OMG Induction Heater / General Science and Electronics / Forums

Forums

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

« Previous topic | Next topic »

OMG Induction Heater

first previous 4 5 6 7 next last

Move Thread

LAN_403

Tonskulus

Wed Jan 23 2008, 11:31AM

Registered Member #1223 Joined: Thu Jan 10 2008, 04:32PM
Location:
Posts: 133

Same happened again as always for me. It seemed to work just fine at 110V supplyvoltage. IGBT's and freewheeling diodes didnt get hot, just slightly warm.
Inverter voltage vs. current:
http://www.elisanet.fi/tonskulus/inverter_out.jpg

And then, i decided to try this at full 220V supply voltage. I also added more inductance for zmatch to keep supply current at reasonable level, still 5..7Amps. It didnt last even one minute before igbt failure happened. Both were shorted. Is it possible that some shoottrough happened there for some reason? I had always the same problem with halfbridges driven at >100Volt supply. they wont suffer.

GeordieBoy

Wed Jan 23 2008, 01:51PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

Although the devices died cold, shoot-through seems unlikely as there would be a lot of noise on the voltage waveform. Diode reverse recovery problems also seem unlikely as ringing on the current waveform suggests switching takes place before the current zero-crossing.

The voltage waveform looks strange though: Rounded leading edges and almost a 60% voltage overshoot! This may be exceeding the rating of your switches at the peaks of the supply voltage. That cosine shaped wiggle on top of the inverter voltage waveform (which should be flat) is usually indicative of excessive inductance in the DC bus wiring to the inverter. Try adding more capacitance across the DC bus right where the IGBT's are located.

Have you checked what your gate drive waveforms look like at the IGBTs during operation? It's also worth checking that your x10 (or x100) scope probes are correctly compensated before making these high-frequency measurements.

Tonskulus

Wed Jan 23 2008, 07:27PM

Registered Member #1223 Joined: Thu Jan 10 2008, 04:32PM
Location:
Posts: 133

I just noticed that there is overshooting problem. Without load at all (inverter output open), there is still current going through igbts. 200mA or so. I did try adding 10ÂµF capacitor to the DC supply rail and it helped smoothing invertrer voltage.:
http://www.elisanet.fi/tonskulus/ih/volts.jpg

GeordieBoy

Wed Jan 23 2008, 08:53PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

That output waveform looks a lot better now.

With no load on the output of a half-bridge (or full bridge) you always get some current drawn from the supply, particularly when running at high frequency and high voltage. This 200mA current does not necessarily indicate shoot-through due to conduction overlap in the IGBTs. It may just represent the power required to charge and discharge the output capacitances of your switching devices.

(The power dissipated to charge/discharge device capacitances alone can get high for large die devices at high frequencies and high supply voltages. Eliminating this loss is the main motivation behind ZVS schemes like Class E, etc in power electronics.)

Tonskulus

Wed Jan 23 2008, 09:58PM

Registered Member #1223 Joined: Thu Jan 10 2008, 04:32PM
Location:
Posts: 133

Ok, i have been running this inverter today for a while with new igbts. Everything seems to be ok at 110Volts. I pushed this to the limits, 15Amps input current and no problems. But im sure it blows up if i switch to 220V. Thats what it always does. Yesterday I was running off 110Volts / 10Amps and decided to switch 220V. IGBT failure happened at only 4Amp supply current.

kim_ladha

Wed Jan 23 2008, 10:12PM

Registered Member #1217 Joined: Mon Jan 07 2008, 11:46AM
Location: Leicester, UK
Posts: 11

Hi Tonskulus, Well done. Looks ok. The waveforms are either shoot-through or too much inductance on the low side drain- otherwise it should be working by now. Scrap that 10u cap though- the waveforms are fine. Geordie Boy hasn't realised the reason for the dodgy looking waveform- its is tracking the mains voltage. ....UPDATE- Yes he did! He has added a thread just after I posted this....This keeps your power factor high. I get 0.99 PF which is as good as a comercial unit!

Shame about the IGBT failures though- just looked at your other waveforms. The 10u won't solve shoot through it will just damp the ringing frequency. You need a dead time- I use 500ns ant 120khz- its high but seems to work reliably. You have some high frequency ringing on the collectors that looks suspicious- look for stray inductance. Finally make sure your oscillator is far from the work coil and shielded and the current feedback to the oscillator is minimal- current transformer saturation led to early unexplained failures for me.

Hi Steve, Awsome idea about the little cap to drop the current- Better than my solution. As for ZCS, it only occurs at high currents with low work coil loading. The work coild current is supperimposed onto bridge output by the ratio on the coil inductance to the inductor. If i have a small inductor i can get good zcs but high device currents. If I have a large inductor my current is almost triangular all the time. My ratio is 10-20 to 1 deppending what power i want.

Geordie Boy, the 3 kW low load power goes like this... all approx- 70w into switches, 100w into inductor (resisitive), 135w into my caps and the rest into water heating through resistive work coil losses (no work piece)- water goes in at 15 degrees and comes out at 70!

Regards to all
Karim

GeordieBoy

Wed Jan 23 2008, 10:16PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

Tonskulus: For debugging unexplained failures it helps if you can run it from a regulated adjustable DC supply to remove the 100Hz envelope blurring. Wind the voltage up slowly towards 340VDC whilst watching all the waveforms on a scope. The clues are there!

At the risk of getting my backside kicked for turning this thread into a lesson on inverter testing I'll leave it there.

-Richie,

Tonskulus

Wed Jan 23 2008, 10:38PM

Registered Member #1223 Joined: Thu Jan 10 2008, 04:32PM
Location:
Posts: 133

Without 10ÂµF DC filtering cap there are huge amounts of ringing at inverter output! Anyway, im using 50Amp IGBT brick instead of HGTG30N60xx's. It drives ok at 100kHz. I have to get rid of some stray inductances and make a shield for oscillator/driverboard if it helps, just like Kim told me to do.
I'll keep working :)

Experimentonomen

Thu Jan 24 2008, 05:40PM

Registered Member #941 Joined: Sun Aug 05 2007, 10:09AM
Location: in a swedish junk pile
Posts: 497

Hey kim! What is your comments on my board design ?

Tonskulus

Fri Jan 25 2008, 03:32PM

Registered Member #1223 Joined: Thu Jan 10 2008, 04:32PM
Location:
Posts: 133

Finally Im able to melt some steel parts with this thing. Still 110Volt supply is used as it seems to work reliable now, I have no reason for higher input voltages to be used. Not for this one..maybe I can make another system for higher powerlevels later but its another story then. Ok, I made a short video about bolt being melted. Input voltage 110Volts and current 10..15Amps, 20A is maximum (limited by isolation trafo).

http://www.elisanet.fi/tonskulus/ih/ssih13.avi

My workcoil didnt suffer, it collapsed and destroyed (full of holes) because there is no water cooling:
http://www.elisanet.fi/tonskulus/ih/bustedcoil.jpg

first previous 4 5 6 7 next last

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