Miller DX200 welder SEMIKRON SkiiP 33NEC125T2 Power IGBT Module

I do not take credit for someone else's work.

Here is the assumed topology of the OEM diagram:

I am doing in this order:

Separate Diode Power Bridge Rectifiers from IGBT components to reduce a heat spread.

OEM PART IMAGES:

Now side by side:

Here is someone else's work so I do NOT take credit for reverse engineering ( also this all just assumption based on this diagram )

Based on the diagram on the image following components is no longer available for sale:

Component released on : July 2007

And last but not leased another member of this forum made a PCB preview ( PCB not complete and needs redesign )

it seems the bottom right corner of the original has 4 IGBTs in parallel which are used for PFC, not really critical if higher amps IGBTs used.

Another thing is I will attempt to measure the threshold value of Vge for main IGBTs. Will publish when I'm done, so we can compare with FGL40N120 on schema above.

it seems the bottom right corner of the original has 4 IGBTs in parallel which are used for PFC, not really critical if higher amps IGBTs used.

Another thing is I will attempt to measure the threshold value of Vge for main IGBTs. Will publish when I'm done, so we can compare with FGL40N120 on schema above.

Under question ALL resistors Base to Emitter.

I am fully aware that floating base should not be floating, however, 1K value under question. ( we do not know yet if OEM PM has it build into module.)

Remember we are assuming at this point. ( I haven't done any test yet )

Also, we need look into other SKIIP product line that designed for  120 -125 AMP 120 240 380V  PM and check datasheet's.

We could get away with making PCB spacer for new SKIIP PM. ( to avoid reinventing something already on the market )

Usually Manufacturers keeps everything similar + - some changes on Gate voltage some less powerful some has more or less IGBT's.

I have found some datasheet for another module by SKIIP

Here is adjusted image of original power module. Resistors measured and they are only under 10 ohms connected in series with each gate, needed for better balancing due to PFC circuit has 4 of them in parallel (Q9 - Q12) and each side of the push-pull output is 3 IGBTs in parallel so again each gate need to have small resistor for balancing. So, if we look at generic NAC schema then it seems that just U-V-W all output connected in parallel and have one output I marked as "Main OUT", so again there are NO any resistors grounding gates 1k there at all. 

Actually since IGBT inside is a similar to darlington where first transistor is Mosfet and second regular BJ, the gate is simply like a capacitor which works off input voltage charge therefor the most importantly will be to find a perfect match by Vge threshold (Gate-Emitter threshold) value, another words we need to know at which voltage the output BJ is fully opened or closed. so will publish my outcome of the research as soon as I'll have time to get it measured as best as I can, just to be sure we're in the needed range for the driver of these IGBTs for PFC & main inverter drivers.

NOTE: I'm still in dark on role of Q7 & Q8, but will get there as soon as full schema of original PM is complete. If anybody can analyze that would be great, otherwise later then.

Do to inactivity project has been moved.

yes, the last looks about right, except the each of those gates has around 8-9 ohm resistor so 4 of them to be added. And also one or 4 untrafast diods doesn't matter as they will open and take care of back emf coming from inductive reactances in circuit. But in our version we don't need to add those diods externally as each IGBT already has them inside. About those 1k resistors - I don't mind having them as it will only protect ext. board from statics by providing a small leak from gate or will short circuit if static applied, they will not affect functionality of the IGBT.

Images and Design has been moved.

Also I have check numerus SEMIKRON SKIIP datasheets regarding Vge =15 V standard on all of them.

I haven't not complete entire reverse PM to RC pinouts. however, as of right now looks like PM controls by ASSY NO. 207822 and not by Boost Board assy No: 200841

Sure, it's good to know on Vge threshold, just one concern, normally they give a range like 4v - 8v.

As of schematic it's a good start.

And as of PFC part I'd worry about it last as unit can work if PFC is not even implemented or dead, like instead of 200A you may get 160-180A but it should work as PFC only recovers some lost energy due to inductive reactance small phase shift correction. 

F.Y.I. below is fragment of the "Vge threshold" that we're looking for from the datasheet you posted above:

Let me look at the Semikron SKIIP datasheets.

Here what I have found:

NAC vs NAB:

Seems module has build in Chopper - IGBT ( this information has not been confirmed for PM SkiiP 33 )

The first datasheet shows Vge (th) threshold in the range of 5 to 6.5v w/ average 5.8v and that measurement is done at 6ma in load, typical value of Ic is 250uA, not sure why german company decided to go w/ 6mA, probably it is their standard, but it's Ok.
As of second datasheet above does not publish Vge threshold at all, so, you can only guess.

Also, there are 4 generations of those SkiiP Power Modules and all are different as designed at different time,

So, guessing may get us in trouble. Another parameter is Vce sat - also important as size of the radiator directly depends on the value of voltage  drop on collector-emitter at max load let say 50A will get us to let say Vce sat = 3v will be needed to dissipate 150W where 3v drop is a real number from datasheet above, and that is way over for DX200 radiator. So, less value is a preference, 

As I've heard this 33NEC125T2 is heavily customized specifically for welders schematically w/ the best IGBTs used at that time, so I'd be careful, but sure you can experiment with what you have, you gotta start with something. Meantime, I'm thinking about the way to take measurement without destroying good power module which I can't afford, and we need to get a close match by most critical parameters and better match those IGBTs will help us to get a stable outcome.