The AM Forum

I'm new to AM and,  in fact, just appeared on ULS a few days ago (I'm a 'codeless extra').

I plan to build a class e transmitter for 80 meters, any recommendations?
Did some SPICE runs and got LCs of .35uH and .0025uF @250V & 70A.  It's hard to match the switching FET's 10 milliOhm Ron.  Wow!

Tons of info at the links below. I'm sure some of the Class E guys will jump in here soon.


http://www.classeradio.com/

http://classe.monkeypuppet.com/

go to ARRL site select QEX and down load the sample article for Nov 2005. It is actually pretty easy to match them. The hard part is driving the gate but Steve has found some drivers to take care of that.
These parts don't glow in the dark though. They let the light out all at once just before the smoke comes out. Still it is a lot of fun getting 5 KW peak out of a handful of plastic parts.

Thanks for the pointers.

In the meantime, I'm going to crank up an old Swan 500C, unsuppress some carrier and join the 'ghetto' gang.

Hey Guys-This is a repeat of recent post on Steve's site.
____________________________________________________

Hey Folks,
Since I'm retired now,
I finally got off my butt and posted some Class-E design info at

http://www.qsl.net/nu2b/

Look under the "downloads" section.

If anyone finds this of interest or has any trouble with downloading stuff,
Please let me know.
Regards,
BobbyT

Thanks nu2b, the filter software comes in quite handy.

I've been playing with a Spice version for switching power supplies.  I may be able to get a KW with one FET (.01 Ohm Ron and 300 volt).  The only real issue is the rather high current in Pi net which probably requires .005uF caps carry 80A...

vacuuum variable caps handle plenty of RF current.....

Yeah, but I can buy a KW ricebox for the price of those .005uF vacuum caps.  The voltage isn't that high (500V peak) even under no load.  Mica???

Maybe I could construct the two .005uF fixed caps from sheet aluminum...  I think it's around a KV/mm breakdown in air.  Fortunately I only need a few hundred pf variable to tune the 80 meter band and I'll follow it by an antenna tuner to get the load to 50 Ohms.

Wow, a KW on one FET?
What device is capable of that power?
(obviously there are some.... bet they're big bucks though)

We're using 10 FQA11N90 FETs (2 modules of 5) for a max of 500 watt carrier... for 75 and 160 meters.
(QIX's design)
Allowing for a safety margin of course....

Those are 900volt 10.7 amp devices....


For drivers, I'm using the DEIC420 from IXYSRF... they drive 5 FETs (11n90) at 160 and 75 no problem.
There are also IXDD414, they'll drive 2 of those FETs.


Lots of ways to do it!!

Using QIX's design went very easy for me as a first large transmitter project, and works very well.

For this plumbing/sheetmetal project, I'm using a 60NM30 (.045 Ohm on @ 30A).  The very low on resistance is key.

I modelled a circuit without the ferrite xfm'r using only a Pi network.  Lower voltages, but big current.  The ratio of peak FET voltage to output voltage is 1:2 as in your circuit.  It's kind'a like designing a magnetic loop antenna.

The pi net is .005uF, .38uH, .005uF. Any RFC over 10uH will do.  And best of all, the voltages don't go crazy at no load.

Hello AB1GX,
I didn't get the name. But welcome to the AM crowd. It's a wonderful challenge and the Class E Tx is going to be a nice project for you. I bought a complete Class D Tx for 80M from a Ham in Washington State, K7DYY. I tried to put the kit togther, BUT I couldn't see the DA## parts!!!! Sooo I sent the kit back and paid the "upgrade" to assemble the kit.
While you are gathering your parts and thoughts, try to get on the air with your transceiver and a linear or go the Tube route. Lottsa nice receivers and Tx's for sale at the 'Festers and on this forum.
Good luck I hope to work you on AM SOON
Fred KC4MOP

Hi:

I am running Class E at from 125-250 watts on 75m. I am using push pull 3x3 FQA11n90 Fets as per Steve WA1QIX (I have made 400 watt carrier, but do not need to stay there). PDM modulated. Pretty high voltage pulses at the output.

There are other ways of doing it, and am interested to see how your single fet works. Some people are doing Class D and Class H audio.

I believe in push pull RF...the two out of phase pulses add nicely in the output tank circuit.

Please keep us posted!

