Why extra Diode?

ROFL!!!!

That reminds me of an incident that happened in my first job. I was a trainee building prototypes of power supplies, and had just finished building the first of this new huge switch mode thing with 48v at loads of amps output.

Anyhow, so I proudly plugged it in and turned it on - there was small pause as the lights flickered in the lab, then BANG - the mains rectifier electrolytics deposited their entire contents over me, the ceiling and anything within range - turned out I put them in backwards :-)

I never did that again.

Yours, Mark.

Wats> I was examining a defunct ATX PS that had a Really Bad Day; two of the

start up voltage. some switchers require a min voltage to allow for its feed back voltage in the coil to get started. the inrush current on the caps will not allow for enough voltage to be present so a Diode is a good idea to use since it will give you a approximately .6 volts of a drop. putting both together you would end up in the range of 1.2 volts on the average of at least 1.2 volts to be present at the coils on start up when the caps are in initial inrush. anything below that would constitute a short in the diodes which maybe the reason why your caps exploded it the start up didn't have a time delay shut down from the voltage not reaching at least its

Wats> I was examining a defunct ATX PS that had a Really Bad Day; two of the

Mark (UK) top-posted:

I was trying to learn how to build a switcher from some chip's data sheet. I was working with ferrite vendors as well as the chip vendor, and decided on a core, bobbin, turns, and all that. So, I've got this cup core or pot core, I'm not sure which it's called - it's the one with two cylindrical halves and a hole through the middle.

So I've got this carefully committee-designed transformer, about to fire up this unit at 1MHz, everybody's nervous - I'm cranking up the variac (this is when I learned that it's hard to ramp up the input of a switcher and have it behave predictably), and realized, too late, that I had forgotten you're not supposed to fasten a cup core transformer down with a steel screw. FzzzzzPow!

I never did get a switcher designed, or even built. After about three months, and I don't know how many thousands of the client's dollars, we gave up. We concluded, in 1990, that it's impossible to fit a 300W power supply into 75 cubic inches.

Some things have changed, but they probably still say you're not supposed to use a steel screw to fasten down a pot core. ;-)

Ironically, the guy had a brass screw exactly the right size on hand.

Cheers! Rich

Quite possible. You can only really regulate the DC voltages from one output meaningfully - or maybe cross-couple 2 perhaps.

The output voltages on other windings will then be set by turns ratios ( and diode drop voltages).

As it's just a 1 amp output it's probably simplest to drop a bit of voltage this way.

I've seen designs that selectively use Schottkies with their lower voltage drop to get accurate voltage outputs too.

Graham

SMPS supplies can have quite a high voltage/turn and fractional turns are a hassle.

Best regards, Spehro Pefhany