Modulators - Class H vs PWM?

[KQ6F]

Am trying to decide which type of modulator to go with my newly-constructed CREE RF deck.  Have built up small-scale versions of both a series-pass Class H and PWM.  Both seem to do a good job with a subtle edge in fidelity going to the Class H.  But the Class H is horribly inefficient especially when scaled up to produce 100% modulation of a 375-watt deck.  Seems like a shame to marry such a beast with a 90% (ish) efficient deck.  OTOH, the LPF for the PWM seems problematic - especially the first inductor that probably needs to be air-wound to circumvent core saturation problems.

What to do...what to do  

[w1vtp]

It's clear to me: go PWM.  That's what I'm using and everything is running quite cool.

Al

[w4bfs]

there have been several posts on the PWM filter ... single layer solenoidal and large

[VE3ELQ]

 OTOH, the LPF for the PWM seems problematic - especially the first inductor that probably needs to be air-wound to circumvent core saturation problems.

What to do...what to do  

My 1st inductor is a 120uhy solenoid 4 layer air core #14 insulated solid, low profile, works FB driving 350W decks. PW mod is the way to go IMO.
Pic attached showing a 3 layer 125uhy, the 4 layer 120uhy and the corresponding second inductor a ferrite core toroid.
73s  Nigel

edit: seems the pics are not getting through correctly.

[steve_qix]

There is absolutely no doubt that over a hundred watts, PWM is much better.  The input inductor isn't THAT big - the one in my 450W power input transmitter is maybe 5 inches long (maybe smaller, actually) and around 1.5 inches O.D., single layer.   

You *can* use a toroid as the input inductor, and I show this on the class E web site.  That modulator is a 45V 9A modulator (about 400 watts power input to the RF amplifier).  The two cores in the filter are the largest and highest flux units available - I forget if that's Mega-Flux or Hi-Flux.  I do have the part number, and I think it's on the web site in the Construction Projects -> PWM modulator and power supply section.

The input inductor gets slightly (not very) warm.   The 2nd inductor is cold.

Regards, Steve

[KQ6F]

Steve - thanks for the good advice.  Will go with PWM.

Am looking at the latest high-density flux cores from Micrometals Arnold called Sendust.  They appear to have roughly half the core loss as the Hi-Flux ones. 

73, Rod

[KF1Z]

My PWM has both the inductors on HiFlux toroid cores ( CH777060 )

For a 400 watt RF deck these are just fine.
Many have been built using these cores.

[steve_qix]

Steve - thanks for the good advice.  Will go with PWM.

Am looking at the latest high-density flux cores from Micrometals Arnold called Sendust.  They appear to have roughly half the core loss as the Hi-Flux ones. 

73, Rod

The challenge when designing PWM filters using cores is DC bias (current).  This has to be calculated at 3x the carrier DC current to account for the high positive modulation capabilities of the transmitter.  The most important characteristic of the core, assuming it is large enough to deal with the current (which is 3x the carrier DC current), is the stability of the permeability over the current range from near 0A to 3x the carrier DC current.  The Hi-Flux cores excelled in this area - they were the most stable over a wide range of DC bias values.

I don't believe the sendust cores were nearly as stable permeability wise as Hi-Flux, but it's been a while since I've looked.  I used X-Flux cores in my most recent 1kW pulse width modulator, but the input inductor is an air core coil.  I was unable to design a suitable core-based input inductor that would handle the DC current (of the 1kW rig) _AND_ handle the flux excursions that take place due to the fact that this inductor is subjected to the full voltage swing at the switching frequency.

[KQ6F]

Steve -  I think I'll try both the Hi-Flux and Sendust cores.  Or, what the heck - just go for the air-wound solenoid as the first inductor.  I calculate that I need 60uH which would be a pretty big solenoid.  Am trying to squeeze everything including a large toroidal power transformer into a 2U 19" rack.

Making the solenoid multi-layer might be an option.  That's what Nigel is doing.  BTW, Nigel - can you email me a picture of yours.  This site seems to be currently choking on .jpg pictures...

Rod

[steve_qix]

Steve -  I think I'll try both the Hi-Flux and Sendust cores.  Or, what the heck - just go for the air-wound solenoid as the first inductor.  I calculate that I need 60uH which would be a pretty big solenoid.  Am trying to squeeze everything including a large toroidal power transformer into a 2U 19" rack.

Making the solenoid multi-layer might be an option.  That's what Nigel is doing.  BTW, Nigel - can you email me a picture of yours.  This site seems to be currently choking on .jpg pictures...

Rod

60uH for the input inductor seems somewhat big.  What is your filter corner frequency, and what is the DC resistance of your RF amplifier at normal operating power?  With a 12.5kHz corner and 45V @ 9A, I think I used something like 40uH which is still a lot of inductance.

[KQ6F]

Actually the number came out 57.7uH.  It was derived using a modified Butterworth design for a load Z of 3.3 ohms and 11kHz corner frequency.  The modification was done using the Elsie simulator from tonnesoftware.com.  Jim has an excellent paper on his website on the design of PWM LPFs.  I followed his rules and tweaked the values so as to obtain a constant Z throughout the passband.

That was earlier.  With the latest updates to the RF deck my load Z is now 2.95 ohms.  I'll redo the LPF which will reduce the first inductance value..

Rod KQ6F

[VE3ELQ]

Making the solenoid multi-layer might be an option.  That's what Nigel is doing.  BTW, Nigel - can you email me a picture of yours.  This site seems to be currently choking on .jpg pictures...

Rod

Pic Emailed as requested. 
I'm running my decks at 70V @5A, a higher 14 ohms impedance to take advantage of the 1200V FET spec to get better efficiency. So my 1st inductor is a little higher value but still a pretty small profile using 4 layers, you could go 5 or 6 if you want to. Yours would be half the size of mine if air core and very inexpensive. Just another option.
Nigel