No announcement yet.

Circuit Breaker Pops Startup GTSW250

  • Filter
  • Time
  • Show
Clear All
new posts

  • Circuit Breaker Pops Startup GTSW250

    Long post below and semi related to Miller welders... I am trying to troubleshoot why my circuit breaker keeps popping at startup on my Thermal Arc 250 GTSW after I put in a new bus capacitor, which has a similar input circuit to the XMT304. I am curious if it is a breaker sensitivity issue (50A Eaton BR breaker) or it it's really the difference between a newer and older capacitor model.

    I have worked on quite a few different generations of inverter welders, and noticed the evolution of the capacitor inrush current limiting by timing when to use big resistors to charge the bus capacitors.

    The early (pretty much all but the 300SE) POWCONs and the 80's Hobart/Panasonic welders (probably many others I have never seen) used charge "resistors" (the Portatig used a spiral of what looks like bailing wire) that were always in series between the bridge rectifier and the buss capacitors.

    Some other inverters in the 80s and early 90s, and even the smaller Miller units (Dynasty 200, early Spectrum 625) used a mechanical relay in parallel with the charge resistors that switched on some time after the buss capacitors had a chance to charge. These could use smaller resistors, but I am guessing had the cost and reliability issues of the mechanical contactor. If I remember right, my Lincoln Powerwave 355 (and V350 to that point) still used this same design.

    Evolution from there was to use what the XMT304 and a ton of the late 90/early 2000s industrial machines used, which was the Sanrex bridge rectifier with "solid state in-rush" - basically a thyristor in series between the bridge rectifier output and the capacitor. On the Millers and a lot of the later Thermal Arc (400GMS, 300 GTSW/TSW) these thyristors we're in parallel with a charge resistor. At least one Thermal Arc didn't have a charge resistor (will come back to that). The problem with this design from my perspective was if the capacitors failed (or possibly IGBTs), they would ultimately take out the bridge rectifier and power switch.

    The next evolution I saw was with the XMT350 - they basically use a solid state relay (SCR on each leg) so to prevent overcurrent cycle by cycle. They still used the charge resistor and can often "save the input components if the buss capacitors failed.

    So this gets me back to why I posted... I just replaced the bus capacitor in my Thermal Arc 250 GTSW, and it pops my 50A breaker on startup. I couldn't find the exact replacement for the original Hitachi capacitor, so I went with one that had the same capacitance (3300uF), voltage rating (400VDC) and case size. From the original (Hitachi HCG F5A no longer available) datasheet, I am seeing a max ESR of 0.08 ohms, and the TDK B43607 datasheet (replacement), I am seeing a .02 ohm ESR. I can switch back and forth between the caps and the results are the same - old cap no circuit breaker pops (this was true on a 30A Square D breaker too) - new cap the breaker pops. I inadvertently discovered if I turn the machine on with the new caps while they were partially (dis)charged, the circuit breaker didn't pop. I want to come up with a solution to use a new cap, because the original 20+ year old cap is on borrowed time.

    First Question - Is there something with the Eaton BR breakers that would make them more sensitive to current in-rush? I wouldn't have imagined the ESR difference would make that big of a difference on a 120Hz rectified input. At 120Hz, that's only about a 1% difference in impedance, so 1% difference in input charge current. I might be off in my assumption here, and should treat the input as DC given how fast it charges. That would make it 4x the peak inrush current.

    Second Question, was this a common issue with the Thermal Arc 300GMS and 250 GTSW. It seems like the capacitors would have a shorter life subjected to that inrush current everytime they were switched on.

    Third Question - Am I better off changing the input circuit to something like the XMT304 or later Thermal Arc inverters? This welder has an input circuit just like the XMT 304, except doesn't have a charge resistor (like what you happen if R7 on the XMT 304 PC2 would fail open). From the service manual, in 1994 Thermal Arc replaced a mechanical contactor (generation 2 above) with the solid state in-rush circuit. Oddly, it kept the 30W 200ohm charge resistor, but now uses it to charge a 10nF ceramic capacitor across the SCR gate. This has a relay in series that charges the gate ~5 seconds after turning on the machine. I am not one to question the electrical engineers who designed this circuit, but they clearly changed it in the next series of welders (and copied or had their design copied by Miller). I am thinking to use something similar to the SCR gate charge circuit used in XMT304 and repurpose that large 30W resistor to do what it originally did in the mechanical contactor circuits.
    Last edited by jjohn76; 07-15-2022, 07:21 PM.

  • #2
    Here's the Thermal Arc "solid state in-rush circuit." C4 is the buss capacitor. R1 is a massive 30W , 200 ohm resistor charges capacitor C26 to turn on the SCR. Note CR1, which is a relay that closes ~5 seconds after switching the main switch on the welder. There is no voltage across C4 until that relay closes. It takes very little time to charge C26, which does very little to charge the C4 bus capacitor. One C26 is charged enough to turn on the SCR in the module, the C4 is basically connected directly to the output of the bridge rectifier and pulls a lot of current.

    Click image for larger version  Name:	16579153156706838683127424863126.jpg Views:	8 Size:	3.16 MB ID:	620941
    Last edited by jjohn76; 07-15-2022, 07:25 PM.


    • #3
      Here's what the "solid state inrush circuit" replaced. The power resistor was in parallel with the main contactor (MC1), and both were actuated separately. In this design, it appears CR1 actuates first, allowing C4 to charge through the charge resistor, then MC1 actuates, connecting the bus capacitor straight to the rectified input.
      Click image for larger version

Name:	16579162190611337940999130276246.jpg
Views:	67
Size:	3.54 MB
ID:	620943


      • #4
        So, investigating the issue further, mid-day in New Mexico, both capacitors pop the breaker at startup. I tried two different 50A circuits in the same panel, same issue. I am not really seeing how the "solid state inrush" circuit works up above to pre charge the capacitor at a slower rate.

