I may be going out on a limb here (how many EE's in the world have
designed plasma cutters?), but does anyone have an idea on how the
feedback loop works in a switch mode plasma cutter? Is it direct PWM
from the power adjustment pot, or voltage/current mode with the pot
adjusting the bias? The no load voltage floats to the peak voltage when
there is no arc, and to complicate matters, there's a 10 amp pilot
current that needs to be controlled from a microprocessor (or current
sensor). If anyone has full schematics of the control section of a
switch mode plasma cutter to email, I'd really appreciate it.
BTW, it's for a hobby/educational project of mine, to eventually build
a smps plasma cutter.
I would think that the latter is the case, as the resistance of the
load (plasma arc) varies over time due to various factors like
distance of torch to work, etc.
I am building a SCR controlled plasma cutter actually. I already have
this power supply working as a tig welding machine and arbitrary DC
power supply, and use it for that, and am slowly working on also
making it do plasma at higher voltage. Its transformer can be rewired
to produce high voltage. Which I do with contactors. I use a
microcontroller programmed in BASIC.
I think that it is a lot easier to work with SCRs, for me (esp. since
I got a nice SCR controller already), than with IGBTs or MOSFETs for
I would post your questions to sci.electronics.design and
rec.crafts.metalworking. Here we have welding experts, and some
electronics experts from whom I learned, but not as many electronics
gurus as in two newsgroups that I named.
How critical is regulation? Plasma cutting seems fairly tolerant to
variations. I've seen one lincoln (procut-25,
pg 26) without an output smoothing cap. It had a shunt on the secondary
side, which is something I've almost never seen on a SMPS. Maybe
there's a PLC accepting the input from the shunt and the pot to allow
the pilot current/nonlinear power control, then using a DAC for direct
PWM of the controller. You wouldn't happen to have any upclose pictures
of a plasma cutter control board?
I've seen your site, and it's definately a project I would like to do
some day. Good luck with that, would be interesting to see just how
complicated these $5k+ welder are.
What about turnoff? Forced commutation sounds difficult and slow, but
then again there's a company (Arcon) that exclusively makes SCR based
inverters. I bet the sound of 10 kHz gets annoying after a while...The
dual forward topology means the fets will never be exposed to anything
more than Vin, so snubbers are virtually eliminated (although diode
recovery delay allows some excess voltage on turnoff). Gate drive can
become tricky, but there's plenty of literature available online from
both professionals and tesla coilers (sstc's).
Already crossposted to sci.electronics.design, but will post to
A cap does not regulate current... Am I missing anything here?
Um, I have my own control board, that's not what you are interested
My immediate plan is to connect my torch to a source of dry air and
get a pilot arc going with maybe 10a of current.
If I get to this point, I will have to do a few things such as:
1) attach gas quick couplers to the former water valve on my welder
(this valve is not presently used, I use a separate torch cooler that
I built). I iwll use this valve as a gas valve for plasma cutting.
2) Add two relays, one for turning off the pilot arc, and one for
activating the air valve, to the control board
3) Write BASIC code for plasma cutting (I have some code now, but it
is all wrong). By the way, my source code is GPLed and is at
4) I have a plastic carrying case about 20x14 inches, I want to make
it into a holder for the torch, air/electricity adapters etc, so that
I can use it to store the torch when not used. Then when I NEED TO CUT
something, I would open up the case, connect quick connects to
air/water/main power/control and start cutting.
Yeah. My welder is not exactly cheap in construction, I just use
surplus components so that the project is inexpensive, but doing what
I do on a production basis would be expensive. E.g, I have a big high
frequency arc starter module, big transformer, a lot of contactors,
IGBTs for the inverter part (not yet plugged in), SCRs, SCR firing
board, etc. All of that would cost a lot new, but cost little on a
I basically want to build something that would cost over $10k, like a
powerful multiprocess tig/mig/arc welder and plasma cutter and DC
power source in general and power supply with square wave. Out of the
above, I have a tig/arc welder and arbitrary power supply mode
basically working. (have not yet tried tig pulsing yet) Unlike some
other people, I have pictures of sample welds.
No, I think that I mis-stated what I was doing (or did not state it
I have a SCR/transformer based DC power supply. (works ok). It is
controlled by COTS SCR firing system by PCTI. The transformer is a 60
Hz transformer, a lot of iron.
For the welding square wave inverter (to make square wave AC for
welding aluminum) I built a IGBT fired inverter.
Plus there are some good gate drives sold on the market (like Semikron
20 years ago, as an exam project, I build a 150 A welding machine; That
thing used a simple two-transistor (with cascode MOSFET/BIPOLAR switches
because all devices were crap then ;-) forward inverter running at fixed 48%
duty-cycle and used a saturable inductor on the secondary side to control
the current; the control was just an analog PI controller. Nothing complex
at all (except the transformer, switches, nasty voltages and currents,
decoupling caps going open e.t.c e.t.c).
Took about 4 months to do, at eight-to-twelve hour per day. And I had a
decent lab and equipment.
Was this using a flux reseting winding? I bet transients would be
easier with cross coupling diodes, which were probably far away 20
Wow. What was the majority of the work? I was imagining a complexity
similar to http://www.stevehv.4hv.org/12kwPFC.htm
And do you have any pictures/schematics? Igor and I would definately be
interested, and I bet a few more people here.
I think that perhaps we ought to start a power electronics club or
something. I am definitely very interested. I am not an electronics
designer, I just learned what I thought I needed. I am a computer
The converter was similar to a full bridge except with the switches for one
cycle replaced with diodes so the flux reset is via the supply rail and
those diodes. The half-bridge just works here because the supply was 380 V,
The main problem was that the "pulse-capacitors" burned - i.e. lost the
capacitance - because the cascode switches were easily fast enough to exceed
the di/dt rating of them. It took a bit of experiment to get decent caps. A
second problem was that we used MOSFET's and got to learn something about
ESD the hard way - ESD was all new then (here, anyway).
Pretty similar actually. The transformer and filter inductor took ages to
design because we had to come up with a sound design method and there was
not much information to go by (There were lots of how-to books but one has
to show off some analytic capacity to pass university. ;-).
Then there is all the little things like the thermal ratings of filter
capacitors and such - we did not even do anything clever like simulations,
just simple energy-conservation calculations but all takes time and
verifying by measuring takes more time, then one does a better model where
there is pain e.t.c.
We also had to rebuild the switches a couple a times ...
But, 4-6 months is not unrealistic starting from scratch, even now that I
know much more I.M.O.
I will see if I can dig the report out of the attic - it is there
*somewhere* amongst the rubbish.
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.