Is Sinewave Drive as good as Squarewave in Class E????

Welcome and feel free to ask questions about class E, mosfet's, and AM.
You must be registered to post.

Is Sinewave Drive as good as Squarewave in Class E????

Postby WA2WVL » Thu Apr 13, 2006 10:25 pm

I have been looking at the IRFP2907 Mosfet (209A part) for use in a 13.6v Class E amplifier. To decide if there was a penally for using Sinewave drive I did a graphical analysis which showed that there will be NO measureable difference in efficiency. Testing of a 300 watt amplifier at 3.8Mhz has proven this to be true.

The driving power for Class E amplifiers can be reduced considerably by using bias just below the turnon voltage (about 4v in most of todays Fets). Without bias the first 4v of drive does nothing. The key to driveing large Fets is to series tune out the input capaclty. This tuning can raise the input impedance so that conventional 50 ohm sources can drive the amplifier. Also it furnishes the high peak RF driveing current so that Drain/Source saturation occures quickly.

At 3.8Mhz one RF cycle is 263ns long. Assuming a 5v peak drive to a IRFP2907 with bias the analysis shows that the RDS(on) will be .1 ohm in 4ns and .02 ohms in 7.3ns (1v drive). Squarewave drivers generally cannot develop risetimes of 7ns (into 13000pf for this part).

Assuming that peak voltage and current for the IRFP2907 was 25v@80A
nearly 2000w pep (Peak envelope power) would be developed. Testing has shown that the peak drain swing in a well designed Push/Pull Class E amplifier runs about 3 times the DC voltage so the 75v rateing of this part is not exceeded. There is really no need to derate the design since the parts have breakdowns much greater than the data sheet shows.
In 30 years of working with high power RF Fets I have never had one fail from excessive drain voltage (even when trying at twice rateing). The failures were always due to excessive drain current which melted the metalization on the die. A good current limited supply can prevent this.

It should be fairly easy to design a modulator running from 13.6v that could generate 10v offset for the AM carrier and 25v peak under modulation. For a more conservative design two pair of IRFP2907's could be used with "Current Transformer Drive" .

Any comments?

Floyd WA2WVL
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby frank carcia » Fri Apr 14, 2006 9:15 am

Interesting post. The latest thing in QEX did a little bias on the gates.
I had a problem with the series inductor in my 75 M final...C was just too big. When I first tested it the drive transformer was jumpered in with 1/4 inch jumpers. When the transformer was soldered to the main board the match degraded without the jumpers. That little bit of l made a big difference in the load. fc
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Postby WA2WVL » Fri Apr 14, 2006 3:06 pm

Frank, when the input impedance is less than 1 ohm a small inductance change will make a large difference. With Current Transformer drive the impedance of each Fet is stepped up by the transformer and then series
with other transformed Fets to produce a still higher impedance. At this point the input C of each Fet is tuned out with a single coil of several microhenrys. The tuned frequency acheived at no drain voltage should be about 5% higher than the final operating frequency as it will shift down as the amplifier runs power.

Floyd
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby WA2WVL » Fri Apr 14, 2006 3:19 pm

Frank, here is another idea. If one insists on using small parts like the Fairchild 11C--- you could do the following.

For example if 8 fets were to be paralleled, connect them in parallel pairs feeding the center of the strap. Then use a current transformer to step up each pair and connect the 4 in series. Add a series C or L to tune ihe total input to your operating frequency. This should work much better than direct paralleling of all 8.

Floyd
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

zero bias tiodes

Postby VK3KRI » Sat Apr 15, 2006 1:48 am

Ive used tuned gates on both push pull and single ended small mosfet amps with considerable success.
In tuning the gates and biassing closer to turn on , is there an increased riisk of parasitics? I suppose that with biasing the gate, if the linear part of the fets curve now is closer to the point where a sine wave has its fastest transition, so the fets in the linear region for a shorter time, so maybe its less?

One advantage of zero bias is it eliminates the bias supply , which is handy for me as I nomally modulate the -ve rail so I don't need to float the modulator FETS gate drive abouve ground.
VK3KRI
Member
 
Posts: 44
Joined: Fri Jan 31, 2003 8:32 pm
Location: Melbourne, Australia

Postby WA2WVL » Sat Apr 15, 2006 11:29 am

VK3KRI, I don't think that the shape of the driveing waveform has much to do with parasitics. Large Fets have VERY high voltage gain as they turn on. The IRFP2907 has a Forward Transconductance of 130S (130a/v)and as the first 1-2 volts above cutoff does all of the work, beyond 2v the gain is low. My 3.8 Mhz breadboard with a pair of IRFP32N50K Fets has a gain of 20DB (3w in/300w out)(@40v) 26DB @ 80v, so some broadband swamping is desireable. Low frequency resonance in the DC feed system is the surest was to be unstable (VG=GMxRL). I advocate transformer drive since the ground return of the driver need not be grounded near the output circulating current return path.

Floyd
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby frank carcia » Sun Apr 16, 2006 6:24 pm

count me in for the driver transformer vote.
Floyd when are you going to build a current transformer rig for all of us to see. fc
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Postby WA2WVL » Sun Apr 16, 2006 6:51 pm

Frank, my problem is that I have no way to put schematics or pictures on the Forum.

Did you get the Email I sent you a few hours ago?
Floyd
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby N9NEO » Sun Apr 23, 2006 11:55 am

A good place to stay away from, or at least traverse through rapidly, is area where a small change in gate voltage can produce a change in drain current. This is where parasitics can be spawned. The mosfets should be either on or off. You want the external circuit to determine the current, and not the voltage on the mosfet gate.

