Paralleling MOSFETs

The neighbor and I were noodling (danger!). He is an instrumentation guru at one of the local hydrocarbon processing sites. He needs to weld thin type K thermocouples out of raw wire stock. A LOT of thermocouples. Quickly. We goofed around with a 100 amp 6 volt supply and got some fair spot welds pressing them in an X formation between copper electrodes, so now we are looking to produce a bunch and achieve better time/volts/current control.

My thought is to use a simple MOSFET driver circuit controlled by a P1. It will take several MOSFETs paralleled to handle the high currents involved.

Question: Do MOSFETs in parallel inherently current share when their gates are driven hard? Or do they act more like traditional transistors where you have to mess with individual emitter resistors to keep the currents balanced? I suppose I could just try it, but I already spend a lot of time dealing with that pesky purple smoke.

Comments

  • jmgjmg Posts: 14,417
    JRoark wrote: »
    Question: Do MOSFETs in parallel inherently current share when their gates are driven hard? Or do they act more like traditional transistors where you have to mess with individual emitter resistors to keep the currents balanced? I suppose I could just try it, but I already spend a lot of time dealing with that pesky purple smoke.
    MOSFETS have a positive temperature coefficient, so that means they do tend to thermally balance and are not prone to thermal runaway.
    Current sharing will still depend on the RDS variations, and on the parasitic wiring milli-ohms to each FET.
    On a spot-welder, you would also need to worry about possible avalanche energy so you may be better to design a sync converter buck regulator (ie a MOSFET in each possible current path)

    6V/100A is 60 mOhms, and you can buy MOSFETS well under 2m Ohms these days, Digikey has 829 hits < 1.2mOhms, so you may decide to ask 20-30A per MOSFET as a practical wiring operate point ?


  • Bingo. That positive thermal coefficient was the piece I was missing. Thanks @JMG.

    Yeah, I sort of think 20-something amps is about the realistic limit per MOSFET and that will be imposed by the physics of copper traces on FRB. BTW: I saw a really gauche (but effective) setup made by a guy making a solenoid “gun”. He dead-bugged a dozen MOSFETs onto two parallel copper rails and tied the gates together. It was ugly as sin, but the leads were short and apparently it worked well. I may have to explore this option just for fun. :)
  • Some thermocouple welders use capacitor discharge. Depending on the thermocouple type and gauge, some small fraction of a farad is charged up to the neighborhood of 100V, and then the joules let loose thru a carbon or other suitable electrode into the junction. MOSFETs can take quite a surge current for a short time.
  • That's a great idea Tracy. Makes the control so much simpler while still having lots of adjustment. The high-current discharge is a blind on-off that is simplicity itself. And that makes the isolation a breeze too because no feedback involved.

    The charging circuit can be a low current affair that charges to a precise voltage and doesn't get stressed over super precise timing or crazy spikes. Could even use a multi-pole relay here for isolation before discharge.

  • An example commercial instrument is this one:
    http://www.dcccorporation.com/hotspotplus.html

    As a trigger, an SCR might be preferable to the MOSFET. Dump the energy, auto reset. It is not hard to find an SCR that can handle the voltage and a surge current of well over 100A.

    Energy stored, E = C*V^2 / 2
    20000µF charged to 50 volts give 25 watt-seconds. (=joules).
    That is sufficient for smaller (<20 gauge) thermocouples.
    The high end hotspotplus claims to deliver maximum 525 watt-seconds for 12 gauge couples and general spot welding.
  • I like it, Tracy. And the output is self-limiting in the time domain. Once the cap is discharged, its done. No need for precision timing. Basically you can control the process on the front end by how much you charge the cap. Simplicity!