        Edit: I'll draw the XMT304 and TA250GTSW input circuits side by side soon, but just confirmed the XMT 304 uses the same resistor (R7) to both charge the bus capacitor, then positive bias the SCR. The PC1 control relay has it connected normally closed (NC) to the DC bus and normally open (NO) to the SCR gate. When the control relay actuates at ~5 seconds post startup, it disconnects the power resistor from the DC bus and connects it to the thyristor gate. The GTSW doesn't have the NC connection and the order of relay/power resistor are different. There are some other differences in the SCR snubbers (probably has something to do with the input inductor L1 in the XMT), though I'll worry about that later. I think just finding the NC connection (will look at pin 10 on the relay) and routing that to the DC bus gets me to better inrush current limiting. I am pretty certain this is worth it, given how much inrush it takes to pop a 50 amp breaker (albeit at 90 F ambient in the panel) - this is a huge stress on the capacitor too.
        Last edited by jjohn76; 07-16-2022, 02:05 PM.


        • #5
          Hopefully this drawing makes more sense. The GTSW 250 and XMT are very similar, except the GTSW 250 has a filter cap across the rectified DC, the XMT 304 has a RC snubber across the pre-charge/in-rush SCR, and the XMT has a bleed/pulldown resistor on the SCR gate. I showed the inductor on the XMat input because I think it is the reason for the snubber across the SCR (guessing it helps with noise and keeps the bus capacitors happy).

          Click image for larger version

Name:	16580034445823805747649878369419.jpg
Views:	42
Size:	2.33 MB
ID:	620947

          With this, I am going to look for the NC connection on the relay and connect it straight to the bus capacitor+ terminal. Then I will flip the resistor/relay connections on the input bridge module, so it goes Rectified+ -> Charge Resistor -> Relay terminals -> module and add lead to capacitor terminal.


          • #6
            My guess is its busticated.


            • #7
              Yeah, I thought it was my breaker. Checking the timing and voltages though, that input circuit is working like it's designed - No current flows into the the C4 capacitor until the relay CR1 closes. This machine calls for an 80 amp breaker, probably for this reason. I am not putting in an 80 amp circuit, especially since if it has that much inrush, the cap won't be happy. I am chalking this one up as a bad design that lasted only a few years.


              • #8
                A lot of press is given to inrush but I never had it happen even with a couple 300s on 60.


                • #9
                  Originally posted by Sberry View Post
                  A lot of press is given to inrush but I never had it happen even with a couple 300s on 60.
                  Cary, which 300s are you running?


                  • #10
                    I have a 300 nsynch, was using a 330 at another shop and a bud has an Airco. I think the BP and Airco may be wired on correct breaker but I had my synch on 60 a long time and only ever tripped it up continious +200 or so on air cooled tig. I never had it trip on a start, never had a machine do it. Only time I recall ever tripping up a welder other than the one I mention was nickle on a buzzer. Been a long time though and dont recall the circumstance except that it seemed funny it tripped up at 90 A or so. I had welded from that circuit bigger rods hours on end.


                    • #11
                      Mine has made a great cart for plasma. Sort of the current layout, another feeder has a home at the bench. Click image for larger version

Name:	Welder group new.jpg
Views:	32
Size:	505.3 KB
ID:	620959


                      • #12
                        Yeah, I was actually limited to a 30A breaker for a long time and could get everything to run without a breaker trip. The older mains surged at Arc start more than startup unless they had PFC caps - I tended to half or third the PFC capacitance on those machines to get the most out of a 30 amp.

                        I was digging more into the 250 GTSW. There is something probably wrong with the timing on mine. In the circuit above, leads 103 and 8 are tied together with an onboard relay (CR3) at startup, then open sometime later. So the startup sequence is
                        1. ​​​​Switch Machine On 2. Control transformer charges control board circuits
                        3. Control board actuates CR1 - control transformer primary windings are put in series, fan turns on, the pre-charge resistor connects to the SCR gate, which is connected/shorted to the DC+ bus through CR3 on the control board. This prevents the SCR from turning on, while also allowing the bus caps to charge through the pre-charge resistor.
                        4. CR3 on the control board actuates, disconnecting the SCR gate from the DC+ bus, letting the SCR conduct when it is forward biased.

                        When I measure the voltage across the capacitors, step 3 and step 4 are happening very closely. The bus capacitors pre-charge to ~30V then jump to 325V and the breaker pops. There is probably a timing circuit on the board that turns on CR3 after the control transformer doubles the control board voltage. I will look for some other time. In the meantime, I added a lead from the NC terminal on CR1 to the DC+, and swapped around the order of the CR1 terminals and R1 pre-charge resistor. So far, no breaker pop. It charges the bus cap to ~280V on startup, then CR1 turns on, and it continues to charge to ~300V, when the control board CR3 clicks open and the cap jumps to ~325V.

                        The only issue I see with this temporary fix is the bus has some positive charge before the control board is fully powered. I am less worried because that's how the XMT and later Thermal Arcs work, and that's how the previous GTSW 250s worked. Plus, the bus is current limited and the inverter IGBT gates have pull down resistors. We shall see...
                        Click image for larger version  Name:	IMG_20220717_160754.jpg Views:	1 Size:	1.67 MB ID:	620963
                        The red wire comes from the CR1 NC terminal to the +terminal on the bus cap.
                        Click image for larger version  Name:	IMG_20220717_160719.jpg Views:	1 Size:	2.67 MB ID:	620964
                        Nit too visible, but I swapped the yellow and white wires (coming in from left to the middle and bottom terminals) on the module. So far so good.