1
\$\begingroup\$

I've got a fairly beefy induction heater (100-400 KHz, 5 kW output) that uses 1/4 inch copper tubing (and a coil) to a power head to deliver RF power to a workpiece. It's an older piece of equipment, and because it's an induction heater, it has a power output figure but no reflected power. It also has some minimal tuning capability (it auto-adjusts the frequency to optimize output).

I'm repurposing it as a power source for an ICP torch. I've done some work with matching networks at much higher frequences (2.45 GHz, 10 GHz, 20 GHz) and in general have some working knowledge of how to design those for waveguide or coax.

What sort of components are useful for matching at this frequency and power level? I can imagine maybe using a variable vacuum capacitor. But I have to admit my imagination fails me when thinking of how to transition from the copper tube form factor to some coaxial cable (7/16 DIN?). Copper strip brazed to the tubing that gets brazed to the center pin of a solder jack connector?

The secondary question is the implementation of power sensing. My assumption is that you would need to (because of the power level, and the low frequency) implement a coax directional coupler, likely with air as the dielectric (or maybe SF6) to inhibit breakdown / avoid thermal runaway with PTFE. Is that roughly the right approach?

\$\endgroup\$
10
  • 1
    \$\begingroup\$ Do you know how much voltage/field strength is needed to excite your ICP application? At a guess, this frequency is much too low to do anything except at severely reduced pressure (is this at atmospheric pressure?). \$\endgroup\$ Sep 3, 2022 at 17:54
  • \$\begingroup\$ @TimWilliams We've gotten it to work at "reasonable" pressures (significantly higher than a typical ICP reactor at 10 Pa) but without better coupling it's not going to get to atmospheric or higher (which is the goal). It's able to initiate breakdown at ~250A, but we haven't bothered finding the minimum. \$\endgroup\$ Sep 3, 2022 at 18:07
  • 1
    \$\begingroup\$ To achieve higher voltages/currents on a given tank, the multiplication ratio is limited by the Q factor. For most solid wire/tube coils, this is around 100-200. Most power supplies I think aren't made to couple to such high Q factors either, which may push you into a far corner of its operating range, where you then aren't generating enough MMF or EMF despite the high Q; or to modify the tuning network inside the supply; or it simple isn't stable into such a load. Q can be raised using litz in hose (water cooled). Do you have any estimate for what V/I is required to achieve breakdown? \$\endgroup\$ Sep 3, 2022 at 18:13
  • 1
    \$\begingroup\$ Also, what model of power supply is it, or what design does it seem to use? (Might not be very useful, or very apparent how it's wired, but just for information.) \$\endgroup\$ Sep 3, 2022 at 18:15
  • \$\begingroup\$ Not sure about the design. Its an Ambrell EasyHeat. \$\endgroup\$ Sep 3, 2022 at 18:23

1 Answer 1

1
\$\begingroup\$

What sort of components are useful for matching at this frequency and power level?

400 kHz has a wavelength of 750 metres and, given that it's likely you will come close to one-tenth (a rule of thumb) it's most likely that your reasons for wanting to match impedances are highly dubious.

At higher frequencies (like above 10 MHz usually) matching impedances is done to avoid signal reflections corrupting data transmission but, at 10 MHz that would involve a cable length of 3 metres or longer.

So, figure out why you think you need to perform any impedance matching. Tuning for resonance is certainly something you'll likely to have to do but not impedance matching.

I can imagine maybe using a variable vacuum capacitor.

Are you sure you don't mean tuning for resonance?

The secondary question is the implementation of power sensing. My assumption is that you would need to (because of the power level, and the low frequency) implement a coax directional coupler

These rely on the length of the coupler being about the same order as the wavelength and, 750 metres seems unfeasible in that respect.

\$\endgroup\$
8
  • \$\begingroup\$ Bunch of responses, probably revealing that I am at best a hack at RF. 1 - Impedance matching - in ICP literature, this is generally what people refer to, rather than tuning for resonance. Most ICP sources are 13.56 MHz, but the coils tend to be << 3 meters, and still employ L matching networks. I've also seen some references to CCP discharges at 450 KHz using the same. We generally "know" we need to do this because very little power is coupled to the plasma otherwise (measurably). If not able to measure power by directional coupler, what other methods are there for this frequency? \$\endgroup\$ Sep 3, 2022 at 17:44
  • \$\begingroup\$ Also, I didn't do enough googling - there are a bunch of low power directional couplers in existence for this frequency range in small form factors, from Minicircuits and others. How are they able to implement them? \$\endgroup\$ Sep 3, 2022 at 17:53
  • \$\begingroup\$ @phyllisdiller OK, let's run with impedance matching; what impedance values? \$\endgroup\$
    – Andy aka
    Sep 3, 2022 at 18:52
  • \$\begingroup\$ The plasma density (and thus impedance) in the reactor will vary with different input powers / pressures. Would likely need to have something that could tune to Z = 50 + 150j or Z = 75 + 200j. \$\endgroup\$ Sep 3, 2022 at 20:14
  • \$\begingroup\$ Two impedance values are needed plus a schematic of the reactor. Imagine I know nothing about things except L-pads and impedance matching. What gets matched to what in other words? \$\endgroup\$
    – Andy aka
    Sep 3, 2022 at 20:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.