'Tis the season for 10m AM! class-e amp help request

[kf1z]

Hi All,

Thanks to Steve, QIX, I have succesfully put together a pulse-width modulator, and 10 FET 75 meter amplifier........
Not completely finished, or pretty...but on the air.

Now, since the 'ol sunspot cycle will soon be in an upswing, and 10 meter propagation will be improving.... I'd like to start thinking of putting together a 10 meter class-e amplifier.

Since I am not and engineer, and sure there are plenty of other folks who would like to do the same thing...
I am hoping you guys with the "know-how" will start throwing out some more usefull info to get us started....

Anything from which devices would be suitable... how to calculate any shunt capacitance, and most important.... a suitable tank circuit...

So yes, just about any info you care to share on the subject will be much appreciated.


I think it's a good time to start a 10meter thread...

What say you?

Thanks
Bruce G
KF1Z

[steve_qix]

I've got a 100 watt design, sort of on the back burner at the moment, but maybe will be brought out again, soon !

The main requirement is a MOSFET that has a low enough gate inductance (within the package) to allow the gate to be driven at 30mHz. These MOSFETs, an any sort of power at all, are expensive.

There are some smaller IXYS MOSFETs that look promising... more later!

Regards,

Steve

[blackplasma]

I know it sounds high, but is package inductance really so much of an issue at 30MHz?

Sure, the TO-247 packages are tricky... but what about resonant-circuit drive, where one usually adds a wee bit of inductance anyway?

The highest I've worked at (so far) is 17MHz, and I very rarely use low-impedance squarewave drive into the FET -- I don't have the drivers, and stuff gets too hot, so I generally make drivers from discrete components, swamp the gate capacitance slightly with a ceramic cap to make it a little more predictable, and resonate it with a ferrite transformer. =P



What kind of power are you aiming for? If it's <~150W you could probably pull it off with a few IRF510s or IRF520s...?

[steve_qix]

I know it sounds high, but is package inductance really so much of an issue at 30MHz?

Sure, the TO-247 packages are tricky... but what about resonant-circuit drive, where one usually adds a wee bit of inductance anyway?

The highest I've worked at (so far) is 17MHz, and I very rarely use low-impedance squarewave drive into the FET -- I don't have the drivers, and stuff gets too hot, so I generally make drivers from discrete components, swamp the gate capacitance slightly with a ceramic cap to make it a little more predictable, and resonate it with a ferrite transformer. =P



What kind of power are you aiming for? If it's <~150W you could probably pull it off with a few IRF510s or IRF520s...?

The IRF 510 - 100V, 4A device. So, on AM you could run - with a not too big safety factor about 10 volts unmodulated DC. Figuring on 25 volts for 150% positive modulation - then 4x the DC for the peak of the class E waveform - that's 100 volts with NO SAFETY FACTOR. I usually recommend a 100% safety factor - but let's go 50% for the sake of the discussion. So, that's about 7 volts - times 10 for the peak at 150% modulation of 70 volts - not a great safety factor, but better than none!

Ok, let's take the current. The repetitive peak current is about 2x to 3x the DC current (depending on the gate waveform). It's lower for square wave drive. So, let's be generous and say 2x the DC. So, that's 2 amperes max current at the crest of the modulation waveform. Divide 2 by 2.5 (the peak voltage and current relative to the carrier) and we get about .8 amperes at carrier. 7 volts at .8 amperes = 5.6 watts per device ***** FOR AM SERVICE *****. It would be a LOT higher for CW / FM - at least 6 times the power or about 40 watts.

The input capacitance is about 200pF (250 max), so this is practical at 10 meters or even above - much more so than the 3500pF of the FQA11N90. These devices have a gate self-resonance of around 20mHz. The IRF510 will probably have a self-resonance of over 100mHz.

So, small die (like the 510) = low capacitance = low power.

That's essentially the design of my 10 meter rig. A bunch of devices that will run about 10 to 15 watts, put together to get about 100 watts.

Good discussion!

Regards,

Steve

[frank carcia]

I know of a guy who has a 6 meter linear running IRF510s
I think they are running 28 volts. I wouldn't try to modulate at 28 VDC.
My 160 meter rig uses IRF840s at 85 volts peak (30V carrier). Peak drain voltage swings just over 300 volts.

[blackplasma]

Because I found a great deal on some IRF730s ('400V 5.5A") recently, I used them as a nice intermediate FET at medium powers for high frequencies.

