LCLR Matching for induction heating question / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

LCLR Matching for induction heating question

1 2 next

Move Thread

LAN_403

nrhoades

Fri Mar 30 2007, 03:39AM

Registered Member #610 Joined: Wed Mar 28 2007, 09:44PM
Location: Middletown, RI
Posts: 110

I think I am having trouble with a concept, or I'm thinking too hard about
this...

On Ritche's site, he uses a 45uH inductor as his matching inductor. His
tank Lw and Cw say that the res. freq. is 209kHz... so technically at this
frequency the tank looks like R.

The way I am trying to understand this is that the matching inductor does
2 things:

1. Blocks high current spikes from switching.
2. Compensates for the DC blocking cap so that it cancels out the cap's
reactive component.

Since Xc = 1/jwCm and Xl = jwLm and we want Xc + Xl = 0, then Lm=1/(w^2*Cm).

If Cm is 470nF, the equation says that Lm should be 1.23uH. If he is using
45uH then his network isn't matched, and his power transfer isn't as good
as it could be?

Steve Conner

Fri Mar 30 2007, 10:21AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

No, it doesn't resonate with the DC blocking cap. If someone calls a capacitor "DC blocking", they mean that its value was chosen large enough that its reactance is negligible, and it shouldn't participate in any resonance.

When the matching inductor is added, the resonant frequency of the whole system moves slightly such that the tank circuit doesn't "look like R" any more. It looks slightly capacitive, and it is this capacitance that resonates with the matching inductor.

Also, its main function is not to block current spikes, but to regulate power transfer by performing an impedance transformation. It works exactly the same as the "L-match" circuit described in ham radio handbooks. You typically guesstimate the value of the matching inductor, and then adjust it experimentally until you get the amount of power draw you want.

Steve Ward

Fri Mar 30 2007, 04:05PM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

Thanks for that explanation Steve, Nate had emailed me personally about the operation of this circuit, and i think i was a little off in my understanding of the actual function of the L-match.

The important thing is that the inverter "sees" an inductive load (worst case) rather than a capacitive load in worst case, and ideally looks resonant to the inverter.

This reminds me that i still need to critically analyze this circuit in pspice.

nrhoades

Fri Mar 30 2007, 08:31PM

Registered Member #610 Joined: Wed Mar 28 2007, 09:44PM
Location: Middletown, RI
Posts: 110

Haha, fun coincidence... Hi steve. BTW I emailed you after I posted here so I didn't take your advice for granted.

I did some analysis on this problem... Ignoring all real (resistive) components, and assuming Ritche's component values... this is the response of his LC tank circuit WITHOUT the dc cap and matching coil:

Now, can anyone help me understand this from a power perspective? What are you trading when you try to maximize the power transferred to the work coil?

-Nate

PSPICE: V(a) = tank voltage, I(L1) = work coil voltage

(Since we're ignoring real components, the poles will exactly lie on the imaginary axis. The Bode plot is simply a slice of the 3D pole-zero plot through the imaginary axis)

When the blocking DC cap is added, it adds a zero to the response. This is what the total impedance looks like from the source:

PSPICE: V(a) = tank voltage, I(L1) = work coil voltage, I(V1) = source current

The input current, tank voltage, and work coil current all take a dive at DC. But now the tank voltage is no longer protected from switching transients at high frequencies, and it looks even worse for the input current. Lets see what happens when we add a matching inductor...

PSPICE: V(a) = tank voltage, I(L1) = work coil voltage, I(V1) = source current

Now we have increased impedance at frequencies above and below the resonant frequency. Technically our impedance at the resonant frequency still sits on a pole... but as soon as we add resistance to the circuit, the pole will slide away from the imaginary axis. When this happens we will be taking our slice of the 3D plot near the base of the pole, and the high peaks will round out (lower Q). In other words, the peaks wont be as sharp.

Adding the matching inductor adds another pole to the system. In Ritche's case, he chose a large inductor, so his new pole sits very close to the DC blocking cap's zero:

Notice what happens if we choose a smaller inductance value...

Now the new pole is further away from the DC capacitors zero. The impedance profile changes, and the pole is less influenced by the zero.

In conclusion...

The ideal value of the matching inductor should be right at the point where the inductor's pole starts to merge with the DC cap's zero, and not high enough to allow the other pole to shift to the left too much and increase the reactance of the system.

PSPICE: V(a) = tank voltage, I(L1) = work coil voltage, I(V1) = source current

------

Also, its main function is not to block current spikes, but to regulate power transfer by performing an impedance transformation. It works exactly the same as the "L-match" circuit described in ham radio handbooks. You typically guesstimate the value of the matching inductor, and then adjust it experimentally until you get the amount of power draw you want.

Now, I'm still not 100% clear on the deal with power transfer...

The MATCHING part is because:
The source (inverter) is capacitive (because of DC cap) and the tank (also driven at a frequency that makes it capacitive) both need to be matched with a compensating inductance?

