Push-pull output transformer for MOSFETs

[ka1tdq]

The attached picture is from the class E website.  I understand how to construct it as Steve describes, but how would the secondary be constructed if I wanted more than one turn for impedance transformation?

I'm thinking about another 40 meter AM transmitter digitally driven from a DDS VFO and a JK flip flop to divide the frequency in half and have out-of-phase inputs for the gates. 

The output will be class C but I need to match the 50 ohm antenna jack to the low FET drain impedance. 

I'm thinking that I would need to put the necessary number of turns for the secondary (say 4 turns, for example) on each side of the transformer and then connect them in the center.  So, the left set of cores would have 4 turns and the right set of cores would have 4 turns.  Then solder the center two leads.

Jon
KA1TDQ

[steve_qix]

That is single ended push pull wiring.  Single ended push pull is easier to configure and is more forgiving of "overlap" - when both sides are turned on at the same time - than is traditional push pull.  It is also easier to build.

That being said - any overlap will pretty much cause the RF amplifier to be non-functional (or extremely inefficient).  MOSFETs have a turn-off delay which is typically longer than the turn-on delay.  This, in and of itself causes overlap.  Than you have the problem that most drivers modify the timing of the input signal.  And, typically, the turn-off is lengthened.

So, particularly on 40 meters (but on 80 meters as well), it is very important to be able to control the "on" time of the MOSFETs.  Otherwise, there will be overlap.  I do it with an AND gate, with one input being the "raw" signal from the divider, and the other being a differentiated variant of the divider's output. Adjust the time of the differentiated signal by varying the resistance the differentiating capacitor looks into.  I think there is a schematic of this somewhere on the class E web site.  It can also be done with a couple of high speed comparators using the same underlying principal.

Thought I should mention this.

Regards,  Steve

[ka1tdq]

Ok, so it wouldn't be as simple as slapping a JK onto a DDS VFO... good to know. 

Yeah, I've seen your VFO design using the AND gate for duty cycle adjustment.  I'll just copy that rather than trying to reinvent the wheel. 

I'm actually in between two ideas for my next project.  My current FET 40 meter rig is low power and needs a linear.  The linear I'm using now uses a pair of 3-500's, but I'd like to get off of tubes completely.  So, I could build a higher powered 40 meter AM rig or build a solid state linear.

Communication Concepts offers amplifier kits, and I thought about putting 4 of these in parallel:

http://www.communication-concepts.com/eb104-1/

And using this as a splitter/combiner:

http://www.communication-concepts.com/psc-4h5/

All parts (minus the power supply) come to $2000 and it would be a much more capable linear than what's currently on the market.

Jon
KA1TDQ

[steve_qix]

You can build a high power 40 meter transmitter using only MOSFETs.  I've put various transmitters on 40 meters.  Currently I have a 16 MOSFET (16 FQA11N90s) on 40 meters running a KW and it works just great!  This is a pretty old transmitter, too - vintage 2001, which is old for class E rigs.

[ka1tdq]

Could you draw up an analog drive schematic for your 8 FET class E rig that you have posted on the E website?  From the gates back... For simplicity, I'd like to drive it from a ricebox. 

From there I can cookbook the rest and make small modifications for 40 meters.

Jon
KA1TDQ

[steve_qix]

I have never made any complete schematics of analog drive transmitters.  The idea is, you have an input (driver) transformer with a step down of from 4:1 to 6:1.  You match the primary to the driver (in your case a rice box) using some kind of a tuned network.  That's really all there is to it.

Here are some schematics showing the connections between the various components.

[ka1tdq]

I'm sure I'd have to go through lots of experimentation to get an analog drive to cooperate with an 8 FET rig.  And I really don't have too much time anymore since my wife is pregnant now with #2... Booyah!  

A VK3 ham sent me a schematic for a 100 watt carrier class E rig with analog drive.  Everything is there from the gate bias/drive to the output component values.  Now I just need to start collecting parts.

I just picked up two massive heatsinks at the last hamfest and a nice, new commercially made output coil.  I almost hated to open the box.  It had been sealed since the 1960's and I picked up 4 of them.

I'll worry about modulation later, but PWM is probably the way to go.  Right now I need get plans ready for some weldin'.

Jon
KA1TDQ

[steve_qix]

It would take too much experimentation, I think!  The matching network is just an L network, so not too bad.

Regards,   Steve

[ka1tdq]

Maybe something like this?  Using four BN-43-202 cores for 8 FETs (FETs are paralleled).  Primary uses 5 turns and secondary is 1 turn.

Center conductor of the SO-239 is connected to the cores as shown in the schematic.  Transformer phasing is noted.  

An LDG auto-tuner would make the rice box happy.

Jon

[steve_qix]

Personally, I would not run the primaries in series.  Could result in uneven drive between MOSFETs or groups (modules) of MOSFETs.

Also, it's not a good idea to break up the drive to the MOSFET gates within a single module (if that's what is shown - not sure what's happening with the drains).  Better to use a single transformer for 4 MOSFETs.

The reason is, there will be phase shift between the driver transformers.  It just happens, and can't be eliminated.  If you have a single transformer per module, you can be assured that at least within a single module (of, say, 4 MOSFETs), the drive to the gates is the same, and the gates are all in phase with each other.

You can the balance the phase between the two modules by observing the drain current.  The phase can be varied easily by squeezing or separating the turns on the primaries of the driver transformers.  You can actually achieve pretty good balance of drain current if you're careful.

Regards,  Steve

[ka1tdq]

So if I used two cores (say BN-43-7051's) and used a 6:1 turns ratio using 4 FETs in parallel per core, that should work? (Basically duplicating the post of the push-pull diagram you posted earlier while observing proper phasing).

Then just copy your class E transmitter (drains to output) while making slight adjustments to the output network for 40 meters.

Jon

[steve_qix]

That should work.  1 core *should* handle 4 MOSFETs in a driver capacity.

Hook your oscilloscope to the gates and you should have 30V P-P (+/- 15V Peak)

[ka1tdq]

Ok. This'll be my next project. I'll putter at this and post more once I've made some notable progress.  I actually have my shack the way I want it right now with a quality sounding AM signal on 40 and a home brew CW rig. A class E rig is just gravy, so I'm in no real hurry.

But, I've done everything else so this is the next step.

Jon

[PD0RTT]

This might be help:

http://ludens.cl/Electron/mosfetamps/amps.html

[ka1tdq]

Thank you, that's some good reading.  It's before 6am and too early to digest right now, but I will read it. 

I just picked up some copper sheet that I can use for the source, drain and gate busses.  I'm going to mount the IRF11N90's on the heatsink, solder the busses (transorbs and all) and then install the gate input network.  I figure this will all be minimal cost to ensure that I can get the +/- 15 volts before making a heavier investment in the steel and output network. 

I've made a picture of  my parts so far.  I do have the LDG tuner for the input.  Next I'll order the output cores and the 2 binocular cores for the input.  Then get some FETs and Transzorbs...

Jon
KA1TDQ