One more question: Would it be possible to use a class-D amp (like the Wondom 100w or some of those TPA3116's) as a modulator for the output stage of the Commando (also from Dave Martin, schematic below)? The Commando is only class AB but seems easier to build than let's say the U-LuLu.

Edit: Or would this only work effectively with the LuLu, as it's class-E?

You would have to run the modulator from 24V. Just connect the +ve speaker output to the top of L4, omitting the mod transformer. With no audio, the + output of the amplifier will sit at 12V. When you apply audio, the voltage at that point will vary. Forget the variable power bit to the drive stage - run it at full power all the time. Make sure that the 12V supply rail for the oscillator and driver is regulated. You might have to apply a little forward DC bias to the gate of the output FET, to push it towards conduction. You should probably get around 6 - 7 Watts of unmodulated carrier, peaking at around ~25 Watts on mod peaks. Quite a lot of the smaller shortwave operators run this kind of power level. If you have a reasonable aerial and earth, you can get substantial coverage from this kind of power level.

Make sure that your audio bandwidth is limited to 6 kHz maximum, and compress the audio as much as you can to maximise the "talk power" as the Leftpondians like to say.

The 3116 is by far the best audio amplifier, i’ve had them all, suppose your using audio amplifiers because its hard to make amplifiers ones that don’t just generate heat

https://www.ebay.co.uk/itm/334520754836 is the amplifier I've used as a modulator a few times. I put a simple audio compressor / limiter and audio filtering before the amplifier module, and some of the rigs have had MP3 players built-in too.

The rig that I've been building a few of uses a five-chip CMOS synthesiser (using 74HCT logic to get the speed) with the 74HC4046 oscillator running at twice the output frequency feeding into a 74HC74 to divide the oscillator by two and give an accurate 1 : 1 M/S drive. I've used a 74ACT240 with all the gates paralleled to drive the gates of the Class E output FETs. The basic version develops 25 Watt carrier / 100 W PEP.

When I'm happy with the design resilience and make fully certain that it's easy to reproduce, I'll put Gerbers and circuit diagrams up here for all to use. It's designed (at the moment) for 5.8 - 7 MHz in 5 kHz steps, but the design principles will be easily applied to other frequencies up to a maximum of about 15 MHz. With further work, it'll be possible to increase the power a lot!

I use the same method as any class D amp will make a very efficient PWM modulator.

Why are you lot obsessed with mod trannies?

They not used anymore and haven't been for years.

For (upto) a 20W carrier the (Texas chip) class D amps are fine.

For any more power you need a simple half bridge and butty filter. There's also a single chip solution for audio to pwm. So a couple of chips 2 Fets, a handfull of bits and some L n C's and you got yourself a modulator that will mod a 300W carrier.

Weighs nothing, 95+% efficient and flat audio response.

You dont get any if that with a mod tranny!

Pic shows modulator and integrated filter.

stretchyman wrote: ↑Fri Mar 15, 2024 11:31 amWhy are you lot obsessed with mod trannies?

Everyone's obsessed with mod trannies - I've decided it must be some sort of rule...
Nice pic. What's in the black box? DC-DC?

Yes 18-72V in 12V out. Rather useful.

This is the hest circuit to use, lot of really good fets around now, GaN being the best. I used IRFB4227 here but GaN on the latest design.

stretchyman wrote: ↑Fri Mar 15, 2024 5:15 pmThis is the hest circuit to use, lot of really good fets around now, GaN being the best. I used IRFB4227 here but GaN on the latest design.

I agree, but why do you need to use GaN for the PWM FETs? Is there anything to gain?

shuffy wrote: ↑Sat Mar 16, 2024 2:37 pm I agree, but why do you need to use GaN for the PWM FETs? Is there anything to gain?

You don't have to but they're a bit more efficient if you're trying to squeeze out every last watt. And easier to drive although the IR2110 in stretchy's pic will drive normal silicon fets just fine

The GaN's are leadless and fit under the PCB, cooled by a Cu strip.
VERY easy to drive but I guess a bit OTT.
We're dealing with 100's of KHz here and I can drive them to 20MHz!

PWM frequencies are so low virtually any old thing will do, you just have to watch out for 'ringing' (oscillations on the gate).

There's also infinitely better PWM half bridge drivers other than the good ol' IR2110, however (again) fairly pointless as this is a very undemanding application. I've used the devices for Skyworks (Sixxxxx ex Silabs) but they keep going obsolete as they keep making a newer better one.
The IR2110 is pefectly fine (been around for years and even the cheap clones work) unless you want to drive 6V (Vg), 100V(Vd) GaNs.