Steve Conner

Sat Mar 31 2007, 10:45AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

If you ignore all real components, how in the world do you expect your analysis to tell you anything meaningful about power transfer? If you wanted to calculate the matching inductor value, you'd need to assume a load resistance shunted across your work coil, and a source resistance for your generator (which for a voltage source inverter is usually the DC bus voltage, times 4/pi, divided by the maximum current you feel happy drawing from it)

Once you know these two real resistances, you can apply the L-match design equations found in ham radio textbooks. This gives you the inductor value. The L-match will "borrow" the required capacitance from the tank capacitor (so you must make sure the capacitance value that falls out is small compared to the tank capacitance)

In practice, nobody can be bothered calculating or measuring the effective load resistance due to a workpiece in an induction heater, so the matching inductor is adjusted by trial and error.

nrhoades

Sat Mar 31 2007, 12:05PM

Registered Member #610 Joined: Wed Mar 28 2007, 09:44PM
Location: Middletown, RI
Posts: 110

Ahh right. OK I understand now.

Anyone know of good resources for building MOSFET or IGBT driver circuits, specifically for an H-bridge configuration?

Dr. Shark

Sat Mar 31 2007, 05:20PM

Registered Member #75 Joined: Thu Feb 09 2006, 09:30AM
Location: Montana, USA
Posts: 711

Firstly, let me say it's good to have you on 4HV, you seem to be putting some effort into what you do. I am sure you will get your heater working nicely.

If your question is specifically about driving the transistors and not the frequency tracking... I suggest you sample some gate driver chips from Texas Instruments ((UCC37321/22) or from Microchip (TC4421/22). They work very well for this kind of application.

nrhoades

Sat Mar 31 2007, 05:53PM

Registered Member #610 Joined: Wed Mar 28 2007, 09:44PM
Location: Middletown, RI
Posts: 110

Yes, I have a bridge driver chip from Intersil (I'm sure its similar...) The Intersil one I have can drive an H-bridge (with shoot-through protection and programmable dead-time) with 2.5A drive per gate and can work upto 75V. What steps can be taken if you want to use a higher voltage on your DC bus? Should I use something like a totem-pole arrangement if I want to increase the current drive?

Also (and I just know about the theory of MOSFETs and less about the practical limits) don't I need to be careful about the dv/dt... i.e. I can only switch the gate voltage so fast before I risk latching the thing up? This would make me think that I would need to place resistors on the gates of the MOSFETs to limit the current that charges the gate, thus limiting dv/dt.

May I depend on the intrinsic body diode to recover fast enough, or is it common practice to add freewheeling diodes between the drain and source?

And my last question: What are your experiences with H-bridges you've made regarding voltage vs. current? I have some pretty nice MOSFETs from IRF that are rated for 80V and 170A, have a Qg of about 180nC and a dv/dt of 16V/ns plus a fast recovery time... would these work well for my induction heater?

I wish that the rockclimbing forum I use was as professional as 4HV!

-Nate

Dr. Shark

Mon Apr 02 2007, 10:18AM

Registered Member #75 Joined: Thu Feb 09 2006, 09:30AM
Location: Montana, USA
Posts: 711

Sounds like the driver chip you have is a combined high side driver that is internally isolated to the voltage the hot side of the H-bridge sees. If this is the case (maybe you can post a part number), 75V is way too little. What people here usually do is to wind gate drive transformers to isolate the driver from the hot side.

I have never heard about latching MOSFETs, apparently it was a problem with older IGBTs, but nobody seems to worry about it nowerdays.
The intrinsic body diode of MOSFETs is generally veeery slow, so you would need to bypass it with a Schottky in series to the MOSFET and a fast high current diode antiparallel to the two. It is way easier (and cheaper) though to just use a IGBT co-packed with a fast diode.

And my last answer: Generally people tend to use 600V 20-50A devices for this sort of thing. An 80V device limits you to a Vcc of 30V or so, and it is not easy to get considerable power to flow at that kind of voltage. The impedance of the whole circuit will be extremely low, every piece of wire will act as a resistor etc., so you will be much better off working at a higher voltage.
Of course low voltage MOSFETS off unique advantages like a very low R_ds, so if you feel up to the job of designing a very low impedance circuit, go ahead!

BTW, the archives contain a lot of information that would be useful to you, both regarding induction heating as well as general H-bridge design. It may take some time to find the relevant threads, but I assure you it is worth looking for.

Steve Conner

Mon Apr 02 2007, 10:31AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

75V would be about right if it were a HIP4082 or suchlike. They're meant for battery type applications. The IR2181 is a similar chip that functions up to 600V. However, we prefer gate drive transformers as they're simpler and more robust. You only need these chips if you're trying to synthesize a gate drive waveform with a low frequency component which a GDT can't pass. An example of this would be a motor drive or Class-D audio amp.

MOSFETs don't latch.

The body-drain diode seems to work fine unless you are driving a capacitive load at 100s of volts and 100s of kHz. Then forced recovery overheats it and the recovery spikes rip the rest of your circuit apart too. Newer MOSFETs boast that they have body-drain diodes with fast and soft recovery.

I have induction heated stuff at 200kHz

using an H-bridge of 55A, 100V TO-220 MOSFETs running off a 30V DC supply at up to 7A. The FETs needed no heatsinking and coped with capacitive loading fine. The power I could run was limited by the tank capacitor overheating.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob