Homemade spot welder keeps blowing MOSFETs

[Refrigerator]

But that's before the gate resistor. After the gate resistor the fall time alone increases up to almost 200ns.
So i'd say it's pretty much impossible to get rise and fall times on the gate any lower than 100ns since it's limited by the gate resistors. 

To go faster, you need lower value resistor, more current. The problem is that most likely it will start ringing. If you want to push high current into gate, you can, but you need very low parasitic inductances to prevent ringing. That means a good layout, very short traces. Also, not all mosfets are equal, you might be able to find one with lower gate capacitance. Still, this is not the whole story, because there is miller effect, which can prolong the turn-on. Also there is parasitic source inductance, so high drain-source current spikes cause the actual source potential to rise above ground and can cause slow down of turn-on. That is why you can find mosfets in 4 pin packages with kelvin source pin used only by gate driver, but not carrying drain-source current. All in all, high speed, high current mosfet circuits can be a pain to do good.

But in the end i question whether it's even worth putting in the extra effort to get super fast switching. 
A 1µs rise/fall time only increases the dissipated energy by less than 5mJ.
And a 10 Ohm gate resistor is already very low, in fact i don't have anything lower than that on hand, other than a couple 5W ceramic resistors of varying values.
I think i might decrease the resistance of the top side PNP transistor base resistor to maybe 470 Ohms to help bring the rise time below 1µs and will see how things go from there.

[Avelino Sampaio]

In this article the author provides a technique, without the use of the gate resistor, seeking maximum response speed.

http://prosje.be/Projects/BelastingSchakelen.html

[Refrigerator]

In this article the author provides a technique, without the use of the gate resistor, seeking maximum response speed.

http://prosje.be/Projects/BelastingSchakelen.html

The problem with my build is that i'm trying to drive 10 MOSFETs at once all in parallel.
A single MOSFET would be no problem.

[Refrigerator]

While i'm sat here, sifting through my biggest diodes, it kind of makes me wonder how do these ebay kits last that long when they don't even have any visible protection for the MOSFETs. 
Well for starters (pun intended?), nobody uses massive overkill batteries and huge wires like i do but even then there should be significant voltage spikes.


On a different note, i found three spare 80CPQ150 diodes that should work just fine i think.
80A continuous and 1930A peak - only for 5µs pulses, but in parallel they should share the load just fine.

[magic]

A freewheeling diode in parallel with the electrodes will dissipate inductive energy stored in the electrode cables.
The energy of power supply (and its cabling) inductance will still go into the FETs, I think.
Wouldn't an RCD snubber closely in parallel with each FET be the right solution?

[Refrigerator]

A freewheeling diode in parallel with the electrodes will dissipate inductive energy stored in the electrode cables.
The energy of power supply (and its cabling) inductance will still go into the FETs, I think.
Wouldn't an RCD snubber closely in parallel with each FET be the right solution?

I would need one massive snubber to handle a 1kA spike and not let it exceed 24V.
And placing several on each MOSFET wouldn't be any better than placing a large one on a single MOSFET just because all MOSFETs are on a single copper bar.

[Refrigerator]

Bought a 2m length of 6mm² wire and it's alot smaller than the 16mm² i used previously.
This should give me about 5mOhms of resistance.

[Refrigerator]

Assembled the welder again and added the diodes.
But while testing i noticed that the diodes do just about nothing. 
On the other hand my zener mod works perfectly. 
Currently i'm using a car headlight bulb on the high beam as my load.

[Refrigerator]

Previously i only had a zener on one MOSFET just for a quick test, now i've connected all gates to the zeners through a shottky diode on each MOSFET.
Replaced the bulb with a 330mOhm ceramic resistor for load and everything seems to work fine so far.
So i think i should be able to test the welder on some nickel now.

[Zero999]

I use a spot welder at work, which I've also repaired on a couple of occasions. It uses a DC:DC converter to charge a couple of large capacitors, which are discharged by a huge SCR, rather than a MOSFET. The amount of energy going into the weld is controlled, by changing the output voltage. The SCR is rated to 300A, at 400V, if I remember rightly and don't think it has any overvoltage clamping.

