Switch-mode bench supply problems / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Switch-mode bench supply problems

1 2 next

Move Thread

LAN_403

uzzors2k

Sun Aug 26 2007, 12:43PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

My switch mode bench supply is nearing completion. It's features are variable constant voltage or current and short-circuit protection. Regulated voltage range is 4 - 50 Volts, regulated current range is 4 - 13 Amps. Output power should be 700W peak.

Having lashed everything up yesterday, I ran a short test with a 10 ohm load to see that the voltage was regulated. Everything was fine up to 30 volts when suddenly something started hissing and the voltage dropped. The power LED was fluttering and dim. I didn't think much of it at the time, as the 10 ohm load was only rated for 10W, and it was running hot enough as was.

Eager to load the supply as heavily as possible, I put together a Mazzilli driver. Cranked the voltage to 50V and flipped the switch. Short-circuit protection goes. Great, I think, and turn the voltage down to 10V and turn on the power. This time I get the same hissing/buzzing, the voltage has dropped and the flyback isn't arcing. I feel the IRFP450s (on the smps) and they are sizzling hot.

It turns out the mysterious hissing noise comes from the GDT and messes up the drive signals. The IRFP450s are in the linear region and little power is sent through the main transformer. The same hissing noise came on and off while just testing the GDT and PWM section, but it seemed to come at random then. Like when powering off the circuit and then quickly reapplying power, which causes errors in most circuits I've played with earlier.

So somehow heavier loads cause the PWM section or GDT to send faulty drive signals, which make the GDT hiss/buzz. Anyone have a take on this?

Here are the schematics for everything. Things not mentioned in the schematics: No mains input filter, no primary side over-current detection installed and the PWM section is powered by a common 18V wall wart with 1000ÂµF output filter.

1188131935 95 FT0 Bridge Recticfier Stage

Dr. Dark Current

Sun Aug 26 2007, 01:14PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Just a small problem I can see, how do the bipolar transistors like -16V on their bases? In the datasheet, there is stated maximum Veb of 5 volts.

Marko

Sun Aug 26 2007, 01:25PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Hi

The first thing I notice, your buck inductor is way too small. It may be a difficult thing if you want to vary output voltage a lot.

I found this nice calculator

you could use. Largest inductor is one for 0.5 voltage ratio and I think you should start from there.

I don't know what frequency are you using, but example for 50kHz would be about 33uH.

The 'no mains filter' thing is also disputable, and that's the one question that interested me too.

Designers use boost converters for power factor correction in front of these circuits.
I thought, could one get rid of this and still have a high PF supply?

By using no input filter and moving all the filtering to the output I thought if you could keep power factor high and complexity down. Anyone has thoughts on that?

Without scope shots I can't tell you much more. You should best go step by step, first assure your gate drive is working well, then carefully apply voltage to the bridge without anything on the transformer secondary.

Then use a simple load like a bunch of bulbs and load the thing down to power level you want.. If everything is fine proceed to output buck and filter section.

Marko

Steve Conner

Sun Aug 26 2007, 10:08PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Do you have adequate decoupling on the 16V supply? I'd recommend a several hundred uF electrolytic, physically close to the gate drive transistors and TL494.

Marko

Sun Aug 26 2007, 10:24PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Steve what do you think, is it possible to correct cos Ï€ by using no input filter and larger output bulk capacitance. With feedback it would, as I think, behave similar to those PFC boost converters? From the other side I guess there must be a reason why everyone uses separate boost converters. I only seen small flyback using the topology described.

Steve Conner

Mon Aug 27 2007, 08:46AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Hmm, well that is an interesting idea but I doubt it would work as-is. The feedback loop will try to keep the output voltage constant, and in doing so it will draw all sorts of weird current waveforms from the line. Probably the inverse of what you want.

Also, because the circuit is a buck converter, it can't ever draw a perfect sine wave no matter how it is controlled. You need a boost converter to reach down into the troughs of line voltage that are lower than the reflected voltage on the output filter capacitor.

My approach to this problem would be to redesign it as a flyback converter controlled by a boost PFC chip such as the L4981. The penalty would be a lot more 100Hz ripple on the output, much poorer response to load transients, and a bigger magnetic core than a forward converter would need. But if you just want eleventy billion watts for blowing stuff up, you probably won't care.