    Thanks for this!
  • frank freedmanfrank freedman Posts: 1,664
    edited 2020-07-30 - 02:44:00
    You could always charge a cap bank in a reasonably gentle manner and when charged turn off the FETs. Let the electrodes making contact enable the discharge and when done turn on FETs for next charge cycle. Lot easier on the electronics, lower power requirements and lower cost. Maybe a series resistor in line with the caps so the spot isn't vaporized.
  • I suspect it would be unreliable closing the welder electrodes with them hardwired to the charged capacitors. It would probably blow out a chunk of the thermocouple wire on the open electrode side and not even weld them. Or just vaporise right through, leaving nothing between the electrodes.

  • Maybe, maybe not. Check youtube for home built spot welders being made with car batteries and relays for spot welding strips to nicad and other battery types. Search on rebuilding battery packs for older power tools.
  • No transistors even for charging the capacitors? That'll be bloody rough on the relays!

  • Yeah, but c'mon, it's youtube after all......
  • Tracy AllenTracy Allen Posts: 6,504
    edited 2020-07-31 - 19:58:19
    evanh wrote: »
    I suspect it would be unreliable closing the welder electrodes with them hardwired to the charged capacitors. It would probably blow out a chunk of the thermocouple wire on the open electrode side and not even weld them. Or just vaporise right through, leaving nothing between the electrodes.

    I wonder about that too. If you play the video from the link above to the hotSpot welder, or look at the capacitive discharge welders from Omega Engineering TL-WELD-Series, it looks like the arc weld is indeed initiated when you touch the wires to the big carbon electrode while holding down an "enable" switch. The capacitor is directly connected to the two sides of the contact point and there is no FET or SCR or relay in series to initiate the discharge event. Both of them use an insulated long-nose plier connected to the power to hold the wires. (have to be careful with up to 80 volts?!) The Omega device holds the touch electrode in a well inside the instrument and allows for an optional argon purge, but the DCC device uses an external block of carbon or whatever. There is probably a transistor involved in the charging process while you wait for the green light to come on between events.
  • JRoarkJRoark Posts: 480
    edited 2020-07-31 - 19:57:45
    Yeah, but c'mon, it's youtube after all......
    Laughing. :)

    Just a quick update and a word of appreciation for all the feedback:

    We hit a bit of a wall yesterday using capacitive discharge to fuse the thermo's. It works a treat mechanically now, but the resulting thermocouples are not even close to spec outputs. They are all over the map with no discernable pattern except they all read low.

    There are two working theories for this at present:

    1). There is contamination either from the atmosphere or from the copper contacts of the weld jig, or,

    2). Something is happening to the wire during the weld process that causes there to be more than one point of generation.

    So for the moment we are forced by time constraints to fall-back to a quick buzz from the TIG torch in an inert atmosphere.

    I'm going to keep noodling at this because it intrigues me, and I'll follow up with findings when they come.

    Thanks to one and all for your insights!

    EDIT: Tracy posted while I was writing this. That bit about the optional purge gas sort of confirms this could be a consideration. Thanks, @Tracy Allen
  • what guage of wire are you using? Wonder what result you would have if the wires were the spot electrodes. Bring them together at the ends of the wire under argon with enough of a charge to spot them rather than vaporise and see if the results get more consistent. Thinking out loud for what it is worth.
  • You need to size the capacitor and the capacitor voltage to the wire gauge and bring the wire ends together in a quick firm motion. A solenoid or spring mechanism can can bring the wires together and the capacitance/voltage can be determined by trial and error. A tedious time consuming task, but once done the resulting welds are very consistent.
  • Here is a video found of all places, youtube.

    https://youtube.com/watch?v=-8cBCjJJcB4

    Note one if the comments regarding twists and the reply.
  • That's interesting, and some of the questions clear some things some things that were unsaid. I found it strange that he didn't identify the name of the spot welder, but it is evidently the hotspot I from DCC Corp. US$690. Their product pages have a link to their own informative promotional video. Note that they are touching the wire pair (grounded) to a carbon electrode (hot).

    I've tried it DIY with mixed success using old motor brushes or carbon poles pulled from old C-Zn D cells, hooked to a lab supply. It does take patience to find the optimal energy/wire mix.
Sign In or Register to comment.