Inverter Welder Design Using R48-3000e3

[offbyfour]

Hello, I'm looking for recommendations for an Inverter Welder design using R48-3000e3 supplies. The specs of the welder I'm trying to hit are (Multimatic 220):



The R48-3000e3 supplies have a few characteristics that I think make it nice for this application:
1. Cheap and easy to find  (about $100-$200 but even cheaper in some places)
2. Programmable output voltage (42VDC-58VDC)
3. Hot-pluggable and Stackable (for additional power support)
4. Wide input voltage range (85 300VAC) for better flexibility and so that it can work with dirty supplies like generators.

But the main reason I want to use these supplies is because they dont require significant exposure to high voltages which make them a little easier to DIY. Although I understand the drawback that this is probably not the most efficient way to supply the circuit.

For this design, I'm thinking 2 paralleled supplies should be good enough to meet the circuit requirements.


As far as the converter, from what I have found online it seems inverter welders can have the following topologies:



I'm thinking of going with the Full-Bridge topology since that seems it would result in the most flexibility and efficiency. I would also like to be able to run this off a standard house/garage outlet (240@ 30A breaker, 24A continuous) so better efficiency for this stage (>95%) seems worth the extra cost.

The switch requirements in order to meet that seem tough though. For max power, it seems I would need to support either 58VDC @~100A or 42VDC @~140A. The closest part I have found that may give good performance is the GS61008P (100V 65A@100C) but I would need to put two of them in parallel (or 8 total for a FB topology).

To drive these, I'm thinking of using an FPGA/CPLD as that would give me the most control over timing as well as load balancing the paralleled switches.

Although there are still quite a few details to work out, I wanted to get feedback on what I have so far to make sure I was starting in the right place.

Any recommendations, reference designs or documentation is appreciated!

R48-3000e3 Datasheet
https://www.vertiv.com/globalassets/products/critical-power/dc-power-systems/netsure-731-a412/r48-3000e3-datasheet.pdf

GS61008P Datasheet
https://gansystems.com/wp-content/uploads/2020/09/GS61008P-DS-Rev-200402.pdf

[moffy]

If using a Full Bridge configuration I think I would look at using a bridge rectifier on the secondary as that would halve the number of high current turns on the secondary of the transformer and improve the copper utilisation but it does increase losses in the diodes.

[offbyfour]

Anyway to tell if a bigger transformer would result in less losses than the diodes? I dont 100% understand the output stage transformer and diode configuration so I'm not sure which would be the best.

From what I can tell the transformer takes the PWM from the Full-Bridge and steps down the voltage/ups the current and I can see how the ferrite and the caps smooth the ripple but I dont understand what the diodes are doing.

[moffy]

The transformer has two windings, center tapped, which are the inverse of one another i.e. while one is positive the other is negative, or they are both zero together. The diodes only conduct when that winding is positive and block when the winding is negative (they will also conduct when 0V for some time due to the output inductor). They just make sure you only get a positive output voltage but each winding is only used 50%(I am including the off time) of the time depending on which is positive which under utilises the copper on the secondary.

[offbyfour]

Whats the benefit of center tapping the windings? I would think you could get positive only by going across the entire winding?

[moffy]

Whats the benefit of center tapping the windings? I would think you could get positive only by going across the entire winding?

A full bridge rectifier does that for you, the diodes do the switching, the important thing about transformers is that they can only handle AC, no DC, so if the primary is positive for some time it needs to be negative for the same amount of time to remove any DC component. That means a single winding will only be positive at most 50% of the time if we define one end of the winding as 0V
The following article might help a little: https://www.ti.com/lit/an/snva006b/snva006b.pdf

[NiHaoMike]

The very cheapest would be "telecom rectifiers" on the surplus market. Then add a multiphase buck converter with all the MOSFETs on the low side and a relatively small, high voltage film capacitor on the output. (The high voltage rating is required to handle the huge transients.)

You can use lipo batteries for the power source although one capable of such high power is going to be quite expensive. But maybe cheaper than getting a 240V outlet installed and it can double as a general purpose portable power supply.

[offbyfour]

How would I get isolation using a multiphase buck? Or is it not needed since the supply is isolated already?

[NiHaoMike]

It's an isolated supply, no need for additional isolation.

BTW, really read into how to make your design tough against EMI, you're essentially dealing with a very high power spark gap transmitter.

[offbyfour]

What are your thoughts on doing a buck/boost circuit instead? 15-30V for welding and 145V for plasma cutting?

[robby60259]

Hello, I'm also working on a welder with a full H-bridge design. Here is a some documents that might help. Also the Youtube channel "Power Electronics with Dr. K" was very useful.

[offbyfour]

Oh awesome. Appreciate the links! I'm sort of branching out on this project and going for more of a universal buck boost supply that will hopefully be able to weld as well.

But if you need any help with your design and don't mind making it open source, I'd be happy to help out. I know my way around KiCad and can do some PCB layout and things like that if you need it. I'm also learning Q-Spice but  I'm not 100% with that one. Completely fine if you dont though!

[Dinesh6252]

Hello, I also have 3-4 emerson rectifier modules 4000 watts(R48-4000) which I have hacked for 40V-58V output with a potentiometer. 1. Can these be directly used as a welding machine?
2. How about paralleling two for higher current or connecting them in series for higher voltage?

[offbyfour]

hello Dinesh, I know for sure that they can be paralleled for higher current but Im not sure about series.

In terms of direct welding, with my limited understanding, you may be able to start an arc but might not get good penetration since the current is not too high.

The other thing is you want to configure the supply for constant current mode that way the over current protection doesnt get in your way.

I have not had the chance to work on this as much as i would want to, but i am keeping track of all Ive looked into here. if you have any instructions on how you hacked the rectifier id love to take a look!

https://github.com/offbyfour/DC_Supply_5p5kW/tree/main