Hi,
So Im working on what is basically a fancy capacitor discharge welder, and I keep frying the mosfets so they get stuck on.
Everything works perfectly at low voltages (~12V) but only for a few times then the mosfets are stuck on and no longer function as switches. Ive gone through so many of these things its time to ask for help.
Attached is a wiring diagram of the power circuit. the lower part is the driver/mosfet/capacitor bank I'm having trouble with, I included the top circuit (which is just a way to sense head contact for the controller) just in case it's causing something funny to happen.
Basically the weld head comes down, both electrodes make contact with the metal, then after a short dwell period the caps fire for about 300us.
The resistance of the weld head/ and wires is very very low so I made a .05 Ohm "resistor" out of resistance wire used in heaters which I confirmed to be ~.054 Ohms. I originally had 12 mosfets in parallel but after a few failures doubled down and put 24 but they still fail closed at the same voltage level so I don't think it's a current problem.
I'm thinking maybe there's a voltage spike which is killing the mosfets but have no way of measuring it and they die when the capacitors are only at 15v so it would have to be a big voltage spike... I also might just be doing something obvious and dumb lol
Any ideas or pointers would be greatly appreciated!
Thanks guys!!

 

Here's a very lengthy thread that led to development of a semi-commercial capacitor discharge welder controlled by mosfets:
http://www.eevblog.com/forum/projects/guesses-on-what-i-am-attempting-here/msg1232857/#msg1232857

The author (tatus1969) has a site to offer the parts: https://www.keenlab.de/index.php/portfolio-item/kweld/

He wants to keep the code private, but I believe everything else, including the schematic has been published.

 

A single FET exceeds the gate driver capacitance rating at 20 nS,
so that tells me the MOSFET is in the active region for some time
because total Cload of driver >> than its ratings. That in turn causes
real high Pdiss.

But in your case your duty cycle is real low ? Ie. you only fire at a very
slow rate ? Like << 1 Hz ?

The other thing is you have no gate damping Rs in series with each
MOSFET gate, and or a ferrite bead. This to prevent HV gate failures
from inductive transients, from layout, wire runs.....

Curious do you know what the ESR is of the bulk caps is ?

Do you have a Digital Storage Oscilloscope, such that you could capture
a HV one shot sweep if thats is what is occuring ?



Regards, Dana.

 

......the caps fire for about 300us.....

This seems a ridiculously short pulse , to get enough energy into the weld you would need a current of over 10KA enough to fry anything ...

No need for such a short pulse , adjust it to 10ms and your troubles will be over I'm sure.

I've just finished building a small spot welder for fixing nickel strip to 18650 cells , my pulse is about 50msec @ a few hundred Amps

 

Post a photo of your wiring and general setup.

Your gate driver should be located close and tight to the MOSFETs, as in keep the leads as short as possible. (like less than an inch!)
It matters when you are trying to drive them fast and hard. (read the application notes for the gate drivers)
The ground return path is equally critical. The inductance of even a few inches of wire will wreak havoc with switching speeds.

As designed, all your energy will be dissipated in the 0.05 ohm sense resistor, it's value should be a very small fraction of the overall weld circuit impedance.

Consider using a current transformer here.

 

A single FET exceeds the gate driver capacitance rating at 20 nS,
so that tells me the MOSFET is in the active region for some time
because total Cload of driver >> than its ratings. That in turn causes
real high Pdiss.

But in your case your duty cycle is real low ? Ie. you only fire at a very
slow rate ? Like << 1 Hz ?

The other thing is you have no gate damping Rs in series with each
MOSFET gate, and or a ferrite bead. This to prevent HV gate failures
from inductive transients, from layout, wire runs.....

Curious do you know what the ESR is of the bulk caps is ?

Do you have a Digital Storage Oscilloscope, such that you could capture
a HV one shot sweep if thats is what is occuring ?



Regards, Dana.

Hi thanks for the reply,
The ESR of the caps is somewhere around 30 milliohms (Theyre ALF20G223EH063)
It will be low frequency (1 weld every second ish) however I eventually the controller has the capability to switch the weld on/off at up to about 20 kHz for different applications. Just not doing that yet.
I do have a oscilloscope, thats how I know my driver circuit and everything upstream of the mosfets is working properly. I tried to capture the weld last go around but I only got one shot (because the mosfets failed) and the oscilloscope didnt catch it. its a few hours of wiring up new mosfets to try again and I know ill get the same result without changing something. You think maybe this is just a matter of adding some damping resistor to the gate pin? Im a structural engineer, so a bit out of my element here...

 

This seems a ridiculously short pulse , to get enough energy into the weld you would need a current of over 10KA enough to fry anything ...

No need for such a short pulse , adjust it to 10ms and your troubles will be over I'm sure.

