Class H modulator with a single Class C FET

[ka1tdq]

Will this work?

I'm thinking about building a single FET class C transmitter coupled with Steve's class H analog modulator.  I've calculated an output impedance of 48 ohms using 60 volts carrier voltage.  120 volts would be applied to the drain bus of the modulator.  Power output would be 72 watts carrier.

Using the schematic on the Class E website, Steve has 6 FETs/transistors on the drain/collector buses.  Since this rig is only 72 watts, I'm thinking I could get by with 2 on each bus.

I want make a high fidelity low powered transmitter that I can also take mobile.  This scematic that I've posted here would work perfectly in an old locomotive radio chassis that a friend just gave me.

25 VAC voltages for the modulator could also be pulled off of the power supply.

Jon
KA1TDQ

[WBear2GCR]

John,

It's unclear to me why you specify multiple small transformers in series to make ur voltage? I think it will be bigger and heavier than a single toroid of the appropriate size, and much much bigger and heavier than a switching supply...

Also the breakdown voltages in the small xfmrs might come into play.

For "72" watts at 60 volts you need just a little more than 1 amp, assuming 50% efficiency, then 2 amps... not much. A toroid that size is little.

I'm not comfortable with a full wave bridge set up that way to make 0 - 60 - 120 vdc... it doesn't look right to me. Also the taps for the other voltages sitting way above the ground point of the transformer string looks like trouble to me, unless they float and have no ground reference.

Perhaps you might consider a simple "E" class output stage, less heat...
Getting rid of 70 watts of heat is non-trivial, that's a lot of heat.
I can't recall if class C does better than 50%, even 35 watts of heat is not inconsiderable, takes a pretty decent sized heatsink and/or a fan.

You will need high voltage Mosfets to do that job, and some protection stuff too... something to keep the gates safe, etc... Looks like ur planning on using that driver chip to do the bias, is it designed to do that? Or do you need/want fixed bias at the gate?

                              _-_-bear

[ka1tdq]

Actually, I was thinking that class C was around 70%. But, if 50 is the number, then yeah, that's a lot of heat to dissipate. Better than tubes though...

The mosfet for the final is from the E website. Some crazy number of 900 volts on the datasheet. Seems odd, but I guess the number is right.

Class E rf decks confuse me and C seems much easier. And I get the analog class H modulator.

Your suggestion of the toroid power transformer makes more sense. I hadn't thought of that. I want to fit the power supply in a box about 3.5 inches tall and 10x 12 inches.

As for the gate bias, I'm not too sure. I think I need to look that up.

[kb3ouk]

I used to think class E was confusing too until I really studied a schematic enough to realize that they're not really that complicated. http://classeradio.com/class_e_topologies.gif shows 3 class E output circuits.

[ka1tdq]

I guess it wouldn't take too much to make a few changes for a class E single mosfet RF deck using the H modulator.  Using the deck that's on the class E website, I could use a single XXDD616 on the final FET.  I could also bare-bones the VFO schematic that's on the site (west coast AM uses different frequencies than the crystals I have).  Plus, there is still room in the box for a VFO.

I'll post pictures of the railroad radio chassis that I have too.  It's really neat!  A bottom chassis for a power supply and a top one for the RF deck.

My work schedule is hectic for the next few days but I'll think about the plans and post a schematic.

Jon
KA1TDQ

[WBear2GCR]

I think an "E" rig is harder slightly to make work.

No reason not to try the Class C version at least first.
Breadboard.
See what it does.

The FETs that QIX specs are cheap and ought to work fine afaik.

Just my thoughts...

                      _-_-bear

[steve_qix]

Class E is very forgiving which is why it is so easy to make it work.  Primarily, that's why I use it.

[ka1tdq]

I'd like to make sure I understand how a Class C mosfet final works.  I've added a photo showing the basic stuff.

Biasing using a mosfet driver such as a TC4421 would need a dc blocking capacitor.  Then, depending on the voltage swings going to the gate, the Rgate value would depend on mosfet load line for the given conditions.

Without getting into heavy math, could you simply start out with putting a variable resistor in the gate and then slowly increasing value from zero (Vgs=0 and Id=0) to whatever value is needed to get to your calculated power output?

This would assume sufficient drive was available too.

After the RF is flowing, then just keep adding audio until you get to 100%-ish modulation.

Jon

[kb3ouk]

That's pretty much the exact same design I came up with one day when I was thinking of homebrew solid state ideas.