[Refrigerator]

I use a spot welder at work, which I've also repaired on a couple of occasions. It uses a DC:DC converter to charge a couple of large capacitors, which are discharged by a huge SCR, rather than a MOSFET. The amount of energy going into the weld is controlled, by changing the output voltage. The SCR is rated to 300A, at 400V, if I remember rightly and don't think it has any overvoltage clamping.

SCR's don't need any protection because they only turn off when there's no longer current flowing.
The biggest problem with CD welders is finding the right caps.
Edit: and you need to wait for them to recharge with each weld.
Chunky SCR's are easy enough to find, i have a couple big ones in my parts bin, no idea what their specs are.
Made in USSR in '87, model number PM50 but internet doesn't seem to show any specs for it. 

[voltsandjolts]

Excuse me for slight off-topic, but just want to point out these capacitor
discharge welders that pop up on flea-bay every so often. This one is a
1998 vintage and its a well built work horse that makes perfect battery
tab welds every time. The capacitor bank is 4 x 1500uF 250VDC which
dumps into a very nice custom output transformer. The output winding
is 5 turns of 10mm x 5mm solid busbar There are also various primary
taps, selected by the large front panel 'pulse width' knob.

I think Hughes were the original manufacturer (guessing from the 'HCD'
name) but there are HCD125's from Hughes, Micro-Join, Palomar. I think
Amada-Miyachi may have the product line now and new equivalent
welders are in range of 4K to 10K GBP....gulp [1].

The double-pulse output works very well, cleaning up the workpiece with
the first hit, then full weld with the second. Both pulses have controlled /
programmable energy.

I need to do a rough schematic of this unit some day.

[1] https://www.amadaweldtech.eu/products/resistance-welding/capacitive-discharge-welding-power-supplies

[Refrigerator]

Excuse me for slight off-topic, but just want to point out these capacitor
discharge welders that pop up on flea-bay every so often. This one is a
1998 vintage and its a well built work horse that makes perfect battery
tab welds every time. The capacitor bank is 4 x 1500uF 250VDC which
dumps into a very nice custom output transformer. The output winding
is 5 turns of 10mm x 5mm solid busbar There are also various primary
taps, selected by the large front panel 'pulse width' knob.

I think Hughes were the original manufacturer (guessing from the 'HCD'
name) but there are HCD125's from Hughes, Micro-Join, Palomar. I think
Amada-Miyachi may have the product line now and new equivalent
welders are in range of 4K to 10K GBP....gulp [1].

The double-pulse output works very well, cleaning up the workpiece with
the first hit, then full weld with the second. Both pulses have controlled /
programmable energy.

I need to do a rough schematic of this unit some day.

[1] https://www.amadaweldtech.eu/products/resistance-welding/capacitive-discharge-welding-power-supplies

That's pretty cool. I was also wondering if it's possible to dump a capacitor bank into a transformer because i have a bucket full of SMPS primary filtering capacitors.
How are the primary taps selected? I assume there's an SCR going to each tap with a single common tap so that you can choose which SCR to fire to dump the capacitor bank into whichever tap.

[Refrigerator]

Received the nickel strips i had ordered so i could try my welder with fresh nickel instead of the stuff i pulled off the battery pack, which i took apart.
At first i couldn't make any kind of decent weld no matter how long i set the pulse. Max pulse on my controller is 100ms.
The nickel strip was visibly melted between the two spots but it just wouldn't stick to the battery.   
I even tried refiring it multiple times on the same spot with a 100ms pulse and that made no difference, other than making the 6mm² wire quite warm.
So i started wondering whether i had actually introduced too much resistance into my loop and even thought about trimming my wire.
Good thing i didn't because i found out why my welds weren't sticking. As i turns out i was applying too much pressure on my electrodes, go figure 
If i apply only gentle pressure on my electrodes i get perfect welds. If i pull the nickel strip off i get very nice teardrop shaped nuggets left over on the battery.
I read on a blog some years ago when researching battery tab spot welders, that a teardrop shaped nickel nugget left behind is the optimal weld. So i'm pretty stoked 
I guess it's time to finish up my battery analyzer so that i can re-test and cycle my batteries, so that i know for sure that their capacity is accurate.

