Class E Amplifier Harmonics / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Class E Amplifier Harmonics

1 2 next

Move Thread

LAN_403

bob800

Mon Jan 06 2014, 07:22PM

Registered Member #18429 Joined: Sat May 18 2013, 02:05AM
Location:
Posts: 19

I finally achieved "breakout" with my Class E tesla coil setup, based on this schematic.

However the amplifier is still not working quite right and the IRFP450 MOSFET gets hot pretty quickly... Usually this is normal before the circuit is tuned such that the drain-source voltage is at 0V before the MOSFET switches on.

I'm having a hard time tuning the circuit as "harmonics" (correct me if that is not the correct term) keep appearing in both the gate-source and drain-source waveforms. This is evident in the attached waveform (drain-source 50V/div, gate-source 10V/div). This is actually the best waveform I have achieved, as in the past everything was plagued by extreme noise/harmonics (see second waveform).

This doesn't seem to be a "normal problem" among various Class E tesla coil projects I've studied on the internet. In fact from the waveform it seems like the circuit is mostly tuned----the gate turns on when the drain is "off". But due to the harmonics/ringing, it looks like in certain places, the drain and gate do begin to overlap...

What usually causes this sort of behavior? I've attached pictures of my setup in case physical layout might be a factor...

Other information:
--Supply voltage was rectified 50V from isolated variac, with ~1000uF of filter capacitors
--I am using a 1kV 470pF drain-source cap instead of the prescribed 330pF
--For the "RF choke" I am using this inductor. I tried using several other types, including chokes from power supplies and winding on a real "RF Choke" material, but the linked inductor seemed to work best.
--Breakout begins almost exactly at 50V, though not always... (still very sensitive)

Sigurthr

Mon Jan 06 2014, 09:49PM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

Which channel was your scope's trigger set to? When I was scoping a TC of mine a while back I got something similar, where the DS waveform was irregular (there's more GS peaks than DS peaks) and I was trying to figure out why I was getting odd ringing and such. I fiddled around with the trigger and got a point where the waveforms were clean and as suspected. The TC was fine, but my probing was not.

Your DS looks clean with no real ringing or under-shoot, but your gate looks pretty bad in terms of ringing. How much of the ON gate waveform is under saturation threshold I don't know, so it may not matter too much. But I would guess the heat issue lies with the gate drive, and the timing discrepancies are related to that and funky scope triggering.

Dr. Dark Current

Tue Jan 07 2014, 02:56PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

I think this was discussed here, with respect to drain voltage peaking only every other cycle (with double amplitude). Do a search on the TC board, think it will be a thread on the first page. May have something to do with a too high Q but I don't remember correctly...

bob800

Wed Jan 08 2014, 01:46AM

Registered Member #18429 Joined: Sat May 18 2013, 02:05AM
Location:
Posts: 19

Sigurthr wrote ...

Which channel was your scope's trigger set to? When I was scoping a TC of mine a while back I got something similar, where the DS waveform was irregular (there's more GS peaks than DS peaks) and I was trying to figure out why I was getting odd ringing and such. I fiddled around with the trigger and got a point where the waveforms were clean and as suspected. The TC was fine, but my probing was not.

I think it was triggered by the gate. Your suggestion about the triggering is really helpful--I didn't realize how crucial the trigger settings can be!

Dr. Dark Current, thank you, I think that was part of the problem (high Q). I added some turns to my secondary, and, most importantly (I think), I used a smaller plastic form for my primary to increase coupling. Immediately I got a very nice class E waveform and hot bushy discharge!!! Thanks for your suggestions!

Sigurthr

Wed Jan 08 2014, 05:28AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

Very glad I was able to help! And excellent news that you got it running!

So you said you're using a different inductor for the dc choke, and a different drain-source cap, I'm curious what your coil's resonant frequency is, and the inductance of the inductor you're using. I'm looking to try out class-E again soon (and have a thread up asking questions about it) and could use any input you can offer as far as data points of your design implementation.

bob800

Wed Jan 08 2014, 11:36PM

Registered Member #18429 Joined: Sat May 18 2013, 02:05AM
Location:
Posts: 19

Hi Sigurthr

I attached my ExpressPCB file if you are interested. It is entirely based on the schematic I linked to, but note that the 2n3906 and 2n3904 transistor LABELS are reversed in the schematic (I fixed in on the PCB).

