4mhz coil transistor choice / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

4mhz coil transistor choice

Move Thread

LAN_403

tritium

Sun Mar 21 2010, 05:31PM

Registered Member #2591 Joined: Fri Jan 01 2010, 11:58AM
Location: netherlands
Posts: 76

Hello coilers,

I wan't to build a powefull 4mhz coil, what do you think of this mosfet: IXFK 21N100F

I am aiming at 2KW

ScotchTapeLord

Sun Mar 21 2010, 05:47PM

Registered Member #1875 Joined: Sun Dec 21 2008, 06:36PM
Location:
Posts: 635

2KW might be hard with the .5 ohms of on resistance. That's the only downside of this particular MOSFET I can see. Richie Burnett has some advice for paralleling drivers for getting more power at that frequency at the bottom of his 4MHz coil page,

but you've probably seen that page if you're interested in 4MHz coils.

tritium

Mon Mar 22 2010, 04:20PM

Registered Member #2591 Joined: Fri Jan 01 2010, 11:58AM
Location: netherlands
Posts: 76

yes, I have read the page of Richie Burnett several times, but Richie uses normal mosfets so that anybody can build/copy the thing, I wan't to use special parts that are made for this task like this mosfet, designed for rf-switching.

I have also noticed that the coil of richie is 3.8 Mhz without break-out, so if there is a big corona flame, the frequency will be lower than 3.8 Mhz because of the extra capacitance on top.
This is one thing I don't understand for his coil, but he is a specialist and i'am not, so...

My coil is selfresonant at 4.5 Mhz, but maybe I have to redesign my resonater, all tips are welcome.

GeordieBoy

Mon Mar 22 2010, 05:07PM

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

Hi tritium.

The MOSFET you specified should be able to produce about 2kW of power at 4MHz provided you can sink enough heat from it. It also looks like a large die device so you may find it needs quite a bit of drive power to slew the gate.

If it was me I would probably use a pair of IRFP460LC devices and parallel up the outputs of two discrete amplifiers. I don't know how much you're paying for that device but standard switchmode MOSFETs are more than capable up to about 10MHz when operated in Class-E. Using multiple Class E amps in parallel also helps spread the heat dissipation over several MOSFETs, and provides a greater total surface area for conduction to the heatsink.

The 3.8MHz frequency was the self resonant frequency of the HF-SSTC secondary when analysed with a sweep generator feeding the base. However, when the resonator is driven in the HF-SSTC it is driven from a Class E amplifier using link coupling. The magnetic coupling between the resonator and the primary ties up some of the resonators inductance in normal transformer action. This leaves less un-coupled secondary inductance to take part in resonant action therefore the effective resonant frequency rises. In this case it rises to around 4MHz which is the fixed drive frequency for this system.

You are right that corona does lower the resonant frequency of the secondary slightly, however the effects of the tight magnetic coupling dominate here. The detuning in the downward direction due to corona is measureable, but quite small because you don't get a huge discharge at this frequency with only 500W or so.

-Richie,

tritium

Mon Mar 22 2010, 07:06PM

Registered Member #2591 Joined: Fri Jan 01 2010, 11:58AM
Location: netherlands
Posts: 76

I can buy them for 20 euro/ps, I think not so expensive, if you don't kill them to much

question:
the semiconductors are made for the job, so why use standard, the RF mosfets are more capable for this job, or not? maybe I don't understand the class E/F enough, but the RF semiconductor range is not so expensive, so what is the difference?

ragnar

Mon Mar 22 2010, 07:37PM

Registered Member #63 Joined: Thu Feb 09 2006, 06:18AM
Location:
Posts: 1425

tritium wrote ...

the semiconductors are made for the job, so why use standard, the RF mosfets are more capable for this job, or not? maybe I don't understand the class E/F enough, but the RF semiconductor range is not so expensive, so what is the difference?

You are going to be adding shunt capacitance to whichever device you use -- there is little difference between shunting 500pF output capacitance "Coss" with 500pF, or shunting 100pF Coss with 900pF.

The advantage of the RF MOSFETs comes into play for me only when the output capacitance of normal devices already exceeds a usable value. In commercial products with good margins... EMC requirements... paid staff... they're perfectly justified.

Their input capacitance "Ciss" is not dramatically lower yet drive is easier as RF MOSFETs are designed with lower reverse-transfer "Miller" capacitance. I like to pretend to feel better 'challenged' when facing projects with cheaper components.

Cost is more a defining factor than you think. The type of semiconductors you obliterate will be the difference between a month's rent or a day's lunch! That's notwithstanding the fact that RF MOSFETs are hard to source in the first place.

I expect multiple failures during development. Others with proper backgrounds and careful simulation can minimize their 'semiconductor collateral', but Tesla coil resonators remain pretty cranky loads to design for.

GeordieBoy

Mon Mar 22 2010, 10:46PM

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

Another thing that is often an important distinguishing factor between dedicated RF MOSFETs and run-of-the-mill switchmode MOSFETs is the actual gate structure.

Power MOSFETs for switching applications usually have a polysilicon gate which results in the gate-source impedance model having a significant number of ohms of resistance in series with those nF's of gate capacitance. Since MOSFETs are field effect devices it is the voltage across the capacitive part of the input impedance that actually controls the conduction, but it takes real current to charge and discharge this large capacitance millions of times every second. Since this current also flows in the undesirable series resistance it causes power to be dissipated here. The higher the gate series resistance the more RF drive power that is required to slew the gate voltage over the required range at a particular frequency.

For polysilicon gate devices you quickly reach a frequency at which the required gate-drive power (which is all dissipated in the gate structure of the MOSFET) starts to degrade the overall efficiency of the amplifier. Remember that gate-drive power is largely dissipated in the die of the MOSFET, very little actually couples through to the output. (It is even possible to burn out a standard power MOSFET from RF drive power alone, before you even turn on the supply if you drive it too hard or heatsinking isn't up to scratch!)

Dedicated RF MOSFETs usually have a metal gate structure which greatly reduces the effective series resistance. This means that the required gate charging current results in much less power dissipation in the gate structure. Therefore you can get the same output power from the Class E amplifier without needing anywhere near as much RF drive power. This becomes more and more important as the drive frequency increases because the capacitive current into the gate increases with frequency and resistive losses are proportional to I-squared!

This means that metal gate RF MOSFETs typically have a much higher frequency for viable power amplifier operation. They are also often available in mirror-image packages for easier layout of push-pull amplifiers.

If you want to learn more about Class E amplifiers i'd recommend looking at the Rutledge Amateur radio amplifier designs and reading Sokal's explanation of Class E.

-Richie,

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