Pics:
1 - Test using salvaged nickel. You can see how it melted but i pulled it right off with minimal effort. The strip was cleaned with acetone beforehand. Looks like a smiley face now that i've had a closer look at it. 
2, 3 and 4 - After i realized that less pressure makes for better welds you can see how much better those have turned out. Second pic with 0.2x8mm nickel strip, last two with 0.15x12mm strip.

[parasole]

I would recommend malectrics, bought his PCB time ago (I think it was V3.1) and added 1F capacitor bank which solves the power supply problem. I use 12V PSU and charge the bank via car light bulb as the current limiter, charging time is about 5-7 seconds which is just ok time interval you need to move for next welding position.
Below is the picture of my setup, since that time I supplemented it with oled display, but that is more as a service function, you really need only a pulse generator for proper work.
Now I may see there is  PCB V4 with added capabilities, but again, most important is power electronic arrangement which is done in a proper way so you won't blow up your MOSFETs.

https://malectrics.eu/product/diy-arduino-battery-spot-welder-pcb-set-v4/
https://github.com/KaeptnBalu/Arduino_Spot_Welder_V4

[Refrigerator]

I would recommend malectrics, bought his PCB time ago (I think it was V3.1) and added 1F capacitor bank which solves the power supply problem. I use 12V PSU and charge the bank via car light bulb as the current limiter, charging time is about 5-7 seconds which is just ok time interval you need to move for next welding position.
Below is the picture of my setup, since that time I supplemented it with oled display, but that is more as a service function, you really need only a pulse generator for proper work.
Now I may see there is  PCB V4 with added capabilities, but again, most important is power electronic arrangement which is done in a proper way so you won't blow up your MOSFETs.

https://malectrics.eu/product/diy-arduino-battery-spot-welder-pcb-set-v4/
https://github.com/KaeptnBalu/Arduino_Spot_Welder_V4

5-7s between welds seems like a pretty long time to wait. With my welder i can fire as fast as i can get my electrodes on.
That arduino welder sure looks cleaner than what i have 

[Refrigerator]

Made myself a better foot switch yesterday and played around with the welder some more.
Seems like i've got the hang of it and now my welds are super strong.
Also it's fun welding nickel strips to anything metal. I even welded a strip to a steel bracket i had laying around.
I didn't have any stiff switch for the foot switch so my pedal has a lever mechanism that presses on the stiffest microswitch i could find at home.
It feels ok but i think there's still room for improvement. It's certainly better than what i had before that's for sure

Pics:
1 - old foot switch, was made in like 5 minutes.
2 - new and improved foot switch.
3 - new foot switch opened up, it kind of flops open like that 
4 - tried to pull of a nickel strip i welded and it tore right off, this was with a 50ms pulse.
5 - leftover nickel nuggets can be seen on the metal bracket after i pulled the strip off.

[parasole]

5-7s between welds seems like a pretty long time to wait. With my welder i can fire as fast as i can get my electrodes on.

well, that is my personal setup which depends on PSU current capability (and it is massive as being server PSU) and ballast resistor (bulb), and I easily may shorten that time by simply replacing the bulb with more powerful (one or just a few in parallel, but I found that time to be just ok for me while building my bike battery. 

[Mechatrommer]

1 - old foot switch, was made in like 5 minutes....

thanks for letting me know i'm not the worst builder here... btw, congratulation on your 6mm.sq wire purchase

[Refrigerator]

1 - old foot switch, was made in like 5 minutes....

thanks for letting me know i'm not the worst builder here... btw, congratulation on your 6mm.sq wire purchase

Yeah now i'll know not to go overboard with copper thickness. 
At least it finally works now! 
The new footswitch is on it's third revision already and i think i have it dialed in pretty good.
Can't wait to get my battery analyzer up and running to test all the 18650 cells i have and assemble a nice big lithium pack. 