During my troubleshooting of this project, I was messing around with several variables at once so it's hard for me to give a definitive "this works" and "this doesn't work". However here is what I believe is important:

--DO NOT CONNECT THE POWER GROUNDS AND SIGNAL GROUNDS. Maybe this is obvious but it didn't occur to me for awhile and messed up my gate drive waveform a bit. Also now that I think of it, they are connected on my PCB so you should remove the connecting tracks.

--If you want to include all the power components (IRFP450, etc) along with the signal components on the same PCB, use a double-sided board so you can use the top copper layer as a ground plane. I ended up removing the power components from the board and soldering everything directly on the MOSFET to minimize ringing; however with a ground plane you shouldn't need to do this.

--Twist wires together whenever you have excess length (ie connecting wires between GDT's, primary coil, etc_). I'm not sure exactly why this helps but I think it minimizes inductance?...

--I didn't use any special winding techniques for my GDT's. I simply took apart old power supplies, looking for ferrite beads (not sure if that is the right terminology...) Anyway they are NOT the colored/powdered iron cores commonly found in switching power supplies, instead they are the type power cables are wrapped around for EMI suppression (see my PCB shot in the original post to see what I mean). These worked very well.

--Don't be too concerned about the waveforms in the low-power drive circuitry... at first I got all concerned about ringing and such, but everything turns into a nice sine wave once connected with the drain capacitance of the IRPF450.

Here is the original project http://uzzors2k.4hv.org/index.php?page=4MHzclassE1.

My source cap is a random 1kV 470pF that I found on an old TV board, it seems to work well... This is the inductor I used, though good PC power supply toroids should work as well (in my experience they did not work quite as well, though I didn't bother to rewind them... I just guessed on the inductance).

I'm not sure about the res. Frequency of my coil... I honestly just wound a toliet-tube full of 30 awg wire (can give dimensions if necessary), and then removed turns until I got breakout. Of course this is not as easy at it sounds...

Also coupling is important, I found a 2" PVC form works well for the primary. I used 6 turns of 22awg wire.

Good luck!

bob800

Wed Jan 08 2014, 11:39PM

Registered Member #18429 Joined: Sat May 18 2013, 02:05AM
Location:
Posts: 19

Didn't realize the file types were limited... just remove the ".jgg" to use the PCB file.

bob800

Wed Jan 08 2014, 11:41PM

Registered Member #18429 Joined: Sat May 18 2013, 02:05AM
Location:
Posts: 19

Sorry for the triple post!! I can't upload an attachment on edit.

Looks like jpg is rejected also. Last try...
]classe.zip[/file]

Sigurthr

Thu Jan 09 2014, 12:29AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

Thanks very much! I actually have that exact inductor.

As far as the ground issue goes; you used a GDT then yes, connecting the grounds across sides of the GDT breaks the isolation the GDT grants.

It seems you used the "fiddle with it until it works" method, haha, that seems to be the most effective method for Class-E coils from what I've seen. I'm going to try a GDT-less design for mine, which really helps to eliminate ringing and stray inductance.

Btw, the reason twisting the leads helps is that increased coupling between the wires results in higher capacitance to offset the stray inductance. Also, like in data lines, common mode interference is reduced.

GeordieBoy

Mon Jan 13 2014, 03:33PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

You often get this sub-harmonic oscillation break out in Class E amplifiers if the RFC inductance is sized too large. RF engineers call this squegging. I think it's a kind of relaxation oscillation between the non-linear MOSFET and the drain choke parasitics. In a radio transmitter it is very bad because it broadens the spectrum of what is being transmitted and causes much interference. Of course it also wrecks the nice high efficiency switching waveforms too!

Keep the drain choke to a reasonable size: Just enough to keep the supply current ripple low and RF out of the power supply wiring. It's also worth checking that you are driving the gate hard enough too. At high MHz frequencies it can be a challange getting enough voltage swing on the gate to fully enhance the MOSFET during the "on time". You want the gate drive to be sufficiently strong to slam the device between pinch-off and hard saturation as quickly as possible. Lingering in the linear region is a recipe for all sorts of bad things to happen.

-Richie,

1 2 next

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