It ended up they were the ST version, which has a slightly lower gate capacitance (530pF) to the IR and other brands (IIRC figures up to 700pF).

I don't think you should have too much trouble swinging an IRF730 at 30MHz.. but I always seem to find reliable gate drive the hardest step.

I've got no idea how IXDD414s behave at these frequencies, but for what it's worth I've pushed UCC37322s up to 18MHz into two IRF510s, (but any higher than 7.5V supply voltage and cross-conduction heats the chip up in no time). The UCCs of course cannot source the current for square-wave drive - I tune them such that the 5V supply voltage resonates up to ~16Vp/p, and then float the MOSFET's gate at 6V.

If all else fails, maybe you can scrounge a DEIC420 from somewhere, but I've never even set eyes upon one. =-P

If IXDD414s can handle 30MHz, I'd recommend coupling them to the gates with a transmission line transformer on a binocular core... good luck and happy tweaking!!

[cableman]

I often test a single IRF510 at 30V to see what the max power is,but they never break.(some of them 60Watt,not all equal!!)
Then i modulate them at 12V(12Watt carrier,not 150%)
They let them nice drive by a MC34151 at 7mhz,or the EL7202(inv ad non inv.)
Both drivers trigger on 2,5V or higher on the input.
I get the best results at 8V at the gate.
7V on the drain will give 7Watt.
Now trying to drive a IRF840 with the same configuration,but no succes at this time.

[frank carcia]

840s have a lot more gate c I use them in my 160 meter rig at 85 volts modulation peak and around 30 carrier. gfz

[WA2WVL]

The IRF540N looks like a better part for 10 meters than the IRF510.

I tested a pair in push/pull at 28MHz 9 months ago and the info is in my spot "28 MHz Class E Amplifier" posted on Jan 30, 2006.
Input capacity is of no concern if you series tune the input and match up to 50 ohms. The input impedance of the IRF540N is also listed in one of my spots.
I belive that even large parts like the IRFP32N50K will work well on 10 meters with tuned input.
Floyd

[blackplasma]

I gave up on series/parallel resonance on the gate... the polysilicon gates seem too lossy to resonate properly at all at this frequency. And ack, looking at those waveforms, I'm not sure if I'd get away with them... the transition through 4V>8V seems like it'd do. There's effectively half an ohm of resistance added to each IXDD output... I should reduce that.

Well, this sure isn't 30MHz, but with ordinary IXDD414s I've finally had some success swinging an IRFP450 at 12MHz.

So on second thoughts, where do you guys buy your DEIC420s from? =-P




I think > 13.56MHz really implies a discrete driver (such as a smaller class-E stage) or a proper RF driver.[/img]

[kf1z]

You can now buy the DEIC420 ONLINE.....
(We had not been able to do it online, and had to call IXYSRF, but I see the store works now.)


http://www.ixysrf.com/products/MOSFET_Driver_ICs.html

they are $35.12 us each.

[frank carcia]

Like the F4 (flying brick) Apply enough thrust and you can get anything to fly.

[WA2WVL]

BLACKPLASMA:

If you will let me know what part you want to run at 10 meters I will design the input match for you.
Floyd WA2WVL

[kf1z]

BLACKPLASMA:

If you will let me know what part you want to run at 10 meters I will design the input match for you.
Floyd WA2WVL

This input match is, I assume, for sine wave drive?
And not for use with a driver IC right?


In any case, If you don't mind, why not tell us how to go about figuring the match.
I saw your diagram, but have no real idea how to determine the input components etc.

thanks
bruce g
kf1z

[WA2WVL]

Bruce, the first thing you need is the input impedance at the frequency.
I have measured this on some devices but maybe not the type you are using.
This is done by first assuming that the real (resistive part) is 1 ohm and calculating an "L" network to match this up to 50 ohms. Using a bridge
(I use an MFJ259B) to adjust the network for a perfect match. No circuit just connecting to the Gate to Source.
When you get the match take the shunt C value in the computer network back into the computer program and adjust the R to find out what the R really was. I use the ARRL program TL which comes with the Handbook. Forget the C because it is series tuned out by the L in the network. Now using Xc=1/2PIxFxC calculate the series Xc of the FET. For the IRF540N this comes out to be 1.51-j2.8 @ 28.5MHz.
Make sense?
Floyd