Possible "Class H" like SS HF linear amp implementation idea??

[WBear2GCR]

Ok, so I was thinking, and I am not too clear on this, but somehow I think it can be made to work.

Here's what I have in mind.

I have a set of 4 solid state linear boards based on the MRF 150 output device. Afaik they run in class A. Which is fine. I can combine them all and get about 300 watts max from each board, which would give about 400watts of carrier and 1200 watts peak, maybe a little less in practice.

But all 4 are going to cook since they are class A.

So, I was thinking, if I could "stack" them, so that the second two (for example) ran only on those nice peaks, saving powah, and the main two ran the carrier and NEGATIVE modulation, still nicely in Class A, that would be neat.

There is some transition between the two levels, but presumably that could or might be mitigated (if needed in the first place) by the use of RF negative feedback.

It was suggested by QFX that RF negative feetback will improve the IMD substantially, so I was going to try that anyhow.

Any gurus here think that this is workable?
Then the question is by what means should this be accomplished?

I can think of two basic ideas.

The first is a straight "stacking" like the Class H audio modulators use, so unless the voltage input exceeds the first rail, the upper rail does not amplify. The problem with this is that the basic amplifier board is pure Class A, and we don't want that, we want to save power. So maybe that looks like Class B on the upper ones, Class A on the lower??

The boards invert one half of the input signal so that the output can be PP... afaik that is...

The other option, which is a bit foggy in my mind, is to do something at low level so that there is essentially clipping/limiting (at low level RF) so that the lower level can not exceed max but can downward modulate, and then put a reverse circuit (in essence), some sort of bias I'd expect, on the upper so that it is only sent signal above the threshold... presumably this could be made a bit "soft" so that there are not hard and sharp artifacts.

In this situation, the outputs of all 4 modules would sum...

In the previous one, I think the outputs of each 2 modules would be in a series config, and then summed... not quite clear on this.

Maybe I am too confused on this... probably.  but anyone who has a roadmap here would be appreciated...

The goal is simply to reduce the constant power draw. Rebiasing a set of modules ought to be pretty straight forward.

These modules are somewhat similar to the EB-104 app note. Although not exactly so.

                     _-_-bear

PS. to make it clear, by doing this, the lower modules could still run 200watts each for 400 watts (maybe more) of "carrier" but never be over driven, so 400 watts of carrier is still available. The 200watt balance could be either used for some positive modulation or not, depending on what works best. I was speculating on virtually NO positive modulation on the lower modules, leaving the positive for the upper modules - so a pair of upper modules would make the positive peaks in this scheme... presumably.

PPS. the other thought is that in class B or AB, compared to a Class A PP amp, one gets much more monkey swing (more voltage swing).

[kb3ouk]

Sounds like you are trying to build a Doherty amp but instead of using tubes for the carrier and peak amplifiers, you are using the SS amp modules.

[WBear2GCR]

The Doherty amp seems not an option, since it seems to be rather narrow band, and requires pesky delay lines.

I was hoping to come up with a method that did not require the transmission lines for matching... I'll have to read some to see what and why those 90deg delay lines are required to make it work...

If it is unavoidable, then it's back to simple and hot.

                   _-_-

[KK4YY]

Bear,

The MRF150 datasheet specs the class A output at 50W PEP per device. So that's 100W PEP per module, 400W PEP total or 100W carrier. In class AB 150W PEP per device, 300W PEP per module, 1200W PEP total or 300W carrier.

IMD in AB @ 150W is -32. In class A @ 50W it's -50.

Communication Concepts, Inc sells combiners if you want to put them together in the usual fasion.
http://www.communication-concepts.com/index.php/splitter-combiners/splitter-combiner-psc-4l.html

Don

[KK4YY]

Wait a minute...
I assumed 2 devices per module. You said the modules were like the EB104. They have 4 devices per module so the above numbers would double.

So using regular combining methods...

Class A 200W carrier @ -50 IMD and lots of heat.
Class AB 600W carrier @ -32 IMD and low heat.

Or if you want to run the so called legal limit lower the drive...
Class AB 375W carrier and about -40 IMD and low heat. Not bad.

These IMD numbers are from the datasheet at 30 MHz. YMMV.

Of course this whole set up would give you 2400W PEP on SSB should the need ever arise.