[vishaldotgupta]

hi, i might be late on this forum.  Me too is working on home made welder using big lead acid battery as current source.  my design as follows

1. i will be using a miniature relay instead of gate driver.  The timing pulse is from a low cost programmable timer from aliexpress
2. Gate drive voltage will be sourced from 12V, but there will be diode and a capacitor.  during the welding process the 12V will drop and may drive mosfet in linear region thus blowing them.
3.  just after welding pulse the relay will connect gate to ground thus discharging the gate capacitance asap (through a 10ohm resistance).
5. i intend to use small filament bulb as snubber by placing it across mosfet drain and source.  it has a very low cold resistance and should absorb the spike
6.  i also intend to use a led across welding probe.  The advantage is, that in case a mosfet is shorted (from previous weld attempt), the led will glow up.  so you can prevent an accident before actually welding is attempted with shorted mosfet.
7.  i also suggest adding MCB / delay type fuse in the battery circuit as  a protection mode.

i would like to have comments on this thought

[DavidAlfa]

350A MOSFET, but crappy driver, you need a very fast driver or it'll just blow like that.
Also a 3KW TVS would be help with inductive spikes.

[Zero999]

The spot welder I use at work uses a large capacitor bank and a boost converter to charge it. The capacitor is discharged via a huge SCR, not a MOSFET. A complex arrangement involving floating the ground of the controller is used to trigger the SCR, which I think is to enable the battery 0V to be on the same ground as the clamp on the workpiece. It would have been much easier to have used a pulse transformer to fire the SCR.

[shapirus]

Made in USSR in '87, model number PM50 but internet doesn't seem to show any specs for it. 

Better late than never!

It's "ТЧ50", quite searchable. High frequency thyristor, up to 25KHz, max current is 50A. Max 700V repeated reverse impulse voltage, and the rest of the marking isn't visible.

[Refrigerator]

The spot welder I use at work uses a large capacitor bank and a boost converter to charge it. The capacitor is discharged via a huge SCR, not a MOSFET. A complex arrangement involving floating the ground of the controller is used to trigger the SCR, which I think is to enable the battery 0V to be on the same ground as the clamp on the workpiece. It would have been much easier to have used a pulse transformer to fire the SCR.

SCR's are easy to drive, because as long as you can provide a suitable trigger, the SCR will self sustain on a closed state. I agree, a transformer could do that quite easily.

hi, i might be late on this forum.  Me too is working on home made welder using big lead acid battery as current source.  my design as follows

1. i will be using a miniature relay instead of gate driver.  The timing pulse is from a low cost programmable timer from aliexpress
2. Gate drive voltage will be sourced from 12V, but there will be diode and a capacitor.  during the welding process the 12V will drop and may drive mosfet in linear region thus blowing them.
3.  just after welding pulse the relay will connect gate to ground thus discharging the gate capacitance asap (through a 10ohm resistance).
5. i intend to use small filament bulb as snubber by placing it across mosfet drain and source.  it has a very low cold resistance and should absorb the spike
6.  i also intend to use a led across welding probe.  The advantage is, that in case a mosfet is shorted (from previous weld attempt), the led will glow up.  so you can prevent an accident before actually welding is attempted with shorted mosfet.
7.  i also suggest adding MCB / delay type fuse in the battery circuit as  a protection mode.

i would like to have comments on this thought

1> Won't work due to contact bounce. After i popped my controller (by connecting the power backwards  ) i just used an L9110 motor driver IC as a gate driver and it honestly works amazingly well. The rise and fall times are much sharper than with any other gate driver i had on hand and on par with my high current discrete solution.

2> Fine, but relays will draw a lot of current so your storage caps will need to be large. Assuming you go for relays regardless of above point.

3> Your relay will be too slow for that, the inductive spike will take nanoseconds to knock a hole in your FET's

4> 404 point not found

5> Well i mean some 600A is a walk in the park for a bulb to handle right? On a more serious note, it won't work. You need a very stiff snubber to not allow the voltage to rise above what your FET's can handle. My solution was to use the FET's themselves as snubbers through a diode arrangement, where the inductive spike would power the gate just enough for the FET's to keep it below V_DSmax. It's a simple and cheap solution that i've found works well and has kept my FET's from blowing ever since i got my welder working.

6> Sure, that sounds ok. Personally i find the sparks and burned holes give enough of an indication that my FET's are gone.

7> The fuse will limit your max current by alot. Also most things become fuses at spot welding currents, including the thing you're welding.