Sorry the explanation is a little weak.

If your gate drive voltage is say 10v and you go to 9v then probably nothing in the external circuit is going to change. So you can put a 1v sine wave riding on top of your 10v square wave and no prob.

If your gate drive voltage is say 8v then you go to 7v then you may inhibit your exernal current, the drain voltage could go up some and that could be coupled into your gate and there ya go - you got a parasitic.

If you drive the gate with a sinewave and the gate voltage starts to look a bit fuzzy then you might possibly have a parasitic in there.

I like the transformer drive also, where you can get one discreet winding on each part. Problem there is now you have to drive the gate negative and positive so your gate drive power goes up some. For multiple fets in parallel maybe the best way might be to use a smaller driver chip on each fet and use one transformer with multiple secondarys which drive the control line on the driver chip.

I would also like to hear more about the series transformer design.

NEO
User avatar
N9NEO
Class E Enthusiast
 
Posts: 322
Joined: Tue Apr 08, 2003 11:52 am
Location: Uxbridge MA

Postby WA2WVL » Sun Apr 23, 2006 6:18 pm

Bob, thank you for your knowledgeable response to my post about gate driving power Mosfets.
There are several points that I want to make about using Sinewave drive.

1.When using bias just below turnon the first 2v takes the fet into full saturation. A sinewave of only 5v peak can produce saturation in less than 10ns. This doesn't mean that the drive must be limited to 5v peak but is meant illustrate how effective sinewave can be in turning on the fet in the shortest time. The fet I was describing in my spot (IRFP2907) has an Rds(on) of .005 ohms so Sat losses are almost nonexistent.
With transformer drive the fet input capacitance (and inductance) is tuned out on the primary side which is simple to do. With the large fets I have been working with, a range of 1.8-21Mhz is easily done.

2. I belive negative gate drive is a good thing in that the input capacitance must be discharged very quickly and drawing current from the gate is accomplished by a tuned input. Also I belive that Squarewave drive waists most of the driving power in driving the fet
input capacity in reigons where no fet action takes place.

3. Amplifier stability. There are several causes of amplifier instability that must be addressed. My experience is that very few Mosfet power amplifiers can stabily operate with gains above 17db (50 times power gain) and input swamping must be used to keep the gate to source impedance as low as practical (I like to use 22 ohms) A well designed amplifier can be tested with a little turnon bias and small signal drive to asses the gain and stability. The "Miller" capacity will cause on frequency or nearby oscillations if the input impedance is not low enough.

A second type of oscillation is well off frequency and caused by not keeping the drain shunt impedance low at ALL frequencies. Excessive choke values or poor bypassing is usually the cause.
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Sinewave Drive continued

Postby WA2WVL » Sun Apr 23, 2006 6:33 pm

A third type of instability is caused by common ground impedance between the input and output circuits. When using a Squarewave type of driver the source lead to ground is commom to both the input drive and the output circuit. With the high circulating currents present in Class E designs feedback occures and effects stability.
With transformer drive the transformer secondary can be connecter at the terminals of the fet and the source lead to ground is in the output circuit ONLY. The primary can be grounded well away from the fet.

Both Squarewave and Sinewave drive can produce a good Class E amplifier but more and more I belive that Sinewave drive has the edge.

Floyd WA2WVL
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.

Postby frank carcia » Sun Apr 23, 2006 8:45 pm

Floyd,
I'm running 25 ohms across phases in the 75 meter rig and 16 ohms each phase to source buss in the 160 meter rig. That resistance does tame things. I have never tried a little positive bias because I'm a bit chicken. I agree with Bob the drive through the miller land needs a good vertical slope and who cares what the peak looks like. Swinging negative with sine drive makes more vertical slope on the high side a little positive bias would help. Swinging negative with a positive 4 volt switch point forces the duty cycle below 50 percent.. a trick I used to discover class e
on my own by mistake. fc
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Re: Sinewave Drive continued

Postby VK3KRI » Mon Apr 24, 2006 5:57 am

WA2WVL wrote:A third type of instability is caused by common ground impedance between the input and output circuits. When using a Squarewave type of driver the source lead to ground is commom to both the input drive and the output circuit. With the high circulating currents present in Class E designs feedback occures and effects stability.
With transformer drive the transformer secondary can be connecter at the terminals of the fet and the source lead to ground is in the output circuit ONLY. The primary can be grounded well away from the fet.

Both Squarewave and Sinewave drive can produce a good Class E amplifier but more and more I belive that Sinewave drive has the edge.

Floyd WA2WVL


Ive always been worried in low voltage / high current stages about the imedance of the source lead. The books say there's 13 nH from the die to 1/4 out on the lead. Even assuming you solder the source lead connections straight at the package body and can halve the inductance, at 3.5 Mhz thats .14 ohms of reactance in series. Is this a problem with 20A of peak drain current ?
VK3KRI
Member
 
Posts: 44
Joined: Fri Jan 31, 2003 8:32 pm
Location: Melbourne, Australia

Postby WA2WVL » Mon Apr 24, 2006 9:39 am

The voltage drop in the source lead does represent negative feedback,
lowering the gain in the active region. With saturating drive this is overcome and shouldn't be a problem. Poor layout can result is much larger feedback voltages as well as phase shifts resulting in oscillations.
Floyd
WA2WVL
Class E Enthusiast
 
Posts: 167
Joined: Thu Dec 22, 2005 1:40 pm
Location: Homosassa, Fl.


Return to Class E Discussion

Who is online

Users browsing this forum: No registered users and 1 guest

cron