Exploding high current MOSFETS

Thread Starter

stobby

Joined Jun 20, 2018
26
You do know that term "pulsed Mig" is a misnomer and marketing hype don't you? By its nature, all Mig is pulsed it is so by the way the filler wire melts on contact and then feeds into the work again. Doing this make it a series of pulses.
Adding a electronic pulse won't in any way make either your welds or you as a welder better. That only comes from practice and good training.
8930.gif
Yes it is a real thing, And it is not to do with making myself or anyone a better welder, It reduces heat input into sensitive materials like aluminum whilst the high peak currents maintain a spray arc condition.
 

Thread Starter

stobby

Joined Jun 20, 2018
26
MOSFETs can self-balance when in parallel, only when they are turned ON and OFF at essentially the same time.
The gates on these MOSFET's are not being driven hard.
Therefore, the chance of "balancing" is ... not likely.
The Gate Drive Circuit shown in the schematic is a design that will blow-up MOSFET's.

Use a pair of Dual MOSFET Gate Drivers ( something like this ) ...
https://www.infineon.com/cms/en/product/power/gate-driver-ics/irs4427/?redirId=65104

Maybe, even consider a Gate Driver that sink / source: 4 amps or even 5 amps.
Add Low ESR capacitors to provide the amps needed to drive these gates very quickly.

How to select the proper MOSFET Gate Driver
https://www.electronicproducts.com/Analog_Mixed_Signal_ICs/Selecting_the_right_MOSFET_driver.aspx

NOTE: IRF1405 PACKAGE LIMITATION IS 75 AMPS per MOSFET !
The first MOSFET on, or the last MOSFET off, has a big problem.
So with these gate drivers i would have presumably as close to a 20vdc source as possible, And they accept the signal from the signal generator? I always presumed mosfet gates didnt require any current?
 

wayneh

Joined Sep 9, 2010
17,152
So with these gate drivers i would have presumably as close to a 20vdc source as possible, And they accept the signal from the signal generator? I always presumed mosfet gates didnt require any current?
They need very little current for DC or low speed switching but they’re capacitors. At higher speed they need sizable current to switch them very quickly both on and off. Your schematic shows a large resistance in the discharge path and so they’ll spend way too much time in the linear region where they pass large currents but also show resistance. You’ve seen the result.
 

crutschow

Joined Mar 14, 2008
27,237
Below is the LTspice simulation of your circuit.
Note the large delay in the MOSFET turnoff due to the large gate capacitance.
The peak MOSFET power dissipation approaches 400W during this turnoff (green trace) which can zap the transistor.

That's why you need a high current (in both directions), low impedance driver for the gates.

upload_2019-2-24_0-5-19.png
 

kubeek

Joined Sep 20, 2005
5,753
@stobby I don´t think that switching the electrode off as you plan it is the way to go. The current must keep flowing just at a lower level, otherwise you won´t get that two stage waveform, plus you will have all the problems with inductive kickback that you are having now. I think that the welder machines achieve this by lowering the set voltage in the control circuit. With a tapped transformer, I would go with switching between taps.
Do you have a diagram of your welder? That will help a lot to fit what you want to what you actually have.
 

drc_567

Joined Dec 29, 2008
1,113
... is V[N001] the voltage across one of the M4 mosfets?
If that is the case, then there is essentially no power loss in the individual mosfets until the slow gate response creates a problem.
... seems kind of remarkable that the individual mosfet power loss during the current pulse is next to nothing, even at 50 amps per mosfet.
... datasheet gives a mosfet on resistance of 4.6 milliohms, so that would explain the minimal voltage drop.
 
Last edited:

mvas

Joined Jun 19, 2017
538
Hey everyone,
I have made a circuit to convert straight dc to dc square wave at relative low voltage (30v) but high current (around 200amps). I have trialed this circuit at 12vdc with a spotlight as the load and it works perfectly. however when it i trialed it at higher power levels the Mosfet's almost instantly exploded and or cracked. These mosfet's are rated at 55v, 169amps.
You say, " ... the Voltage is 30 Volts Straight DC ... "
Could the voltage actually be more like 120 Hz Full-Wave Rectified DC and not Straight DC?

So, what is the "Peak Open Circuit Voltage" when not welding?
You might consider MOSFET's that are rated 2X your "Peak Open Circuit Voltage"

My friend had a 200 Amp Welder and it was rated at 60 Volts Peak Open Circuit.
And you used MOSFET's were rated at only 55 Volts?

If there is an inductor in series with the output, then calculating Max Voltage gets really complicated.

One DC RATED (50-100 Amp?) Fuse per MOSFET might protect your MOSFET's during your experiment?
 
Last edited:

drc_567

Joined Dec 29, 2008
1,113
You say, " ... the Voltage is 30 Volts Straight DC ... "
Could the voltage actually be more like 120 Hz Full-Wave Rectified DC and not Straight DC?

So, what is the "Peak Open Circuit Voltage" when not welding?
You might consider MOSFET's that are rated 2X your "Peak Open Circuit Voltage"

My friend had a 200 Amp Welder and it was rated at 60 Volts Peak Open Circuit.
And you used MOSFET's were rated at only 55 Volts?