Marko

Mon Aug 27 2007, 11:50AM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Yes, I see what's your point. Forward converter would not be able to work when transformer secondary voltage drops lower than output voltage. I thought if it could be made usable by using a smaller ratio transformer: if secondary output is 40V and bucked to 10V, PFC would work all way down to 80V. This would make efficiency inversely proportional to repair in PF. So meh.

It's in most cases probably just easier to use a small inductor in seres with input.

I know it's done with flyback converters, but that's another story.

To return to the thread: I don't know how much use will 'constant current circuit' be, when you have a large bulk capacitor on the output... You would have to have load always connected unless it doesn't mind a bulk cap discharged into it with full energy.

It will also behave strangely without load; it will charge the capacitor to maximum voltage transformer can put out, about 67V in this case which is bad for your 63V cap.

If you wanted to use the supply in voltage mode after that again you need to bleed this from the cap somehow. Waiting for bleeder to do it before you connect a load is sort of goofy.

From other side, I don't think you'l need such a constant current source in your life, unless you mess with exotic loads like large laser diodes.
If you ever do, I'd rather build a special SMPS only for that purpose.

Small linear source is much better and probably more than you'l ever need.

uzzors2k

Mon Aug 27 2007, 09:00PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

I found the problem. All it took was putting a capacitor and resistor between the inverting input and the compensation input. (Pins 2 and 3) What exactly does this do? Reduce the gain and add an RC time constant, so the feedback is less trippy? I also increased the output inductor to 40ÂµH, and gave a decoupling cap to the TL494, which helped as well I imagine.

So far I've just been able to retune the voltage adjustment and test with a 10 ohm load. Voltage didn't drop at all (100mV resolution on my multimeter) with 220W load, so I'm optimistic. I pretty tired ATM, so I won't be able to test the ZVS until tomorrow. School always seems to interrupt the things that really matter in life...

About the current regulation, I think I'll just route it into the other comparator and use it for current-limiting. That way I can run high current loads without having to set the voltage just right. I had that problem with my MOT supply, quite often current spikes would trip the crow-bar when there was no real hazard. The RC network on the current-limit comparator and current-limiting regulation should make those annoyances a thing of the past.

Well, tomorrow I'll see how it handles with the Mazzilli driver. Oh, and for anyone looking to make a smps with the TL494, be sure to read this.

thedatastream

Mon Aug 27 2007, 09:17PM

Registered Member #505 Joined: Sun Nov 19 2006, 06:42PM
Location: Yorkshire!
Posts: 329

Compensation reduces the AC gain so that at frequencies where the gain of the feedback loop is more than 1 the phase of the feedback loop is less than 180 degrees. There are some rare exceptions to this, which can occur with Power Integrations TOPSwitch controllers.

When gain > 1 and freq > 180 degrees then these are the conditions for oscillation.

There have been many texts written on compensation in SMPS, Texas Instruments have a particularly good one -

- especially for voltage mode feedback like your TL494

uzzors2k

Wed Aug 29 2007, 02:18PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

I was unable to post yesterday due to problems with my laptop, however I was able to test the power supply. The testing of the Mazzilli driver went without a hitch. The voltage didn't so much as flinch. The power wasn't too high though, I was running the flyback at 100kHz and the arcs were long and spindly. So I upped the resonant capacitor from 50nF to 220nF, and turned on the power supply. The flyback started hissing and sparking much more, and I knew it would be a memorable arc. Then, just as the spark initiated, before the arc could form, both half-bridge fets exploded.

I replaced them, but the problem seems much deeper. Upon turning on the power supply nothing happened. The gate drive was working and the mains rectifier was working, but no output voltage. So I rechecked connections, turned on the power again and the power LED lights for a split second and the breaker trips silently.

I suspect the secondary rectifiers have failed, which I won't be replacing anytime soon. I'm keen on giving up on this project as I've learned and seen what I wanted. And quite frankly I prefer running stuff directly off-line. I think I'll just make an isolation transformer/variac combo instead, which should be much more durable.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob