The AM Forum

Long overdue update at:

http://www.w1vd.com/375wattclassD.html (http://www.w1vd.com/375wattclassD.html)

Excellent break down, Jay. Certainly well thought-out and explained clearly if even I can understand it.  ;D

TNX fer posting it.

Nice Job Jay
Guys running class e might consider TLTs. I used them in my MRI linear project. Running DC through the transformer is possible if the primary is the shield since it is usually less resistance than the center conductor.
I'm having no trouble getting amps to work 160 meters through 6 meters.
TLTs have tighter coupling between primary and secondary. Smaller 25 ohm coax can be made using #24 single shielded teflon wire. I'm running 300 watts PEP class AB through it with 600 watts DC going through the shield with no real heating. slipping a couple wires through a core makes for a crappy transformer with lots of leakage inductance. Brass or copper tube primary is only slightly better.

Neato....

but the link for the 75m meter deck is broken?

just comes up to your homepage.

Thanks Jay.

Very nice work Jay along with the great theory of operation writings. I must admit that your "prototypes" look to good to be a prototype!! 

Thanks for the contribution!

Joe, W3GMS

Jay, Thanks so much for the update!, I been working on building one, and your design is
a big help!   73-Chuck W3FJJ

Beautiful work, Jay!

Well software documented, excellent layout, pictures - good explanations.

It makes me think of how far we have come in the last 15 years with this class E / D  AM ham technology and implementaion.


As usual, a FB engineering effort, OM.

T

That's a better explanation that I've seen elsewhere. nice looking board as well.

Jay,

Thanks for the design update. I am working on a 500Khz transmitter and this may be the next project.

Pat
N4LTA

Very nice.  You do exquisite "prototype" work.     I was wanting to get a really good look at your 40 meter unit, but when I click the picture for enlargement, it links to an unrelated page ...

Chris.

The link to your class D final for 40M worked fine for me.

Joe, W3GMS

Excellent write-up, Jay. Impressive results from noodling-out the improved efficiency on 40m, and then backfeeding the design down to 160 and 80 meters.

The combination of strong documentation and clearly done photography really enhance the text presentation, too.

The 600 pixel wide photo of the 40 meter prototype with one driver per fet is an html link.  When I click that, hoping for a photo enlargement, it takes me to "An experimental LF station WD2XNS" showing space weather.

Chris

That html link has nothing to do with the photo of the 40M RF deck.  Its simply a link at the end of his posting to allow you to go to Jay's web page. 

"Back to W1VD / WD2XNS home page"

Joe, GMS

Jay,

   I wonder how you are measuring the gate and drain waveforms. Specifically with your probe, I presume a compensated 10X, I wonder how you attach the scope ground return?

   I ask because with the commercial class E 3 KW RF AMPS I'm testing (13.56 Mhz), I use a little spring clip ground attached to the probe (about 3/4" long), and that gives me the most believable scope shot of the gate drive. I get a very different picture with the conventional 3" ground lead and alligator clip.

Jim
WD5JKO

Jim,
Make sure your scope probe is rated for the voltage you are running and the scope is matched to the probe. At 13 MHz you want a real short ground lead. I like the old Tektronix RF adapter with the spring loaded ground clip about 1/4 inch long. It slips over the end of the probe in place of the witch's hat. Be sure to check the probe compensation.

1st ... thanks all for the positive comments!

2nd ... links have been corrected or eliminated as appropriate.

Jim ... I remove the 'witches hat' as Frank called it and have a piece of #14 solid copper wound to accept the probe ground 'sleeve'. This is attached to a #4 solder lug that temporarily mounts under one of the screws that attach the circuit board to the heatsink. Length between ground and sleeve is very short. The assembly can be twisted around so the 1/4" fine probe tip can touch either the gate or drain.

What do your waveforms at 13.56 MHz look like? Wonder what FETs you are using up there. My next missions is to move up to 20, 15 and 10 meters with CMCD. Any pointers would be appreciated.       

 

Not that link.  The photo above it showing the wonderfully constructed 40 meter unit IS A LINK.  Usually when a photo is a link, it opens a larger photo.  I mentioned it only because I suspect it was an oversight, and I was wanting a really good look at the construction.

