Power supply help

[Oneillcz]

Hello,

I have a question about a power supply that I tried to build yesterday that went wrong.

I'm trying to build myself a +-15V power supply, but I was not really comfortable to work with mains voltage, so I used a full-wave voltage doubler design which was powered by a 20VAC wall-wart. I basically used the same circuit as shown here http://sound.westhost.com/project05.htm, but in the prototype I omitted the diodes except D1 and D2, and used 1N4007. Also I want a +18V output for my LM380 (I could probably use +15V, but I figured that there would be just a tiny bit more headroom with 18V), so I put another voltage regulator for the +18V in parallel with the +15V regulator. And because this was the prototype on a breadboard, I left the outputs open so they were not powering anything.

Now this is what happened. I connected the wall-wart to the breadboard and the output from it dropped to just 15VAC, so the +18V regulator could not work, and even the +-15V regulators after a few seconds showed no voltage output. This measurement lasted just a few seconds then I disconected it. I checked the bredboard maybe five times, but it was fine. So I pluged the wall-wart again (which was a mistake, I know now) and after a few seconds, it went dead. I unpluged it and measured infinite resistance on the primary, so the transformer got destroyed. No component on the breadboard heated up. When I opened the wall-wart the transformer was really really hot.

So my question is this. What I did wrong? Is the +18V regulator in wrong place? Are the smoothing capacitors too big? Was it a problem with the inrush current? Or the fact that the regulator outputs were not powering anything? Could it be just a bad transformer and if so, in what way?

I decided to rebuild the wall-wart with new transformer and try again, but since I don't know what blew the first one, I don't want to just experiment with a new one. And yes, I will use a fuse next time.

Thank you for any help.

[R.G.]

First, go read "Power Supplies Basics" at geofex.com. Think about it a bit, then re-ask your questions.

[Oneillcz]

Thank you for your reply.

I have read your article. So basically I did what is in the last picture in that article, then I attached a +15, -15 and +18V regulators (so that the + regulators shared the +V raw supply). And then I ended up with a blown transformer. As I said, I checked the prototype circuit thoroughly (checking it again as I'm typing this) for mistakes, there were none, I mean, there must be some unintentional one otherwise it would work and I would not be embarrassing myself here.

So I built a half wave voltage doubler, not full wave like I said in my previous post.

I have also thought about it a bit. You mention in the beginning, regarding the very first picture, that "we'll pretend there is some load parallel to the capacitor ...". That's what I did, isn't it? Pretended there is some load, but there wasn't, the regulators don't count as load right? I should have put some resistors between +15V and GND and all the other outputs as well.

Regarding the capacitor size, the ones I used were bigger than what was needed for this Vsecondary used, but surely that alone could not have blown the transformer.

[R.G.]

I have read your article. So basically I did what is in the last picture in that article, then I attached a +15, -15 and +18V regulators (so that the + regulators shared the +V raw supply). And then I ended up with a blown transformer. As I said, I checked the prototype circuit thoroughly (checking it again as I'm typing this) for mistakes, there were none, I mean, there must be some unintentional one otherwise it would work and I would not be embarrassing myself here.

Well, don't sweat the embarrassing yourself - I have made mistakes on power supplies that you would love to watch the videos of - if there were videos. 

I have always found that simplifying things helps. You have the whole mess built, right? If you remove one diode, then you get a half wave rectifier supply on either the positive or negative side, the one you didn't remove the diode from. And if you remove the link from filter cap to regulators, you have just the rectifier/capacitor. So there are two levels of simplification you can do pretty easily. More on this in a minute.

But up front is one of my favorite mistakes. 79xx regulators do NOT have the same pinout as 78xx regulators. The center pin is V- (input), not ground. One of my favorite kill-things-now tricks is to forget this and wire the + and - regulators alike and have Mother Nature remind me of the error by smoke signals. 

I have also thought about it a bit. You mention in the beginning, regarding the very first picture, that "we'll pretend there is some load parallel to the capacitor ...". That's what I did, isn't it? Pretended there is some load, but there wasn't, the regulators don't count as load right? I should have put some resistors between +15V and GND and all the other outputs as well.

That's not your problem. A rectifier/capacitor works fine with no load, it just takes nearly forever to run the cap down when you remove the power. Regulator chips do have some internal current leakage, but it's deliberately made pretty small. You could put, say, 100K resistors across the filter caps during the testing phase just to make sure the caps don't stay charged too long, and remove them later once things are fixed.

Regarding the capacitor size, the ones I used were bigger than what was needed for this Vsecondary used, but surely that alone could not have blown the transformer.

No, that's not the problem.

It is possible that (1) the transformer was already damaged, you just happened to be there for the grand finale. Not likely, but possible; (2) the negative regulator was mis-wired; (3) one of the filter caps was reversed (I've done this one, too); (4) a host of other minor issues.

A small transformer is almost always primary-resistance limited. It has such a high wire resistance that not much current can get through. They also generally have thermal fuses inside the primary, so if it gets too hot, the fuse opens before the smoke signals can be sent out.  They also are wound with higher open circuit voltages to make up for the sag the high wire resistance causes. So if your transformer says something like "20Vac  300ma", it puts out 20Vac ONLY when it's loaded with 300ma. At 0ma, it's probably 15-20% higher. 20Vac*1.20 = 24Vac, and that rectifies to 34Vdc under no load. That's suspiciously close to the max voltage on some regulators, although as I remember the 7818 is 35-40V.

My advice:
1. disconnect the regulators from the filter caps
2. disconnect the (-) diode
3. Hook 100K resistors temporarily across the filter caps
4. test your new transformer with your DMM by placing the probes across the output, set to >20Vac scale; if that works
5. unplug the transformer from the wall, connect it to the now-positive-only power supply rectifier/cap, and clip your DMM probes across the filter cap, set to a 50Vdc scale; plug the transformer into the wall and immediately remove it. If the trannie works, the cap will show a blip of voltage in the correct polarity.
6. If that worked, plug the trannie in and leave it, watching the voltmeter; you should read 24-34Vdc.
7. Hold your hand on the wall transformer output, sensing temperature; any temp rise noticeable? If so, unplug.
8. If all is well, remove the transformer from the wall and connect up ONE of the positive regulators. Re-plug the transformer. Does the filter cap stay at the same voltage from before the regulator was added?
9. If yes, move the DMM probes to the regulator output; is it correct?
10. If so, put the other regulator on and test as in 6-9.
11. Remove the + side diode, put the - side diode in, and repeat 5-9 for the negative side.
12. If all is well so far, hook them all up and test.

[Oneillcz]

Hello,

so I bought a new transformer and tried it just as you wrote and everything went fine, so I put it on a pcb and was done with it. But I can still only speculate as to why the original transformer went dead.

Thanks for your help, I really appreciate it.