Don

[WBear2GCR]

Don,

These modules must be running the output in AB, and the input drivers in A... They're not EB-104, they are actually MRI modules. Frank, WA1GFZ (hope I have his call right) is running them currently. I think he quoted the 300W/module figure. He was kind enough to supply me with schematics and some mod notes, otherwise I'd likely have steered clear of them, as I don't really have the time to figure it out from scratch.

I'm presently a bit sleep deprived, and my brain feels like a mushy bowl of oatmeal, so whatever weak skills I normally have are dulled considerably. I need to look more closely at the schematic and see what is what, as well as try to figure out why the Doherty requires that phase shift to work.

One of the differences between Doherty and Class H (at first glance - I may not have read enough yet) is that the Peak amp in the Doherty sits on ground, whereas in the Class H, the upper rail sits on the lower rails B+. Potentially the upper could also be bootstrapped (modulated) on the lower amps signal...

Thanks for the info.

                            _-_-bear

[KK4YY]

Bear,

MRI modules, eh? I don't know what freq they run at. HF I suppose. With 2 stages you'll be driving them with milliwatts on 75M. Prolly about 50dB of gain there. Good opportunity to drive them with a clean low level signal.

Even with conventional combining you've got a good thing. At 375W there's plenty of headroom for positive peaks and the opportunity to play with NFB to clean it up even more. With copper heat spreaders and big heat sinks you might get away without a fan and go silent running.

I'll have to plead ignorance of class H. If it's viable I guess it might be used on BC transmitters so I'd look there to see what the big boys have done with it.

Is it your goal to experiment with class H or is that just the means to an end?

Don

PS - I'm constantly sleep deprived and my porridge filled brain would be rejected by Goldilocks even if she were starving. (Hmm. Nice. I snuck in a "bear" story!)

[WBear2GCR]

They have three stages, iirc...

by 375 watts I think you mean with 4 modules??

yep copper spreaders are the idea, if they don't have them now.
But I like larger blowers run slow to move a lot of air and work into a backpressure. So that is the tentative plan as of now...

My goal is to not experiment if I can avoid it. But the idea of running an extra few hundred extra watts in heat makes me want to try to drop that down, and not get nuts doing it.

You can see a class H audio modulator on Steve, WA1QIX's site for class E...

Goldilocks came by, but I was out at the time...

                 _-_-bear

[WBear2GCR]

It looks like this... higher res pix later...

half of it gets jettisoned as it is control stuff for MRI, keep the side of the board that has the RF transistors...

[KK4YY]

Bear,

Yes, 375W with all 4 modules combined should be running just about half power capability. That would give plenty of headroom by my calculation and would be in the devices (MRF150) lower IMD area. Best I can figure from the datasheet would be around -40 which ain't bad. Then NFB could be something to play with for even lower IMD. Fire one module up and make some measurements before you dive in. Check that all are working.

I don't know how much power can be saved by going class H. You'd have to weigh the power savings against the added complexity. Seams it's may be practical at very high power levels under CCS use but for the power level and amount of key-down time in amateur use maybe not-so-much. Still, it might be fun to play with if you had a mind to.

I don't see very much heat sink on the pics. Maybe there was a low duty cycle in the MRI or they used a lot of air? Looks like 3 stages, alright. Tiny amount of drive from a ricebox should be enough. Like from the transvertor port, maybe even padded down!

Quite a few guys have built the EB104 kits and posted their construction on the web. Check those out too.

Don

[KK4YY]

Bear,

Looking again at your pictures... there are only 2 devices in final stage. NOT like the EB104. We're down to half the numbers again as in Reply #3.

Don

[KK4YY]

Okay, so it looks like there's a total of 8 MRF150's. 8x150=1200. 1200W PEP, 300W carrier, -32 IMD. Similar in performance to an Ameritron ALS-1300 (which sells for $2899) but with very very high gain.. Run at say 200W carrier there would be good headroom and maybe -40 IMD (from datasheet @ 30Mhz).

You'll need to add a LPF for harmonics suppression, a splitter/combiner, and of course a heavy PS (or 2 or 4 smaller ones). If there's no frequency compensation in those amps they could have a TON of gain at 75M perhaps 65+ dB considering there's 3 stages. So, well under a milliwatt of drive and careful layout of splitter/combiner to prevent oscillation.

