For a given output voltage & current, how much space is saved by using 220V instead of 120V input?

Specifically, I want to know is it expected for 220V only welders to be much smaller than equivalent ones that have to work with both 120V & 220V. Here's a comparison. It's not exactly fair because the 220V machine has less cooling and only allows 25% duty cycle. Another comparison I could've done is compare a 220V only computer power supply vs a 120V one, but it appears there are no 220V only PSUs.

  1. 220V: Bossweld 200 Amp TS200 (25% duty cycle @ 200A)

    • 145mm x 280mm x 410mm = 16.6L

    • 7kg


  1. 120V: Weldpro ACDC 200GD AC/DC 200 Amp, 40% duty cycle @ 200A

    • 447mm x 201mm x 406mm = 36.5L

    • < 27kg


So the 120V machine is 2.2x bigger.

Assuming a welding power supply uses the same design as a computer power supply like described here and here.

Then it's obvious a 220V supply would only use 0.55 the current in all stages before and up to the primary coil of the transformer. Everything after that would be the same for both 120V and 220V.

So is it fair to conclude that the high voltage, non isolated side will be 1.8x bigger in a 120V design? What fraction is taken up by the other isolated, low voltage side?

I'm also assuming having to support variable input voltage shouldn't cost anything extra, because you don't need any extra hardware, just a different duty cycle.

  • \$\begingroup\$ Difficult to answer, but one major volume consumer is electrolytic capacitors. If you take some samples from datasheets of current ones and plot specific energy density versus rated voltage, you will find that 230 V AC input -> 400 VDC rated capactors has a local maximum. If you are doing any comparision, make sure to sort out wide-range ones since that will make 120 V suffer greatly. \$\endgroup\$
    – winny
    Commented May 10, 2019 at 9:30
  • \$\begingroup\$ Very interesting. So the variable input 120 to 220V supply suffers a double disadvantage of wasting the 400V working voltage of the capacitors when running with 120V (4x energy density penalty). And even if it only needs to work for 120V, the caps would have a lower energy density like you said. LNX2G222MSEF 400V, 2200uF, D=51mm,H=133mm -> 0.65 J/cm^3 LNT2D472MSE 200V, 4700uF, D=63.5mm,H=103mm -> 0.288 J/cm^3 \$\endgroup\$
    – Yale Zhang
    Commented May 10, 2019 at 10:08
  • \$\begingroup\$ Exactly so. A PFC from the 120 V AC to 400 V DC capacitor bank in modern power supplies should mitigate the problem. \$\endgroup\$
    – winny
    Commented May 10, 2019 at 10:29

2 Answers 2


The weight is due to the transformer. Higher is the frequency, smaller the transformer is. I have a hobby MMA welder, it is a forward converter with two IGBTs and two diodes, it uses SG3525 PWM controller with a PWM frequency near 50kHZ. It is rated at 140A for 30% intermittent duty.

A smaller welder, TIG/MMA Fronius transpocket TC1500 has a full bridge converter with 4 FET transitors, fast gate drive circuitry clocking at approx 140kHz , a tinny transformer, MCU/FPGA control. It can weld almost continuously at 140A, fully protected.

The price between these two welders is 1:5. Higher power densities are possible with higher price (TC1500 is almost 15yrs old, mine hobby welder is 3yrs old) and even better efficency - the TC1500 doesn't heat as much.

So, conclusion: higher switching frequency -> less weight -> higher price.

  • \$\begingroup\$ My question implied the switching frequency for 120V is within a factor of 2 of 220V, but good point. This begs the question, why are welders still using IGBTs when MOSFETs have even lower losses @ higher Hz. Let's compare the losses for 400V input @ 15amps (avg) & 100% duty cycle (unrealistic but for sake of maximizing conduction loss). IGBT (STGP15H60DF): conduction loss=15A*1.6V = 24W, switching loss = 30KHz*0.35mJ = 10.5W. MOSFET(STW70N60DM2): conduction loss=42mOhm * 15A^2 = 9.5W, switching loss = ? (should be much lower than IGBT).That MOSFET is only $12, so why aren't they using it? \$\endgroup\$
    – Yale Zhang
    Commented May 13, 2019 at 11:45
  • \$\begingroup\$ @YaleZhang It's still a battle going on between IGBT and MOSFET, back ago MOSFETs didn't have such low Rdson at high voltages. Further, a MOSFET always comes with intrinsic diode as a residue, while IGBT comes without it. A forward converter with 2 x IGBT + 2 x diode has to have a switching element without the intrinsic diode. A MOSFET type has to be full bridge phase shifted PWM, more costly control. \$\endgroup\$ Commented May 14, 2019 at 6:24

The difference between the size is most likely due to the difference in the duty cycle.

A higher duty cycle requires more and better cooling.

Some small welders such as 1 that you mention will weld at close to full power for 5 minutes, then need 15 to cool down - how do I know? Well, I have one that does exactly that, one day I will attach a fan to it...


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.