I've just finished building a small spot welder for fixing nickel strip to 18650 cells , my pulse is about 50msec @ a few hundred Amps

So im not making a spot welder technically, its really a micro TIG or pulse arc welder (thats why I have the contact sensing circuit). The actual welds could be anywhere from 300us to 40ms, I just had the duration turned all the way down right now to try and not break things. Hasn't helped though lol
Welding 18650s is actually the first application, A pulse arc style welder is needed because I'm using copper and/or aluminum to join the cells rather than nickel strips and their resistance is too low for resistance welding to work well.

 

Post a photo of your wiring and general setup.

Your gate driver should be located close and tight to the MOSFETs, as in keep the leads as short as possible. (like less than an inch!)
It matters when you are trying to drive them fast and hard. (read the application notes for the gate drivers)
The ground return path is equally critical. The inductance of even a few inches of wire will wreak havoc with switching speeds.

As designed, all your energy will be dissipated in the 0.05 ohm sense resistor, it's value should be a very small fraction of the overall weld circuit impedance.

Consider using a current transformer here.

Thank for the reply!
man, I had no idea the mosfet drivers should be so close to the mosfets... Not a EE at all so this is all new to me.
As-is the drivers are located in the control box and the mosfets are ~4ft of wire away. Its all buttoned up, i dont have a pic that would really be useful but its 4 ft of 22 gauge wire from the driver to the FETs and the actual weld cable is ~8ft of big ol' 2 gauge. I think that's unnecessary with the short weld durations, but I didn't know that when I built it...
That said, I did get this exact setup to work flawlessly with the same mosfets switching a 24v LED array rather than the capacitor bank and my oscilloscope was showing me exactly what i'd expected speed-wise. You think that would change dramatically with a higher amp load? My main problem now is the FETs keep failing on.

I dont quite understand what you mean by the .05 ohm resistor is dissipating all the energy, would you mind explaining that? the .05 resistor is in series with the weld electrodes because the resistance of the weld circuit was so low the current would have been thousands of amps. with the .05 resistor it should be <1000A.
thanks again!

 

The MOSFETs are probably dying of over-voltage spikes on the drain.

The inductance of the welding cables (albeit small) stores energy, when the current collapses, this inductive energy has to go somewhere.
Your circuit doesn't have a path for this current, the voltage on the drain rises until the MOSFETS break-over, which can destroy them.

You need a fast diode to clamp the inductive spike to the + supply rail. This diode may see high currents, needs to be rated appropriately.

Your wiring scheme makes matters more difficult, you have lot's of parasitic inductance that complicates things dramatically.
What you are doing is not the domain of casual hobby electronics: Fast switching of high currents / energy is tricky.

 

Gate drive ref material -

https://www.infineon.com/dgdl/Infin...N.pdf?fileId=5546d462617643590161d24582c33def

http://www.ti.com/lit/ml/slua618/slua618.pdf

https://toshiba.semicon-storage.com/info/docget.jsp?did=59460&prodName=TK5Q65W

Regards, Dana.

 

Hello,

file:///C:/Users/user/Downloads/TK5Q65W_application_note_en_20170821.pdf

@danadak , you can not link to your local disk. Please upload the file to the forum.

Bertus

 

I really like this forum. You all are super helpful, thanks!

Sounds like I have (at least) 2 major issues right now:

1.) I have lots of distance/ wire between my driver and mosfets and therefore will have very poor switching performance. I need to move the drivers to be as close as possible to the FETs. This wont cause the MOSFETs to fail like I'm seeing, but when I try to start adding a high frequency component to the weld it will cause issues. (thanks for the docs Dana, starting to read through those today)

2.) I have a long length of thick wire going to the weld head and the wire's inductance is likely causing a voltage spike during shut off and burning through the gates, locking them on. I need to add a diode between the mosfet Drain and the positive capacitor terminal. I've done some googling, it sounds like the diode will see the same current as the weld for a short period of time and the same voltage? This sounds a bit hand wavey but maybe find a diode with a voltage rating 2,3X higher than the capacitor voltage as a starting point?
The weld wire inductance is about 7000 uH (from a calculator on this site) so round up to 10,000 uH, can I hand calc the diode I'd need from the voltage, current, resistance and inductance, all of which I roughly know?
I added a diagram of the power circuit with diodes to make sure I'm thinking of this correctly

 

The MOSFETs are probably dying of over-voltage spikes on the drain.

The inductance of the welding cables (albeit small) stores energy, when the current collapses, this inductive energy has to go somewhere.
Your circuit doesn't have a path for this current, the voltage on the drain rises until the MOSFETS break-over, which can destroy them.

You need a fast diode to clamp the inductive spike to the + supply rail. This diode may see high currents, needs to be rated appropriately.

