Re: Help with off-line smps
Marko, Thu Aug 03 2006, 07:34PM

You have a lot more to tell.

Are you using high or low side driver, wich IC, regulation, and etc.

A schematic would also be nice...

Re: Help with off-line smps
uzzors2k, Thu Aug 03 2006, 08:56PM

Currently I'm just trying to get an output from a smps transformer. I have two taken from a computer PSU and one from a monitor. I plan on rewinding one of them once I get the main switching circuitry in place. Its driven by a TL494 -> mini-h-bridge -> GDT -> mosfets -> xfrmr. Pretty crude, but I'm just experimenting.
Here is the GDT output as well. When switching a mosfet it slopes a little more of course, but by just 0.1µs or so.

I just want to know if its safe to plug into the wall yet.

Re: Help with off-line smps
Steve Ward, Thu Aug 03 2006, 09:19PM

Those diodes are only rated 200V at 1.5A. The diodes will see 350V plus any voltage transients (so maybe 400V). Also, the diode should usually be rated about the same as the FET, so a 15A diode would be better than 1.5A. The MUR1560 comes to mind as a suitable diode.

So, dont plug it in until you get new diodes!

Re: Help with off-line smps
Marko, Thu Aug 03 2006, 09:52PM

For such frequencies and power level internal diodes inside the mosfets are good for beggining.
You can bypass them altough if you want 'for insurance'.

Maybe you could lower the deadtime in order not to push everything to maximum instantly.

Take care around transformer, im. use one that was previously used with halfbridge (ATX transformer for example).

Re: Help with off-line smps
Steve Conner, Thu Aug 03 2006, 11:01PM

Just think how much you'll learn for every two MOSFETs you trash. :P

The extra diodes shouldn't be needed. I've repaired and modded several commercial SMPSs that use MOSFET half-bridges, and they all seem to get by with the MOSFETs' own body-drain diodes.

The diodes only carry the transformer's magnetizing current, which is small with a well designed transformer. Even if you have a transformer with lots of magnetizing current, it's always an inductive current. What really stresses the diodes is capacitive current.

The SMPS circuit you chose is a fairly old and well-tried one. It should be pretty hard to blow up unless you make a gross mistake wiring it up, or get the transformer design so badly wrong that it saturates.

Re: Help with off-line smps
..., Thu Aug 03 2006, 11:13PM

It would have been nice if you got the formatting right for the schematic so we could actually read it (quote me to see how it was done) when we click it




But other than that you shoud be fine without diodes I should also point out that you might need to put a schotty diode in series with the fet if you use an external diode as if the diode you are using has a higher Vdrop than the fet's it won't do anything (the schotty adds an extra .5v or so to the Vdrop of the internal diode, so the external one takes all of the current)

Re: Help with off-line smps
Steve Ward, Thu Aug 03 2006, 11:59PM

Thanks for fixing the schematic ... .

the schotty adds an extra .5v or so to the Vdrop of the internal diode, so the external one takes all of the current

I suggest you look again at the direction of the schottkey diode. It doesnt add a .5V drop, rather it adds whatever the reverse stand off voltage of the schottkey diode is (so 40V or whatever). The schottkey is forward biased when the MOSFET conducts, but is *reverse* biased when the current reverses and tries to go through the body diode.

I agree with the rest of the people here, your application is low enough in frequency that you can get by just fine on the internal body diode (though it is particularly slow at over 1uS recovery time!).

Re: Help with off-line smps
uzzors2k, Fri Aug 04 2006, 11:56AM

Thanks for the advice. I took out the diodes and plugged it in, only to have the breaker trip. I was about to despair, but I found the bridge recticfier had shorted. I forgot to replace it after the first failure. (my avatar) Now would it be ok to just feed the smps with half-wave recticfied mains?

Re: Help with off-line smps
Steve Ward, Fri Aug 04 2006, 06:02PM

Now would it be ok to just feed the smps with half-wave recticfied mains?

I would not personally do that, but if you really cant wait...

Re: Help with off-line smps
uzzors2k, Mon Aug 14 2006, 11:33AM

I replaced the bridge and I am glad to say the PSU works excellently. ( with only two mosfet deaths! ) I'm using a PC smps xfrmr, and I got 25volts on th 12v line, which is exactly what I got from the xfrmr while it was in the original PSU (with volt mods). I haven't stress tested it, but I think it can source enough power for now. The only limit is the transformer which came from a 250 watt supply.

