Yes in my first two or three posts I discussed the HF start.
I am more concerned about the modulatable current source.
I am concerned about this scenario:
Plasma channel forms, current source is correct, but wind/motion/splatter/etc causes the arc to momentarily break, which causes the voltage to go to its maximum span, with a partially formed plasma arc thats basically there for thermal reasons.
If a arc/plasma channel that exists is extended due to potential, does the HF spike form, as it does with a spark being formed in the first place without a initial plasma channel? Or does it electromagnetically ellongate like a lava lamp or something, so there is no HF or less of a HF transient created vs a non-formed spark?
I am worried that the feedback will end up getting a bunch of HF garbage on it in the case of this kind of 'intermittent' connection and cause the circuit to malfunction or other wise damage it.
I would like a series connection to the load, without galvanic isolation. I don't want to deal with transformers.
Should I protect both the output and the feedback path together, or layer in different protection on the output and the feedback path ?
The start circuit is basically a gimick that needs to be tested (like a motor run/start) to make sure it does not damage the current source... It does tie into feedback though, perhaps passive protection could also help shield the arc from the starter circuit.
My idea for the stater circuit was basically to have a HV rail connected to a pulse circuit and a high power fast transmission gate network, either consisting of fast robust semiconductors or vacuum tubes, which would work like this:
1) Current source is isolated by fast switch from the electrode
2) HV rail is passed through to the electrode, by DC pulse
3) the pulse is configured so that a respectably hot conductive plasma channel can form between the electrode and the work piece. For safteys sake it might be better to use a charge pump to limit energy rather then to just short the rail out (with a controlled impedance, but still)
4) once the start arc formed is sufficient, large to have a useful thermal time constant, the high voltage rail is switched off
5) the pass gate is opened to connect the current source to the plasma channel/work piece network, at this point its in weld mode
6) it stays in weld mode and during weld mode can be adjusted with the foot petal or thumb switch or whatever control you have and whatever settings (polarity, frequency, offset, waveform,etc)
But anyway, I would like to keep it linear since I know how to manage the heat and its all familiar to me, its so esoteric that Idon't care to work on a bipolar switching supply
I am working on the following:
1) the isolation switches choice (it seems that the big boys use more of a passive network from patents but I only glanced at them). But, consider the fact that this thing is like 5 amps MAX, so you don't really have to use high-power design for protection.. you are given the opportunity to use something like a make before break switch since these currents and voltages are kind of sneezeable with modern semiconductors, if anything you can limit the size of passive protectors and make the circuit more ideal during operation without parasitic protection elements required by something that can put out 200A/90V
2) the howland current pump buffer design (transistors type/model, bias levels, etc)
3) feedback filtering (i.e. something like a active LPF on the current feedback, time constant that should be held to for the control loop)
4) any possible problems created by trying to turn a kind of unique APEX op-amp into a composite current pump, possibly problems with the topology
5) additional feedback filtering evaluation (i.e. common mode choke on the current shunt, bulk inductor/ferrite on the output signal, pass capacitor, TVS, maybe HV GDT and related impedance network)
The ionization potential of various weld gasses is like <30V. I am not sure why those welders go up to 80V, maybe its just voltage drop across longer arcs? Not sure what to do. I will look in a book about it.