[W4NEQ]

Normally for class C, D, and E you want to keep the gate saturated with drive to avoid spending much time in the heat producing transitions.  Power output is adjusted then by the amount of drain voltage - and, given enough dissipation capacity in the modulator, both the DC setpoint and amount of modulation is controlled by the modulator.  Always good to start low

Chris

[steve_qix]

Do not bias the MOSFET at all.  This is not a good idea, and can give rise to parasitics and other problems.

DC couple the MOSFET gate to the driver, so the driver can pull the gate to 0V when off, and +12V when on.  The rise and fall times of the driving waveform should be as fast (short) as possible.  The driving waveform, ideally, should not be symmetrical - meaning the "on" time should be somewhat smaller than the "off" time.   NEVER the other way around.

[ka1tdq]

Ok, I think I see.

This should just be slammed on and slammed off roughly 50/50 with the "on-time" being slightly less than 50%.  I had a misconception before that this should be linear, Class A or something.  That would make LOTS of heat.

I've attached another photo showing more basic stuff.  So if I used a carrier voltage of 45 and had a target impedance of 50 ohms, that would give me 40.5 watts carrier.

Also, I guess that the RF choke should be around over 50 ohms?  Different than tubes... Everyone is using ferrite cores when a toroid would be more in-tune with the 2-ish uH of inductance required.  Using a ferrite core gives you one turn only. 

In summary:  DC driving the FET with the 50/50 cycle (experiment with the driver to develop this waveform).  Slap in a 50 ohm choke, output filter, modulate the DC, and... you're done???

Jon

[W4NEQ]

I think you're on the right track, but - two points;  a better approximation of the load impedance during normal operation is Vcc^2 / 2PO   ... at least for class C.   Also, the choke needs to have a large enough impedance at the lowest frequency so as not to significantly shunt the output.  This is typically considered to be 10 times the operating impedance.

Start with low Vcc and SLOWLY raise it.

Chris

[ka1tdq]

Okay. So 45 volts carrier would match 50 ohms nicely at around 20 watts.  That's good enough for a mobile unit and could be used with a linear at home. 

The choke makes more sense at 10x the load impedance.

I'm going to start chugging away at this. I'm thinking that I'll use a 50 ohm input jack for the low level input to the gate driver.  This will give me more flexibility if I install this in the car.  The transmitter could be mounted under the hidden panel in the back and then a small frequency synthesizer by the driver's seat. This would make room for the R390 (kidding).

Thanks for all the help!

Jon


 

[steve_qix]

Oh, one more thing about the output..

The main reason for using an RF transformer in the output rather than a choke/capacitor combination is that you eliminate 1 critical component, that being the DC blocking capacitor.

You have to provide the choke anyway, so why not make it a transformer and just take the RF from the secondary - blocking capacitor gone!  Also makes it easier to combine stages.

If you are not running a lot of current, the transformer can have more than one turn in the primary and secondary.  The FB-43-1020 is a very good core for this type of application.  It's big enough not to saturate with 1 turn (using 6 or 8 cores) at 8 amperes of DC current, modulated to 200% positive.  So, with much less current, you can cut down the number of cores and use more than one turn in the primary and secondary.

[kb3ouk]

Steve,
If he builds it like you described, then if he wanted to convert it to class E, wouldn't the only change be to add a shunt cap between the RF transformer and FET the output network?

[ka1tdq]

Ok, I'm sitting in the car and don't have access to reference materials or a calculator. So if I use 45 volts carrier and calculate an output impedance of 50 ohms for whatever value that turns out to be, what Xl do I use for a single FB core that Steve mentioned for the output transformer?

And I guess to make it Class E I would need to do a few things differently on the output network. I haven't read that part of the Class E website in detail yet so I'm guessing the formulas and stuff are there.

Jon
(iPhone typing so forgive any grammar errors)

[KF1Z]

Stu, AB2EZ has a writeup on his single fet E-rig

http://mysite.verizon.net/sdp2/id11.html

[ka1tdq]

Yeah, that's an awesome rig! He's driving it directly with 5 watts but I'll use a FET driver. The rest I can pretty much copy. 

The series 250-ish pf capacitor in the output may be too large for the enclosure I want to use.  Does it need to be rated for 4500 volts?  I've attached a couple photos of the enclosure that I want to use, and it's an old train locomotive radio chassis.  The bottom half will house the power supply (3" tall) and the top will hold the modulator board and RF stuff (2.25" tall).  If need be, I could put the output network in the bottom half of the chassis.

There's a handle on the front too for portability.

I like the bottom part of his write-up where he shows a power divider for a linear amp.  And you can adjust the power output to the amp by the turns ratio.  Cookbook is good.

Jon