# Type of topology

Joined Nov 7, 2008
18
Hello!

At work i repair CNC plasma cutters and we have a SAF oerlikon 300A power source.
The mains voltage is feed to a transformer and from that it gets rectified and smoothed out.
The DC voltage is feed to a so-called chopper unit that is basically a pwm regulator...
the chopper unit has two brick IGBT's mounted on a heatsink... anyway a unit has died (the IGBT's blew) and i had to repair it.... the interesting thing that catched my eye was the switching principle of the IGBT's...
they are 180 degrees out of phase. Each igbt is rated for 1200V and 400A current
so my question is why are the IGBT's switching out of phase and why use two if one can handle the current?

Thank you!

#### BR-549

Joined Sep 22, 2013
4,936
Schematic?

#### GopherT

Joined Nov 23, 2012
7,983
"180-degrees out of phase" & PWM
I would like to see a circuit and maybe a waveform that goes with each for clarification.

Joined Nov 7, 2008
18
i have the waveforms drawn at work. i will upload a photo on monday... i don't have schematics of the driver. I did draw the gate drive section of the control board because it always blows a transistor or zener diodes when the IGBT brick blows.
with the statement 180 degrees PWM i meant that the igbt's are not driven in parallel (both on at the same time or both off at the same time). The switching speed is 13KHz so the period si about 77us long so if the first IGBT is turned on at 0us the second one is turned on at 38,5us (half of the period). The ON period of one IGBT is always below 50%. If it was 50% the output would be continious DC (always on). If it would be more than 50% the IGBT would overlap .
Here is a very basic schematic how the IGBT's are connected...

Joined Nov 7, 2008
18
i managed to simulate the pwm signals so here are some different pulse widths...

Joined Nov 7, 2008
18
First image shows a 5% duty cycle second one 25% and tle last one 40% ....

the controller works as a current source. i drew the basic schematic without the snubber networks and protection diodes and current transformers...

#### shortbus

Joined Sep 30, 2009
6,882
I'd guess that each IGBT is switched on in turn. The gates controlled in a "flip-flop" fashion. This would allow a cooling down period for each IGBT. I've been trying to design a power supply for pulsed EDM(electrical discharge machine) that will use the same principle.

#### #12

Joined Nov 30, 2010
18,076
my question is why are the IGBT's switching out of phase and why use two if one can handle the current?

Thank you!
That drawing doesn't really show an "out of phase" relationship to me. It's merely sharing the load between 2 devices where one of them would melt if it had to do all the work alone. We were suspecting a center tapped transformer to work with a 2 phase approach, but you don't have one, so both IGBTs are carrying current in the same direction, they're just sharing the load.

Joined Nov 7, 2008
18
#12 i'm sorry for the wrog expression ... but thank you all for the explanation... i thought this could be the reason, but i had to check. The current is very large and with 3,5V at saturation the device produces 1050W... so yes it needs to cool down ....
but we can't figure out why the IGBT's blow... it only blows one of the two IGBT's and this is very random. Sometimes they blow at 100 working hours and somtimes they blow at 3000 hours and when it blows it also smokes the gate resistor (2,2 ohms) and a few zener diodes on the controller. Anybody has any experience with IGBT devices? i was thinking transient voltage spikes or maybe the rise time gets too long? we never have to change the RC snubbers and the high current protection diodes are almost always OK....

#### NorthGuy

Joined Jun 28, 2014
611
As it is drawn, when an IGBT turns off, it should produce a huge voltage spike through the inductor which has a potential to blow the IGBT.

#### GopherT

Joined Nov 23, 2012
7,983
As it is drawn, when an IGBT turns off, it should produce a huge voltage spike through the inductor which has a potential to blow the IGBT.
Yes, this is the whole point of a plasma source - high voltage. It is not possible to get the high voltage with a steady DC source so the transistors pulse alternately and alternately discharge (and spike) as the igbt closes.