73
Dan
W1DAN

Hey, welcome to AM !!  Good stuff ! :)

On the class E transmitter - I've found it easiest to keep the impedance above 3 ohms and below about 8 ohms.  The tank components are very reasonable, and the modulators are easier to build at these impedances.  This suggests higher voltage - say, 40 volts at 10 amperes at carrier.  This can EASILY be done with $35.00 worth of MOSFETs (12 of them @ less than $3.00 each) in the RF amp, and $18.00 worth of driver ICs (1 driver per 2 MOSFETs @ around $3.00 each).  4 Multilayer Ceramic caps by ATC for the shunt caps (use the 100C series), and 12 FB-43-1020 cores (at about $1.80 each) make up the output RF transformers.

An inductance of between 4 and 8 uH and an air variable cap of around 500pF for tuning, with a 3 or 4 gang broadcast variable for loading round out the tank.

Throw some transzorbs across the gates and drain busses, just in case; a little overcurrent protection (part of the modulator), and you've got a pretty much indestructable RF amplifier.

All of this is well (hopefully!) documented at the class E web site (http://www.classeradio.org)

Also, feel free to contact me with any questions !

Lots of guys have built class E transmitters, many of these were first homebrew transmitters.

Glad to have you with us !

Regards,

Steve WA1QIX

my final is closer to 2 ohms so the transformer primary is 1/2 copper tubing with 2 turns of #10 teflon covered wire secondary. Low Z is harder to deal with as you see when the cap valuse get high. Also transformers have leakage inductance that will effect tank values.  Lower voltage does offer more safety factor but consider the trade off with caps that need to handle very large currents.

Again, thanks for the info and encouragement!

If I fail obtaining a practical realization of my CAD models, I'll go with one of your tried and true designs. 

And yes my concern is the high current in the caps.  I may not be able to shove enough sheet aluminum into a small enough space to go the air cap route.

I just picked up some .001 Ohm 275A TrenchFETs.  How 'bout a KW on 12V???

For the modulator, I'm using an old 50V 30A switching supply and applying modulation by varying the reference voltage with a 0-50V voice signal.

Glad to be here!

AB1GX, Tom

This Fairchild FET Steve found is hard to beat for performance and price.
I have not blown one yet.

Yeah, but their on-res is .98 Ohms.  I'll need twenty of those babies to replace one low on-res FET.  Those things are more like vacuum tubes!

An update...

So far so good.  A 1/2" copper pipe is around 9" in diameter for resonance at 4 mHz with .005uF caps in the Pi network.  My rf choke is 20 turns of 1/4" copper pipe on a 1" core.  Each .005uF cap is made from 10 500pF caps in parallel to handle the current (maybe???).  No ferrite cores are needed (at this point).  A 100pF variable air cap allows tuning the 75M band.

The only thing bad that happened so far is that I blew out some 200W light bulbs which were the load.

Cool!

Lookin forward to chatting with ya....Class E with an S38!

73
Dan
W1DAN

I went to the fairchild site and was surprised with the number of trench fets. They are low voltage though. I wonder what you are using for a device and who makes it?

For low voltage (which means you need a big impedance step up in your output network), I'm playing with IRF4104s.  They're about .006 Ohm at 70A but are limited to 40V, and they're cheap.  So 500W per FET seems possible in a switching Amp.  Unfortunately, I don't have a SPICE model for these newer 0-5V gate drive FETs.

BTW, I solved a major problem I had with finding 200A 12V power supplies -Now I'm using a single car battery for 13V and four in series for 50V (and of course a trickle charger per battery).

This class e stuff looks like it belongs in a welding shop.

It's looking better for a single FET 1.5KW class e. The key is a FET with on resistance of .050 Ohm (or less) with a peak voltage of 300V.  I built a prototype with 1/10 the power with a 100V FET using a Pi network.  The supply voltage was under 15V.
 
I've been trying to get a configuration that can do 1.5KW with a .01 Ohm on resistance - 100V HexFET, but so far no luck with any transistor I have a SPICE model for.  If I have some time, I'll try to find out what parameter(s) is responsible for this limitation.

BTW, I'd expect the transformer output 1.5KW designs could take advantage of using a single .040 Ohm 300V transistor.  I think you'd just have to use the 1:4 stepup rather than a 1:2 in the multi-ferrite core output transformer.

Quick question - How many dB down do you need to keep the 2nd harmonic?
Anyone use a 2nd harmonic trap like in class F?

