 18:06
  • \$\begingroup\$ @horta I have tried a 22pF between pin (1) cathode and pin (2) ref. Now I'm trying one between pin (2) and ground. \$\endgroup\$ – Chris May 13 '14 at 18:12
  • 1
    \$\begingroup\$ Why don't you keep C3 in place. Seems simple to me or am I missing something? \$\endgroup\$ – Andy aka May 13 '14 at 18:20
  • \$\begingroup\$ @Andyaka Because this circuit powers up an instrumentation amplifier and I believe there is noise on the output, because my IA peaks up a 50Hz noise at high gain(1000x). \$\endgroup\$ – Chris May 13 '14 at 18:25
  • \$\begingroup\$ @horta Just added a 22pF between ref and ground and I got clean 5V on output. Problem solved. Don't know why I didn't thought of it earlier. \$\endgroup\$ – Chris May 13 '14 at 18:27
2
\$\begingroup\$

Moving my comments to the answer space so that others can get use out of it:

Have you tried putting the cap on the reference pin (2) of the shunt regulator? You want the reference voltage to stop oscillating which is why I would think it would do more there. If you put it at the cathode, it's like filling up a cup with a fire-hose and expecting that you'll be able to accurately get a certain amount of water in that cup.

\$\endgroup\$
2
\$\begingroup\$

BE AWARE THAT Vout SHOULD BE CONSIDERED TO BE AT MAINS VOLTAGE AT ALL TIMES.

THIS CIRCUIT WILL KILL YOU (OR OTHERS) IF YOU ALLOW IT TO AND MAY KILL YOU IF YOU DON'T.

If what you say is true ie

I'm using an 1uF 250V Polypropylene Film cap

then the capacitor C1 is vastly under-rated for voltage and is quite likely to die a few 10's of milliseconds before your output circuit and possibly the user start to.
Capacitor C1 MUST be an "X rated" 230 VAC capacitor - at least. ie a capacitor which is designed to have 230 vAC mains applied across it.
If you are doing that already you should say so and NOT just call it a 250V capacitor, as you do. This, if for no other reason than that other users might copy your lethally dangerous "working" circuit and die. Ideally and morally (and probably on this site) all circuits that work as this one does should carry a non isolated mains voltage at output warning.

C1 Vworking will be mains peak =
= 220 V AC x 1.414 =~ 310 V
and you can increase that by say 10% for surges to ~= 350V and then add spikes and general happenstance. So an X rated 230 VAC mains cap is at least 400 VDC rated and may be more and is designed to handle real world mains happenings.

Stability:

You are looking for ad hoc solutions to loop instability in a regulator without a good understanding of the issues. This is a blackish art and you can either take the effort to understand what is happening or try various things that make some sense and MAY work.
You've found one, which is good. Similarly, a small capacitor across R3 also may help.
Make sure that the solution that you have found (cap from Vref to ground) works in all situations that you will experience in practice.

\$\endgroup\$
  • \$\begingroup\$ This will be used only by me and I would never touch the device while is operating. The device is operating only under observation. C1 was taken from a PSU in which it was connected between live and neutral, so it must be an X rated capacitor. I apologize for the misguidance. The capacitor "looks" very similar to this one: alibaba.com/product-detail/… \$\endgroup\$ – Chris May 13 '14 at 22:11

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.