Your wiring scheme makes matters more difficult, you have lot's of parasitic inductance that complicates things dramatically.
What you are doing is not the domain of casual hobby electronics: Fast switching of high currents / energy is tricky.

yes, Im learning that this is pretty complex haha seems simple fundamentally but when every component eats up 100ns of a 2000ns budget it gets wild.
The control system needs to sense head contact and disconnection superrr quickly (like sense and react in a few microseconds) I had a hell of a time getting iptoisolators and a micro controller to be quick enough. But that's working now so moving on to the next issue! Its been very slow progress.

 

Don't they usually use an SCR instead of Mosfets in these welders? Since there is no need to shut down the transfer of current at a certain point/level. Just charge the caps, trigger the SCR on and it shuts its self down when the energy is transferred. The energy level is controlled by the number of caps being used for the weld, not the time it is on.

 

Don't they usually use an SCR instead of Mosfets in these welders? Since there is no need to shut down the transfer of current at a certain point/level. Just charge the caps, trigger the SCR on and it shuts its self down when the energy is transferred. The energy level is controlled by the number of caps being used for the weld, not the time it is on.

For cap discharge welders lots do, but this is technically a pulse arc welder (or micro Tig) The weld time needs to be precisely controlled and agitation of the weld is an important variable to weld quality ( agitation is done by switching another capacitor bank or the main cap bank at 1-20 kHz frequencies during the weld). I just called it a cap discharge welder because the power circuit it essentially the same and everyone knows what that is.

 

For cap discharge welders lots do, but this is technically a pulse arc welder (or micro Tig) The weld time needs to be precisely controlled and agitation of the weld is an important variable to weld quality ( agitation is done by switching another capacitor bank or the main cap bank at 1-20 kHz frequencies during the weld). I just called it a cap discharge welder because the power circuit it essentially the same and everyone knows what that is.

To my limited knowledge of micro Tig( but do know most other forms of welding including Tig, and have the facilities to do them), I thought micro Tig was just for micro welds, like in jewelry making and things of that size. Not for spot welding battery tabs or sheet metal. do you have links showing the process used for bigger welds like your talking about?

That said, there is another type of "SCR" device that is controlable, on and off. It's called a GTO "gate turn off" thyristor. https://en.wikipedia.org/wiki/Gate_turn-off_thyristor

 

To my limited knowledge of micro Tig( but do know most other forms of welding including Tig, and have the facilities to do them), I thought micro Tig was just for micro welds, like in jewelry making and things of that size. Not for spot welding battery tabs or sheet metal. do you have links showing the process used for bigger welds like your talking about?

That said, there is another type of "SCR" device that is controlable, on and off. It's called a GTO "gate turn off" thyristor. https://en.wikipedia.org/wiki/Gate_turn-off_thyristor

Unfortunately I don't know of any diy or commercial examples other than the 250i2 EV welder from sustone welders. That's where I got this idea, it's used to weld copper or aluminum to battery cells since the electrical resistance is too low for resistance welders to work properly.

 

That isn't what you think it is, it isn't the "universal" welder you have been talking about, it's not even a 'micro Tig".
It's a dedicated battery tab welder. You did look at the data sheet on it? Or the operators manual? It does it's job so well for a few reasons. 1. it has an air cylinder to control the force during the weld. 2. it uses argon as a shield during the weld. 3. did you see it's weld frequency?

 

That isn't what you think it is, it isn't the "universal" welder you have been talking about, it's not even a 'micro Tig".
It's a dedicated battery tab welder. You did look at the data sheet on it? Or the operators manual? It does it's job so well for a few reasons. 1. it has an air cylinder to control the force during the weld. 2. it uses argon as a shield during the weld. 3. did you see it's weld frequency?

Yes, what I've made is a 3 axis 1x1 meter cnc table, pneumatically lowered head with separate solenoid lifted tungsten electrode at the center of a brass weld head, solenoid valve controlling argon flow, optoisolated circuit that senses contact/ disconnect within 100ns and a microcontroller that is able to adjust every variable of the weld, including agitation style and frequency. Added a pic if you're curious.
Left all that out because it's distracting and mostly irrelevant to my problem, which is brutally murdering mosfets over and over lol
Everything is working exactly as I'd like except the mosfets firing the weld capacitors. Once I can get that fixed I can fine tune aspects of the weld to make a stronger bond (frequency/ lift distance/ power etc)
I have a few good papers on this style of weld and there's no magic sauce, just precisely controlled timing and a fairly complicated weld head.
That said, super open to suggestions for additional features, I'm learning as I go here

 

Why not place a zener diode across the mosfet? It will significantly clamp any reverse voltage spikes and will clamp any other voltage spikes to the zener voltage. Just make sure the zener voltage is above the normal operational voltage.