Help with sensing Electrode contact on high amp welder

Discussion in 'Analog & Mixed-Signal Design' started by Cfez202, Apr 25, 2018.

  1. Cfez202

    Thread Starter Member

    Apr 25, 2018
    37
    3
    Hi all,
    Ill try to make a long story short...

    I'm making a welder called a pulse arc welder. Nothing very fancy about the power circuit of it, basically a 48v power supply, a large bank of capacitors in parallel equaling about 1/4 farad, and a bunch of mosfets in parallel acting as the switch.

    The mechanical side consists of a brass collar which grounds to the two sheets to be welded and a tungsten electrode in the center which is lifted quickly during the weld via a solenoid.
    Once the collar and electrode both make contact with the surface, an arduino begins a bunch of closely timed steps to lift the electrode and begin the weld.

    I need help with a circuit that will detect when the electrode and brass collar both make and break contact with the metal to be welded and generates a ~5v "contact" signal the arduino will be able to read without frying everything once the hundreds of amps from the weld are dumped into the same wires.

    Any help would be greatly appreciated, I'm no EE, just beginning to learn all this stuff.
    Thanks a lot!!
    -Charlie

    In the attachment:
    Orange is the brass grounding head
    Grey is the tungsten electrode
    Red box is what id like to figure out.
    P1 comes from a mosfet driver circuit
    I do have some flyback diodes across the powersupply just didnt show them
     
  2. jpanhalt

    Expert

    Jan 18, 2008
    7,674
    1,884
    Welcome to AAC.

    Here is your mosfet drawn with its intrinsic body diode:

    upload_2018-4-25_1-39-20.png

    Your Source is positive and the Drain is negative. So, current will flow as soon as the brass contactor and tungsten electrode make contact. Can you just sense that current? Maybe a photodiode will detect the spark?

    Also, an LED with appropriate current limiting resistor will be lit if attached across your electrodes. Maybe use an optoisolator as your sensor.
     
  3. Cfez202

    Thread Starter Member

    Apr 25, 2018
    37
    3
    Thanks for the response!
    yeah I MS painted the power supply incorrectly... sorry, the source should be negative. I definitely don't want it to fire as soon as the electrodes make contact...
    I did have small scale of this operating w/ only 12v, I just couldn't figure out what to stick in the red box.

    For context, the full order of operations for the weld should be:
    - Weld initiated by operator (Only user input, will be CNC controlled eventually)
    - Argon valve opened
    - Brass head and electrode pneumatically lowered to part
    - Head and electrode contact part
    - XX millisecond rest (50-200ms)
    - Electrode is lifted from part via solenoid (brass head maintains contact)
    - Electrode separation detected
    - XX micro second pause (10-100us)
    - Capacitor MOSFETs fire for XX micro seconds (50-200us)
    - Brass head pneumatically lifted from part, electrode solenoid released
    - XX millisecond Delay, Argon off
     
  4. LesJones

    Well-Known Member

    Jan 8, 2017
    2,304
    693
    I don't understand how this can weld if the mosfets switch on when the electrode is NOT in contact with the metal. Have you missed out a step near the end of the sequence of events ? If you add a resistor in parallel with the mosfets there will be a voltage between the electrodes when they are NOT in contact with the metal which could be detected. If you added an opto isolator in series with the resistor it's output would give an isolated sicnal into the controller. You would probably need to add some protection on the input of the opto isolator to deal with any ringing due to the inductance of the cable from the capacitors to the welding head.

    Les.
     
  5. jpanhalt

    Expert

    Jan 18, 2008
    7,674
    1,884
  6. Cfez202

    Thread Starter Member

    Apr 25, 2018
    37
    3
    Ah resistor in parallel isnt a bad idea... what sort of protection would you put on the optoisolator?
    As far as the weld sequence goes;
    Goal is basically to fire the mosfets as late as possible. Once they fire, the inert environment connects the tungsten and weld piece with a plasma stream just like TIG welding. Too early and the tungsten tip melts into the weld piece, too late and it wont light. The tungsten, its holder and the actuator from the solenoid have a fair amount of inertia so it takes a bit of time (micro second scale) before they actually separate from the weld material. Im imagining the the fire sequence happens like the attached picture (made up time scale)
     
  7. Cfez202

    Thread Starter Member

    Apr 25, 2018
    37
    3
    To be honest, I'm a bit uninformed here. With conventional TIG welding the electrode is negative which causes the majority of the heat to go through the work part (not exactly sure why). But this is going to be welding with an extremely small duty cycle and the tungsten/holder has so much thermal mass behind it I'm not too worried about it overheating... it would be nice to cut back on the capacitors needed thoguh... What could I do to swap polarity and prevent the mosfets from firing on contact? Just a different Mosfet?

    I'm specifically designing this to weld thin copper sheets to battery cells (nickle plated steel sheet), which is why I cant just use a CD welder. Conductivity of copper is way too high. This is essentially an automated TIG welder and uses plasma to do the actual material melting.