If there is an inductor in series with the output, then calculating Max Voltage gets really complicated.

One DC RATED (50-100 Amp?) Fuse per MOSFET might protect your MOSFET's during your experiment?
... take a look at the Fig. 3 plot here:
https://en.m.wikipedia.org/wiki/Power_MOSFET#/media/File:Bv_rdson.png
... The mosfet on resistance seems to increase as the breakdown voltage increases, above 50 volts or so. [The plot image did not copy by itself.] ... Probably don't want the mosfets warming up.
 

shortbus

Joined Sep 30, 2009
8,593
Do you have that welder? If not your doing or trying to do what I suggested. And if you have that welder what are you trying to do? And to use that a consumer grade welder, meant for DIYer's is just more fact to my earlier post about leaning to be a better welder by taking some training not just learning on your own. I did the learn myself before getting the real training.

I don't think what your trying to do is how the pulses in that welder are working. It seem not to be pulsing the actual welding current but is pulsing the wire feed. Quote, "Our advanced, synergic programs and the Swiss-made wire feed mechanism featured on the Pro Pulse 200 enable you to feed aluminum wire with repeatable precision through our standard MIG gun. The ability to maintain a stable and smooth arc at 22 amps allows you to weld aluminum as thin as 1/32” (0.031”). Our pre-programmed Double Pulse program allows you to make “MIG like TIG” welds with the desired “stack of dimes” look." That is from your linked page.
 

Thread Starter

stobby

Joined Jun 20, 2018
26
Thanks for the simulated scope shot that has really helped me to understand what's really going here.

No I don't own that machine, that's what I am trying to replicate. And yes it does modulate the waveform. Check some videos on YouTube about pulsed GMAW.

The machine I have is a Mig star 250 amp cv only.

Hence the term constant voltage, the highest voltage I can push from this source is 45v open circuit.

Here's a link to the user manual, the last few pages show the basic design and circuit.

Thanks for everyone's patience.
 
Screenshot_4.png For this type of circuit I used TC4422 as a driver, and there is another important aspect: you NEED to provide a path for the inductive output current of 200A to circulate when the transistors are blocking, oitherwise the spike will guarantee destroy them; for that will be useda set of flywheel diodes in antiparallel with output, before the wires to the welding tips. I would strongly recommend some 5-10 pcs of HER605 in parallel; No Shottky's because are too fragile.

The transistors need also snubbers: each of them a C-R, with C of 10nF, R of 3.3R; in gate to be a 4.7R resistor for each transistor. And the supply line to have a 1000-2000uF capacitor just before the switching transistors rail, and a supression diode n paralel with it, of 35-40V at 1.5kW. Some ceramics of 100nF in parallel are good also.
 

Lestracy31

Joined Feb 18, 2019
1
At what frequency are You switching? You need a gate drive, but have you looked at Igbt? They are much more ideal for high power medium switching applications. Cheap with much higher current capabilities. Nothing works in the linear region though, so everything stated previously needs to be addressed.
 

shortbus

Joined Sep 30, 2009
8,593
When doing this to a welder that was never made this way from the factory where is the "low current" for the pulse coming from?
 

LowQCab

Joined Nov 6, 2012
569
I agree with Adrian,
although my quick scan didn't see the reason WHY stated.
Anytime you create an arc, you will inevitably create RF HASH.
( I've run into this problem on high current, brushed, DC motors )
With that much power going into an arc you are creating a lot of inductive flyback, (no iron core required),
just the collapsing of the giant magnetic field around the arc its self creates a huge "single wire" inductor.
A Common Mode Inductor that can accommodate your cables, both, side-by-side,
may be in order to protect the FETs from the generated RF hash.
This may lessen the need for, or augment, all those Caps and Diodes.

See if you can find a schematic for the type of unit you want,
sometimes the manufacturers come up with some
really weird stone-age "sticks and rocks" solutions,
which might make an electrical engineer cringe and become nauseous,
but they work, and are reliable, never the less.

Unless you are in this for the adventure of Doing It Yourself,
it's just about guaranteed that you will ultimately spend
more money on the learning curve, than you will save with DIY.
The manufacturers have already spent a big chunk of time and money working out the bugs.

Alternatively, look for a used one.
It might take a couple of months to find a deal, but it might be an excellent deal.
.
.
 

Thread Starter

stobby

Joined Jun 20, 2018
26
Below is the LTspice simulation of your circuit.
Note the large delay in the MOSFET turnoff due to the large gate capacitance.
The peak MOSFET power dissipation approaches 400W during this turnoff (green trace) which can zap the transistor.

That's why you need a high current (in both directions), low impedance driver for the gates.

View attachment 170863
Sorry for late reply everyone, have been without internet for weeks.
I have been doing some trialing on LTSpice and have come up the a more suitable circuit. i have simulated a driver to sink and source the gates as quickly as possible.

As you can see the power dissipation has decreased to under 300w, And the rise and fall times are far better.on off time.JPG
power diss.JPG
pulse mig iso drain.JPG
 
Top