Just one more newb question, should I put heatsinks on the IRFP450s? I don't dare touch them while the PSU is running. They can handle 14 amps, and the bridge rectifier is rated for 2 amps, so I think they will be fine. I'm not going to run the supply for more than 10 minutes at a time anyway.

Thanks for all the help.

Re: Help with off-line smps
ragnar, Mon Aug 14 2006, 11:48AM

A couple of clip-on heatsinks seriously wouldn't go amiss. C'mon, think how easy it would be to just grease up a mica pad and drill some metal.

If in doubt, run it for a couple of seconds. Unplug, discharge, feel temp. Then try the same for a couple of minutes and see if you're comfortable without heatsinks

Re: Help with off-line smps
uzzors2k, Mon Aug 14 2006, 12:06PM

I guess some heatsinks won't hurt, I'll see what I can find.

Should I ever run the PSU without a load? Because I was going to temperature test just now, but the output hit 50 volts and my 10 ohm load resistor started smoking in seconds. If I need a load what is sufficient? A 20 watt load?

Re: Help with off-line smps
Marko, Mon Aug 14 2006, 01:25PM

Good work uzzors

You can add regulation later if you come into need of a specific voltage..

Re: Help with off-line smps
Steve Ward, Mon Aug 14 2006, 03:35PM

You have discovered the need for feedback to provide voltage regulation!

Fet and diode ratings (for packages that *can* be mounted to a heatsink) are rated for being properly heatinked. Your fet, if it were running 14A would be dissipating many 10's of watts, and would melt itself without a heatsink. Some of my really low power SMPS stuff, i have gone without a heatsink, but you mention 250W, which sounds like enough to warrant at least a small heatsink for the mosfets and maybe the bridge rectifier. Most computer power supplies use a dinky heatsink, but remember they have a pretty good forced air system as well, greatly increasing the dissipation ability of the heatsinks.

Re: Help with off-line smps
uzzors2k, Sat Aug 19 2006, 07:14PM

Thanks Firkragg.

I've been checking out just about every smps project on the net, so I think I can get the voltage regulation working fine. The problem is voltage regulation relies on modifying the duty cycle, something I've been having problems with.

The problem is the GDT, because input signals are perfect, but outputs are completely different. The GDT is wound 1:1:1 with 45 turns per winding. Here are the waves, 1st one is at 50% duty cycle while the other is near absolute minimum. I've marked the rise and fall of each wave since it might be confusing if you aren't actually watching it change. The phase also seems to shift about mid-duty cycle which makes things even stranger.(entire wave jumps a box(5µs) ahead, just check the pictures) At maximum duty cycle the waveform is perfect however.

I hope someone knows how to fix this because I've spent all week at it.

Re: Help with off-line smps
Steve Conner, Sat Aug 19 2006, 07:55PM

I can see a few things that might be wrong.

1) The wiring between the TL494's output pins and the transistors that drive the GDT is wrong on your schematic.
Pin 8 should connect to the base of the top-left transistor
Pin 9 to the base of the bottom-right one
Pin 10 to top-right
Pin 11 to botom-left

2) The transistors should probably have antiparallel diodes across their collector and emitter to allow a path for the GDT magnetizing current.

3) The driver has a design flaw. It turns all four driver transistors off during deadtime, leaving the GDT primary open-circuit. You're supposed to short the primary during deadtime in order to slam all the MOSFET gates quickly to zero volts.

By the look of your screenshots, I assume you already found 1) and 2) and fixed them, but 3) explains why you get the weird ringing and reversals, as the GDT's magnetizing current tries to find a way out during deadtime. This needs a redesign of the driver circuit to fix it properly, though as a band-aid you could maybe just connect a large 22 ohm resistor across the GDT primary to damp the weirdness (and increase your gate driver power consumption hugely too)

Re: Help with off-line smps
uzzors2k, Sat Aug 19 2006, 09:15PM

Thanks for the quick reply. This is the first time I'm using a tl494 so the tl494 schematic is something I got from someone else. Infact I haven't fixed anything on it, besides adding a 360 ohm resistor in series with the GDT. Without it the ouput is completely distorted.

I know enough about the tl494 now though, so I'll make my own driver. Hopefully with more success.