Non-symetrical SMPS output

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello All,

I just prototyped (on a breadboard) Rod Elliot's Switchmode Power Supply For Cars (Project 69) http://sound.westhost.com/project69.htm and it looked like everything was good, with very little ripple and such. But my outputs are really non-symetrical. I've got 20.6 VDC on the positive side and 13.6 VDC on the negative. I know its not the windings because its a single wound transformer, turn ratio is a little off but I'm tweaking still. Anyway, the only thing I can think of is that the waveform isn't symetrical. I used all 10k resistors instead of 12k, so perhaps thats the problem. Also for C3 I used a 500pF cap instead of a 1,000pF. Perhaps that could be the problem.

Other than that everything is ok, the transformer emits a high frequency squeel but I think thats normal. I encourage other people to build SMPS as its not nearly as hard as I thought (haha, easy to say, mine doesn't work yet).

Thanks,
Pete
 

Attachments

  • p69-f1.gif
    p69-f1.gif
    8.6 KB · Views: 266
Quick update...

The waveform (50/50 duty cycle) looks fine without the transformer connected (I'm talking as measured across pins 11 and 14). When the transformer is connected, all hell breaks loose. It has a much higher peak on the one side, and a much longer duty cycle on the other. Any ideas?

Pete
 
Just an update for anyone whos following this thread. I believe I 'might' have found the problem. The SG3525 datasheet says minimum C3 (Ct) (timing capacitor) of 0.001uF whereas I've got a 500nF in there. That's half the minimum value. Rod Elliot uses a 1nF (0.001uF) cap. I'll try something higher tonight because I think with the high tolerance values of caps (aren't they +/-50%) you should use at least 2nF (0.002uF).

Anyone familiar with the SG3525 that can confirm this?

Pete
 
non-symmetrical SMPS output

Pete,

These circuits from RE's web site seem to be popping up everywhere on this forum. They all have the same problem. They don't work! I've already posted replies in this forum on other sections. The biggest problem is - where is the inductor??? I visited RE's site and he flatly states that he doesn't feel that inductors are necessary. I can assure you that they are. The output rectifiers and large filter caps do not effectively filter the pulses. Instead they form a peak detector and hold the peak voltage, drooping as they discharge into the load. An inductor is absolutely needed to output a smooth low-ripple voltage. Of all of the transformer SMPS topologies, only the flyback converter can work without an inductor, since the flyback transformer core is gapped, and it can store the per cycle energy. There are plenty of sites, Natl Semi, Tex Instr, ST Micro, etc. where you can get application noted on SMPS. Again, without an inductor, you're spinning your wheels. The reason I haven't contacted RE, is that he specifically states that he does not want e-mails sent to him. I can see why.

Also, for two outputs, two secondaries on the transformer are needed, as well as two output filter inductors, preferrably magnetically coupled. Again, use an inductor after the transformer. Best wishes.
 
Re: non-symmetrical SMPS output

Claude Abraham said:
Pete,

These circuits from RE's web site seem to be popping up everywhere on this forum. They all have the same problem. They don't work! <SNIP>

Also, for two outputs, two secondaries on the transformer are needed, as well as two output filter inductors, preferrably magnetically coupled. Again, use an inductor after the transformer. Best wishes.

Yupp, and further:

You can use one center-tapped transformer secondary, but you do need the inductors. A good example is to be found on Texas Instruments (Not SGS Thomson or Nat Semi) discussion of the SG3525.

The other mistake on the RE site -- there is an error amplifier circuit on the SG3525 LM2525 and LM3524 which is used to sample the output through a voltage divider and compare it to the internal reference. An error will cause the pulse-width to enlarge or contract depending on the direction of the error. The way that the error amp is depicted in the RE website is wrong.

I don't know if you looked at the pulses with a scope -- are you sure that you don't have two different duty cycles.

You can get rid of the whining by using a higher switching frequency. (Why do people still persist in using this first generation chip?)

Lastly, the windings do have to be carefully matched. Hopefully you have chosen a core which isn't saturating part way through the duty cycle. There is a MSExcel calculator cobbled together from Linear Tech's information on my website: http://www.tech-diy.com/smps_xfmr.xls
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.