unstable mosfet driver

[jackXallXmasterX0]

This is a long shot, but hoping someone can provide some insight on why I keep popping the fuses on this circuit, it seems to be generally after an IGBT blows. Maybe an idea on circuit protection. The circuit is for ultrsaound cleaning, and cleans well for about 1 minute sometimes. The gate drivers and IGBTs are heatsinked.

attached.

thanks

 

[itsmoked]

Is it your circuit or are you repairing a commercial unit?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Comcokid]

Design and layout of switch-mode power circuits, even one with no inductors requires careful attention to parasitic capacitance and inductance. Even at low frequency the actual layout and construction of a circuit may be more important than the schematic. Although your circuit is working, there could be ringing occuring that is exceeding the component ratings, and results in failure after a few minutes.

A few suggestions:
Lead length and placement around the VS and VB nodes of the half-bridge driver need careful consideration, especially your 0.1 VB-to-VS cap. Check the IRF website for app notes.

Move most of your 0.1u caps and maybe one or two of the 1u caps close to pin 3 of your top IGBT. What you are rectifing from the transformer is very low frequency, while the current rise-time near the IGBT has very fast risetime (high frequency components) even if the function generator frequency is low.

Add a 10uF ceramic cap from VCC to COM near the half-bridge driver and maybe a 0.1uf cap as well.

Remember - lead length is inductance. Wires or leads close together form a capacitor. Learning to spot possible issues takes careful thought and experience.

 

[hacksaw]

why have you placed the 1A fuse in the location indicated?

 

[jackXallXmasterX0]

wow, alot of responses, thanks all for the help

one of the things that is very frustrating is that the gate driver circuit will run all day long on a normal power supply at 60V. Its only when i hook up the higher voltage output (450V transformer) that it starts intermittently blow IGBTs. So it can drive the ultrasound cleaning for 5 minutes sometimes without problems, other times, it blows right when I flip the power switch. During the time its running, at 450V, the output and VCC (10V) and 450V look pretty clean and not crossing the 600V barrier, not even 500V. ..
comcokid - I will give all of your suggestions a shot on monday. I understand exactly what your saying. It has to be one of those issues.
Even with the issues you have said, with a 600V limit, is there a fair limit on how close I can work near it? Maybe 550V, 500V, 450V. I guess it depends on parasitic inducates, capacatiances, etc… but what is a far goal, 450

hacksaw - the 600V 1Amp fuse is in that location as to try to minimize the effect of capactior charaging on the fuse blowing. I think i can more accuractly guage the current after the rectification process based on the load.

itssmoked - this not a circuit im repairing, its a design.

 

[hacksaw]

fuses are notoriously un-reliable at current limiting, if you have any capacitive load in the sonic transducer, when the IGBT first starts to conduct you get enormous current surges, i.e., you are connecting two capacitors in parallel one of which is uncharged.

with 1 ohm of series reistance that would be 60 amp surge when run at 60 vdc, but 450 A at the higher supply voltage.

a 5-10 ohm resistor would help protect your device, if worried about damage to the resistor the use a non-inductively wound device

 

[VEBill]

...why I keep popping the fuses on this circuit, it seems to be generally after an IGBT blows...

The real question is (obviously) why the IGBT blows. Lots of good advice above.

If it was just a fuse problem, then you could use a slow-blow fuse.

 

[geekEE]

I agree that proper layout is critical in a circuit like this. You might be getting shoot-through. You could try changing to the 21844 driver and increasing the dead time.

 

[electricuwe]

Keep in mind that the specified voltage for IGBTs and HVIC divers should be regared as an instanteneous voltage rating, not as the DC-link voltage the are designed for.

600 V devices are intended to operate on a DC-link voltage in the range from 310 V to 400 V from rectified 230 V single-phase (without PFC or With PFC).

Due to parasitic inductances you will have overvoltages in the ns range when they turn off. And even a voltage spike of on 10 ns is expected to damage the IGBT if it exceeds the rated voltage.

Even if you cannot identify any volatge spike with a properly set up scope and probe, operation at such high DC-lnk voltage bears some special risk for such devices.

Furthermore the monolithic driver ICs are very prone to negative voltage spikes at the switching node.

Either change to higher voltage devices or change from half-brigge to fullbridge topology (at reduced DC-link volatge).

 

[xnuke]

I'm not much of a circuit designer, but I'd include anti-parallel diodes across the IGBTs to provide some protection for them against voltage spikes. Also, if I remember correctly, a series resistor is needed between the slow-blow fuse and the connection of the Zener diodes to allow the Zeners to regulate voltage properly.

I do have some questions:

What is the load? Is it capacitive as you show it? What are its ratings?

What is the function out of the function generator?

xnuke
"Live and act within the limit of your knowledge and keep expanding it to the limit of your life." Ayn Rand, Atlas Shrugged.
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[jackXallXmasterX0]

xnuke - the average capactative load is from 4 to 9 nF. The frequency range is around 100khz.

electricuwe - i think you have the general idea, Its just close to the maxiximum voltage. I do have the scope set up to trigger so i can zoom the time axis in to see the rising and falling edges, and they are a little noisy but not bad. But i still think it is the problem.

geekee - ill try some of the 21844 and let you know.

V1Bill -- ive tried some slow blow fuses today without luck. I think the IGBTs are taking them out.

hacksaw - a resistor to drop the voltage and reduce the load is a good idea and Ill try that today. I think this may need to occur in conjuciton with electricuwe idea of dropping the DC link voltage


will get back and post the solution that fixes it first.( there maybe more than one solution)
hank you all again

 

[iop995]

How long are cable that connects load? Try to modify up or down frequency. It's possible to be close to series resonance. Maybe a soft-start circuit help to avoid IGBT demage.

 

[MagicSmoker]

My $0.02: perhaps the fuse is blowing because there is no current limiting resistance before the ~510V stack of zener diodes? I'd personally replace that with a traditional emitter follower series regulator circuit as described on this page (among many others):

http://www.radio-electronics.com/in...rent_limiter/power_supply_current_limiter.php

That said, you also need to put resistors in series with each IGBT's gate (and as near to the gate as possible). The datasheet for the IGBT will usually give a recommended value for the gate resistor (5 to 20 ohms is typical for this size range). Leaving these resistors out is especially bad when using bootstrap driver ICs like the IR2184, IR2113, etc., as this both causes the bootstrapped supply to sag badly every time the IGBT is fired and/or causes the IGBT to switch too fast, resulting in too high a dV/dt (which leads to both IGBTs turning on, shorting out the DC bus).

One other thing: ultrasonic transducers have a complex impedance characteristic which makes them difficult enough loads to drive, as it is, but they also have a high "Q" which basically means they have to be driven with a frequency that is within a few 10s of Hz of resonance.

I've been tinkering around with ultrasonic transducers recently for the purpose of welding plastics. I've some familiarity with resonant and quasi-resonant SMPS design which kind of helps here, but these are still difficult beasts. Good luck!