    Also thanks for the suggestion, I did see it. How would I hook it up to the same circuit that hundreds of amps and 48v will flow through without turning the isolator into magic smoke? Wouldn't it need a voltage source independent of the capacitors to sense contact? I kind of like Les' idea, resistor in parallel with an optoisolator across the mosfets. Just wondering what prevents it from melting. Once a resistor is large enough to protect the isolator from hundreds of amps, wouldn't the voltage across it when not welding be undetectable by the optoisolator?
     
  8. LesJones

    Well-Known Member

    Jan 8, 2017
    2,304
    693
    The resistor would be in series with the opto isolator. I would choose about a 2.7K resistor which would give about 17 mA through the opto coupler. I think just a diode in parallel with the opto coupler should protect against any reverse polarity spikes. The output side of the opto coupler could connect to the arduino input together with a pull up or pull down resistor. I think it very unlikely that the gap would ionize with only 48 volts across it. A tig welder starts the ionization with a high voltage high frequency pulse on top of the normal welding voltage. Protecting the opto isolator from this sort of pulse would be more of a problem.

    Les.
     
  9. Cfez202

    Thread Starter Member

    Apr 25, 2018
    37
    3
    Gotcha, I'll give it a try! Thanks for the help!

    The goal is it make this a lift start rather than high frequency start. The electrode has an adjustable ~0.5-5 lbs of force behind it so I'm hoping I can "sense" disconnect before it actually disconnects (as the electrode's contact area drops and resistance increases.) Then fire the mosfets while there is still some contact. If that doesn't work out I'll just have to do some trial and error to time the mosfets correctly with the lift. It should be pretty repeatable once I dial it in... Head is pneumatically lowered with a preset force and the the electrode is actuated up via a linear solenoid.
     
  10. shortbus

    Expert

    Sep 30, 2009
    6,875
    4,047
    Have you seen how this guy was doing this?
    http://pulsearcwelder.blogspot.com/2012/08/page-1-pulse-arc-welder.html
    http://pulsearcwelder.blogspot.com/2012/09/page-2-first-weld.html
    http://pulsearcwelder.blogspot.com/2012/11/page3-insides-of-abi-pulse-arc-welder.html

    Personally I'd attack it differently.
    1. I'd preset the solenoid out with a small voltage on the electrode. The solenoid and electrode would also be spring loaded inside the gas cup.
    2. When the electrode was pointed where the weld is wanted and touching the work, the welding head would be pushed against that small spring loading of the gas cup. The gas cup would then cause the small voltage circuit to close.
    3.This would turn on the shield gas, and after a small time out the main voltage would fire and the solenoid would release. The spring on the solenoid would retract the electrode causing the weld.

    The reason for having the solenoid out against a spring, is that the spring will move the electrode faster than the turning on of the solenoid.
     
  11. Cfez202

    Thread Starter Member

    Apr 25, 2018
    37
    3
    Hi,
    I actually did see this, I couldn't figure out how hes sensing the weld head contact.
    As far as the solenoid being used to keep the electrode down or lift it up, it was just an ease of assembly/ packaging and adjustability choice. This way I can actively adjust how fast the electrode lifts which wouldn't practically be possible with a spring. Unsure if that's even going to be necessary though
     
  12. shortbus

    Expert

    Sep 30, 2009
    6,875
    4,047
    I only read it fast not studying it at all, but assumed he was doing the discharge manually, only when everything was aligned. Not in a CNC machine. But a 5V sensing circuit should work, protected by a diode from the welding voltage when it gets applied.

    As a Tig welder I know that you can't have too much distance between the electrode and part so the electrode would only need to move very small distance say 0.03 to 0.06 inch during the weld. Don't most of the jewelry welders use a foot switch to energize them? And the have a positioner to hold both the work and the welding head.

    For welding battery tabs what is wrong with a spot welder? Seems to me that the tabs also need clamping force to get good welds, which this type you're wanting won't give you.
     
  13. Cfez202

    Thread Starter Member

    Apr 25, 2018
    37
    3
    Parts for the 5v opto/ diode circuit are ordered, Ill try my luck and see how it goes!

    The electrode movement is set by a stop screw, so I can set it to retract 0 to ~0.5", but like you said, I'm planning on only needing to retract 1/32-1/16. Tabs do require a fair amount of force to weld to batteries. The brass weld head is pneumatically actuated and can apply anywhere from 1-50 lbs through the brass grounding head (~5-10 lbs should be plenty for most welds).
    I added another pic to maybe make it clearer.
    The brass cap and stainless main shaft (which house the blue electrode shaft) are pneumatically pressed to the weld surface, then the blue central shaft which is what the electrode threads into is separately actuated via a solenoid. The green dashed lines represent sliding surfaces. I have the mechanical assembly all machined and set up, just working on the electronics and power assembly now.

    Main reason I cant use a capacitive discharge or normal resistance welder is the high conductivity of the copper sheets I'll be welding. Most semi professional or DIY battery packs have cells connected with ~.005" nickel sheets. Besides nickel being relatively expensive and difficult to source sheets of, it also has much higher resistance than copper. This makes it easy to resistance weld, but in a high amperage pack you need a much greater thickness of nickel to handle the same amperage as copper. If I were to make a CD welder with enough power to weld copper, the weld electrode would just weld itself to the copper every time. With this pulse-arc/ TIG approach the electrode is separated by the time the material melts.
     
Loading...