Any "Spurs" must be 40db down... for "barefoot" operation.... below 30mhz.
(44.77 db down for 1500watt HF)
(30db for >25watts)


That's one reason for using more than one device in a "Push-Pull" arrangement....
Cuts down on those nasty harmonics.....


As you go along.... maybe you'd post some of this over at the Classs-E forum too!!

Thanks!

In other words, the 2nd harmonic radiated power must be 45 dB down.  It's measured from the antenna, right?

Yeah, the third harmonic is way-way down, but the second is only about 25-30 dB down across the 50 Ohm load resistor in my prototype. 

Watch out for package lead inductance. I'm not sure if its actually in the spice models, but for a TO-247 package its probably around 15 nanoH  Thats an Xl of 0.3 ohms. When your looking at a low impedance load it may become quite significant .  Anyway, give it a go and see what happens. Spice is nice to play with, but real parts and solder are reality.
                                                         
As for harmonics, the Caltec/QST (http://www.its.caltech.edu/%7Emmic/reshpubindex/papers/QST.pdf) 40M class-e uses a 2nd harmonic trap that doubles as part of the loading capacitance

                                                   Ian VK3KRI

Ian,

I wish I had seen that article before I started simulation.   It would have saved me some time.

What I am doing different is driving the FET with a 0-10V square wave using bipolar transistors.

Also, I'm using homebrew fixed air caps for high-Q at very large RF currents.  They are each 4"x4" and about an inch thick.

And I appear to be using much lower-inductance/higher-capacitance in the output network,

Yeah, the inductance of the leads do matter quite a bit so I made the FET itself part of the tank circuit.

Thanks,

Tom 

Thanks again for that article!

The 2nd harmonic trap not only lowered the problem spur but also cleaned up the waveforms in the class e amp.  Push-pull, I'm sure, would have done the same, but now I have a simple one-transistor 1.5KW class e design.

The only thing tricky is that two caps (.005uF) have RF peak currents of 80A and building air caps is a pain!

How's Mica?  MultisectionCeramic?  Anyone know?

30 dB down from 1500 watts is 1.5W.

I've been curious....

1.5kw at carrier??

Or capable of 1.5kw with modulation?


building air caps is probably easier than paying for
something to handle that current.....

1.5kW CW, which would also be the peak for AM.  Legal power is 1.5kW PEP, so an AM carrier would be less than half of that.

Regarding the high current caps, 1 mil teflon could reduce the size of the caps by 1/2.  I have no clue where I could buy this dialectric.

Tom

Right, my second harmonic is down 45 dB with the change, so it will be under 100mW (at the amp output) at l500W.

Ok, was just surious if you were designing it as 1.5kw continuous, or for the 375watt /100% mod. thing....


Don't know about teflon.... but Kapton is fairly cheap on ebay....
I think I paid $9. for 3 sq/ft.

That's 2 mil... (other sizes avail... of course)
But, now that I think of it,  that may be too thick to get .005uf easily
You would need 1 mil... still be pretty large surface area..

Maybe Frank GFZ will chime in on that, I know he's used it...

Kapton is 7000 volts per mil. I use 3 mil myself and have never flashed through it. I did lap the mating surfaces with wet dry sand paper to remove any burs that could cut through it. Also hit all the holes in the heat sink with a over sized drill bit to remove any sharp edges around the holes before and after tapping them. I used 3 by 5 inch copper spreaders to act as one plate of the cap and a common drain connection. Check out my QEX Nov 2005 article. Pictures tell a lot more that words. You can down load it as the sample article for free or let me know and I will send you one if you provide an email address.
My mistake was it was the right size for 40 meters but not big enough for 80 or 160 forcing me to add door knobs to get enough shunt C.
I used the simple cap formula in the handbook using 4.2 as the dielectric constant for Kapton. This formula will tell you the total area needed for the cap plates.  My configuration used the heat spreader as a low L connection to the transformer, a way to balance the temperature of the parallel fets and the shunt cap plate working against the heat sink. gfz

Frank,

Nice article!  Kapton looks like a great material.  20 sq in gives me the .005uF.

Tom

once I subtracted for all the holes I needed to add more C to get back to 5000 pf. Each FET has a nylon insulator to isolate the screw from the heat spreader. This hole was about 1/4 inch. Then the void for the transformer primary tubing. I ended up adding another 2000 pf to my rig to get the shunt so I could do 1200 watts at about 47 volts.