Chris

[ EDIT:  Looks like the photo link was removed.  Oh well, no bigger pic ... ]

Why don't you just right-click... "save picture as.."
Save it to your computer, and resize it?

That's what I did.. looks good enough.

Thanks Jay, your work is a real inspiration. I've used some of your designs before with great success. That 375 W Deck is a beauty

73 Frits W1FVB

Chris ... here's a link to a bigger pix of the 40 meter deck http://www.w1vd.com/40meterdeck1800.jpg (http://www.w1vd.com/40meterdeck1800.jpg) . Don't look too close, though, it is a prototype after all  ;D.



 

Thanks for the hi-res photo.  I thought I did nice prototypes.  I WAS WRONG!

I'm building a class E rig that I want to bandswitch between 40/80/160 - with digital drive using one driver per fet.  I'm debating using IXFH12N100F devices rather than the FQA11N90.

Thanks for your inspiration.

Chris

One thing I did differently with the 40 meter D rig I built a couple months ago, was to face the FETs the other way.

This way no "gate buss" is needed.

Soldered the output pin of the driver right to the gate lead of the FET, about 3/8 or 1/2" total from package to package.


I don't KNOW of any ill effects from that arrangement.

Thank you Jay for the larger photo. The little rectangles between the drain of the driver and the gates of the FETS in your schemtiac are clear now. On 40 meters, what did you use as the required drain impedance... R//C equivalent?

Alan

Gee, I bet the spring from a ball point pen could be wrapped around the probe and bent to make a nice short gound. The newer probes are getting smaller. I've also seen an adapter to a BNC in some probe kits so the probe is totally shielded. I've never used it but looks cool. A clip from a BUSS fuse holder might also work.

 Jay,  

I am away on a business trip, but I will get some pictures in a week or two of the gate drive.

These AE amplifiers use RF drive, and two power FET's in parallel to make 3KW out of 200 VDC. They have three power conversions in the box, 3 phase 208VAC to unregulated DC (with small energy storage cap - PFC correction), then DC to DC regulated 0-200VDC, and then DC to RF. Overall they are 55-65% efficient depending on the power level and the individual unit.

  I am new to these technologies and my time with them has been limited. It appears to me though that RF drive has a problem over digital drive. With RF drive in order to get the FET gates over ~ 8 volts peak, the pulse width gets too wide. I definitely see the unit efficiency drop several % as the RF drive amplitude increases. Conversely as the drive amplitude decreases, I see the % efficiency increase. I tune the RF drive (a coil to bend) while monitoring the fet drain saturation voltage and reducing the gate peak level until I see the the fet 'ON' drain voltage start to rise a tiny bit, and then go back until I am 'just' saturated.  There is nothing to tune on the Drain side.

Jim
WD5JKO

FETs turn off slower than they turn on so it sounds like the increase in drive is making the FET turn on more than 50%. In a push pull circuit this causes phase overlap and loss of efficiency. I have played with drive transformer  turns ratios to reduce voltage and decrease drive Z to force the gate on and off quicker. Sounds like you know how to find the sweet spot but not sure if that spot is stable for you.
My 75 meter final (22 FQA11N90s in push pull parallel) I run 4:1 turns ratio with the secondary grounded center tapped to drive two phases. I run about 88 or 89 % efficiency class e. It takes about 70 watts of RF drive.

Agree with Frank - sounds like a gate drive duty cycle issue. On 40 meters it was impossible (at least for me) to break into the 90's without it.

Suspect attention to gate drive is even a bigger deal at 13.56 MHz. Your 'sweet spot' may be the best that can be done in your current configuration ... but may not be optimum. Would be interesting to try a duty cycle controlled setup and look for improvement in drain efficiency.     

I'm noticing the IXDD gate drivers have a Schmitt input.  Has anybody tried sine input to the drivers, varying the level to control the ultimate square wave duty cycle?

Chris

 

Chris

Sounds like 'flirting with disaster' to me. It's pretty easy to have 'positive control' without a lot of circuitry. A schematic of a suitable circuit at:

http://www.w1vd.com/VFOdriver.pdf (http://www.w1vd.com/VFOdriver.pdf)

That 40 meter board looks really nice.  Perhaps you could sell your idea to some big preamplifier corporation that might produce them commercially ... ?