Don

[WBear2GCR]

I also have a pair of EB-104 kits...
Have to look and see if they are 2 device or 4 device implementations, I did not look carefully.

Got those before I had enough of these modules to do full snot.

I was thinking I would skip the first stage probably... Or use it for some other function.
Too much gain. WAY too much gain.

Ok for checking it out with a signal generator though!

I have iron that will supply the required voltages, but I also picked up some switching supplies that ran servers, they do 48vdc reg. One does 3kW off a 200-240vac input, The other does 775W off 120vac nominal and 1200w off 220vac nominal. I have two of them, they're not that large, they're the prime candidates right now. Assuming they don't make a ton of hash (and they might, haven't gotten them to fire up yet), then it's a no brainer for the PS. I'd like to keep it light weight if possible.

As far as the IMD figures, I was hoping that if I inverted the phase of one pair, ran them in PP at the output combiner that I'd get a bit of distortion cancellation, as opposed to running them in straight parallel. That worked for Stever QIX in the class E.

                       _-_-

[KK4YY]

Bear,

Those high gain MRI modules might be a good way to go for anyone with a Flex1500 (like me) to use it's 1 mW low level output. One module should make a clean (-40 IMD) 50W carrier output from the little Flex.

The EB104 application note shows a 4 device set up — PP parallel. I don't know of a 2 device version.

Batteries are a thought for a PS. No hash, pure DC. I have lots of 12V 35AH VRLA batts if you want some.

I don't know if odd order IMD products will cancel in PP. Sound too good to be true. Some years ago I worked tuning feed forward power amps for cellular base stations and it was no easy trick to cancel the IMD products.

Don

[kb3ouk]

I know even order will cancel out in a push-pull design, but I don't think odd order will.

[WBear2GCR]

No, not the odd order - that's what feedback may help...

The other thing is the "Diplexer" - anyone have any experience with it??

Here are some shots of the board(s).
The small parts side of the board gets jettisoned...

Yeah, they would be nice with a very low level output rig like an SDR...
Not so sure about the batteries - they get heavy. Quiet though.

Something to think about.

                   _-_-

[WBear2GCR]

Sorry, these are native shot images, did not resize them...

[KK4YY]

Bear,

What's up with the interstage coupling? Kinda looks funny to me. I think you need to sweep those amps and find the frequency response. They might not be as broadband as you hope or designed for VHF. Some changes may have to be made.

Are you thinking of building a diplex filter for the output?
A good read is the QEX  article by Sabin, "Diplexer Filters for an HF MOSFET Power Amplifier".
www.qsl.net/wm5z/qex199907.pdf

The First stage, MRF134, is a 28V device with about 47dB(!) gain at 75M near as I can figure. The second stage, MRF148, has 20 dB gain in class A. So it looks like about 45dB gain for the last 2 stages if you omit the first stage. All 3 stages, a whopping 92dB of gain on 75M.

Nice pics. I like 'em BIG! Got a schematic?

Don

[WBear2GCR]

Right, the coupling xfmrs I think need to be replaced with other core material, and associated components, etc.

I got the birds-eye-lowdown on these units from another ham, including mods that work and the schematic. Not sure he wants to be mentioned on this or not. He could come in and comment, if he happened to be reading...

...so getting them set up and running should be fairly straightforward. My post was with the thought that it might be possible to get more efficiency and less heat with the Class H idea, but I started with the (wrong) idea that the output is in class A not AB... or if it is in A, then it will be in AB. So, unless there is a real way to do the Class H and there is some real benefit, there's no good reason to pursue it for this project.

Seems like the Diplexer idea is pretty cool, and helps keep things clean(er). I have read up on it before, but as with many things I read about and never use, I had forgotten completely about the idea until I saw it mentioned on one of the EB-104 web pages, and a bell went off. Oh yeah, that, right!

Wonder if anyone out there in AMfoneland has ever tried the diplexer??

