Suppressing interference from high frequency arc stabilizer

[Ignoramus12383]

The plan of my welder in its simplest form can be reduced to this:


DC + AC HF HV Xfmr
------> +----------+ ---->---\/\/\/\/\/\/\/------> TIG torch
| |
|Inverter |
| |
DC - | | AC
------> +----------+ ---->-----------------------> work clamp

The purpose of the HF transformer is to create low power, high
frequency, high voltage between work and tig torch to initiate the
arc.

What I am concerned is whether enough interference would travel back
from the HF HV transformer to the inverter, through the power line,
that would damage its electronics.

Is that a justified concern?

If so, how I can I protect this inverter?

thanks

i

 

[Tim Williams]

DC + AC HF HV Xfmr
------>-|----------|------+---\/\/\/\/\/\/\/------> TIG torch
| | |
|Inverter | === C
| | |
DC - | | AC |
------<-|----------|------+-----------------------> work clamp

(Note the added cap)

What I am concerned is whether enough interference would travel back
from the HF HV transformer to the inverter, through the power line,
that would damage its electronics.

Is that a justified concern?

If so, how I can I protect this inverter?

A cap for one, or better yet a pi filter (Google it), but neither is a
particularly good solution when your inverter wants to switch a square
output!

But clearly, since the inverter has a low impedance output, it is going to
shunt the HF across it anyway (remember the HF start has its own current
loop, "starting" at its transformer, going through the torch, work and
welder supply transformer before returning). That just leaves the voltage
produced between transistor source/emitter and whatever the circuit ground
is defined as. If you ground the control circuitry to the inverter
transistors, which makes sense, then there's no problem, it all follows the
hash with respect to whatever higher plane you're observing form (earth
ground?).

Now, if you require the circuits be earthed, and you can't earth the output
(which as I recall, your output is floating, and all the better to leave it
that way, minus a TVS or so to keep things within limits), you might go with
isolation from transistors to control. Optoisolators could do this. Since
you require variable frequency (down to ~0Hz), coupling transformers are no
good. 10us is fast enough switching time IIRC, so even a relatively shoddy
opto would still be suitable here.

Tim

 

[Ignoramus12383]

(Note the added cap)

I see. I would like to find out how to calculate the capacitance.

A cap for one, or better yet a pi filter (Google it), but neither is a
particularly good solution when your inverter wants to switch a square
output!

But clearly, since the inverter has a low impedance output, it is going to
shunt the HF across it anyway (remember the HF start has its own current
loop, "starting" at its transformer, going through the torch, work and
welder supply transformer before returning).

Yes, that's 100% right that HF has its own path that does not go
through the inverter. I was only concerned with small currents (that,
at high enough voltage, could destroy the IGBTs).

That just leaves the voltage produced between transistor
source/emitter and whatever the circuit ground is defined as. If
you ground the control circuitry to the inverter transistors, which
makes sense, then there's no problem, it all follows the hash with
respect to whatever higher plane you're observing form (earth
ground?).

Well, the control boards that I use to drive gates, have their outputs
electrically isolated from inputs (signals, Vg, and Vcc).

That said, I am going to float the control circuit and will not ground
it to the equipment ground.

Now, if you require the circuits be earthed, and you can't earth the output
(which as I recall, your output is floating, and all the better to leave it
that way, minus a TVS or so to keep things within limits), you might go with
isolation from transistors to control. Optoisolators could do this. Since
you require variable frequency (down to ~0Hz), coupling transformers are no
good. 10us is fast enough switching time IIRC, so even a relatively shoddy
opto would still be suitable here.

There is already isolation in the boards, either 2.5 or 4k
volts. Maybe I am worrying about nothing.

i