                  _-_-

[W2VW]

I have iron that will supply the required voltages, but I also picked up some switching supplies that ran servers, they do 48vdc reg. One does 3kW off a 200-240vac input, The other does 775W off 120vac nominal and 1200w off 220vac nominal. I have two of them, they're not that large, they're the prime candidates right now. Assuming they don't make a ton of hash (and they might, haven't gotten them to fire up yet), then it's a no brainer for the PS. I'd like to keep it light weight if possible

                       _-_-

Ifin you have the HP 52 volt 57 ampere switchers they are quiet. Still need R.F. proofing at B+/- out  because the switcher simply stops dead in it's tracks when confronted with weirdness. Last year I heated the basement bench testing 55 Larcan TV Amplifiers with one of  those. Never heard any racket on the wireless set.

[WBear2GCR]

Dave,

Great info there!

I think I have the same one, it's about 18" long? (guesstimated)... attached is the tag on it.

But the ones I was thinking of using most would be the others I picked up, run in parallel, or to separate modules, because they will go 120 or 220 without jumpers, but are 700w/1200w respectively. They're also not as long, and seem to have two wire fans, not 4 wire, meaning that the fans can be pulled in favor of blowers without having to fake out a tach circuit... Tag and  chassis image shown below for them.

I've got the fire-up control pin jumper info for the big 3kW supply, but not the smaller ones.

Fwiw, anyone who has not noticed these yet - and they had escaped my view - these server supplies are going dirt cheap and put out a boatload of power at voltages spec'd between 5vdc and 48vdc depending on the supply, and from a few hundred watts out to at least these HP that do 3kw off 220vac nominal!

                  _-_-

[KK4YY]

Bear,

I guess you're thinking of using a remote blower system to keep the in-shack noise down.

I tried that once with a noisy blower. I used a length of hose from a swimming pool filter to transport the cool air from the blower which I mounted outside the house. When I fired it up I was surprised at how little air came out. This is when I discovered friction loss in pipe. It's significant.

The bigger surprise was that the spiral nature of that hose caused a resonance in the air movement which produced a loud howling sound. It was worse that the blower noise had been!  As I went outside to check on the blower, my wife inquired about the strange noises emanating from my basement shack and outside of our house. I grumbled something. Outside, the noise of the blower was quite the racket in my quiet suburban neighborhood.

I was completely skunked. This taught me the value of bench testing. I hope you have better luck!

Don

[WBear2GCR]

No, not a remote blower, although that can work, but you'd want to use something akin to the methods used for HVAC... a blower, not fan, working into a suitable back pressure down the tube of appropriate diameter.

These power supplies use a standard high speed fan that winds up like a jet plane. Don't need that in the shack. So the idea is to replace that with a blower. The blower can be physically larger than the little screaming 3" fans that are built into the supplies. They just need to end up with a leakproof connection to the spot where the fans formerly sat. The idea is that fans are not particularly good at working into a back pressure at all, but blowers are. So, running a larger blower at slower speed will be quieter (and lower in frequency no matter what) than the 3" fan for equal air flow.

At least that is the idea. How it works out in practice for this application remains to be seen. Proportional speed control is possible with a thermistor and a DC blower...

[KK4YY]

Bear,

Ok, like a squirrel cage blower. A central blower gives you the opportunity to use an air filter to keep the crud build-up in check and perhaps quiet the intake a bit too.

I found a circuit a while back that used a thermistor and an LM317 to control a PC's CPU fan speed. I wanted to use it on my FT-101 but haven't gotten to it yet. I don't have a NTC thermistor in my junk box! http://www.extremecircuits.net/2011/06/pc-fan-speed-controller-for-low-noise.html

Diplex filters - It seems to be a fuzzy area whether they help with IMD or not. What they surely do is take the harmonic energy and send it to a dump resistor rather than back to the finals to heat them up. Good enough. Several months ago I was looking into diplex filters and found a few hams who had built them. One, I think, was offering kits, boards, or at least component values, I don't remember. I probably bookmarked the pages but they would be on an old PC that I don't use anymore. A search should turn them up for you. I have a project, long on the shelf, that I want to use them for. Let me know if you come up with a good recipe.

Don

[WBear2GCR]

Yes, I call 'em squirrel cage too, but I've never seen a squirrel in a cage, much less in one with a big wheel to run on! Hamsters, Gerbils and mice, yeah... no squirrels!

More properly I think they are centrifugal blowers...

No matter what you need some sort of intake air filter... that works out to be a large area filter as
opposed to a filter the size of the intake of the particular blower especially if one is moving a large volume of air...

One would think from all this discussion that someone is actually building something! Ha!

                         _-_-