New 1-FET, 35-Watt Class E "starter" rig

Post your photos here and write-ups you may have

Postby KM5TZ/6 » Thu Jul 28, 2005 8:02 pm

Tom WA3KLR wrote:I wonder how well the push-pull configuration performs for the 3rd, 4th, 5th, 6th, and 7th hamonics, compared to the single-ended/LPF approach?


Tom,

Heres a spectrum analyzer shot of my Class E output on 80m at 2MHz/div, 10dB/div. First peak on the left is the DC marker. 2nd harmonic suppression is outstanding. This is a 4 x 5 FET. This was measured with a resistive tap into a dummy load.

David.Image

Here's the link in case the pic does not show up. 5th picture from the bottom on this page (limitation of my freebie web host)
http://www.km5tz.netfirms.com/classe.htm
KM5TZ/6
Member
 
Posts: 26
Joined: Mon Dec 13, 2004 1:47 pm

Output coil question...

Postby AB2EZ » Fri Jul 29, 2005 6:13 am

Bill

You observed/asked:

"Also I noticed that your series inductor is huge and edge-wound with insulated wire, have you checked it for paracitic resonances? (tina, and QIX are running copper tubing, I'm running #4 AWG), all with wide turns spacing, is your series coil getting smoking hot? "

I turned the rig on, and ran it key down this morning for 15 minutes (no modulation) into a dummy load.

Here is what I measured and observed:

RF Output power: 39 watts
Input power (to the RF stage): 40.5 volts x 1.21 amps => 49 watts
Efficiency (output/input) of the rf stage: 80%
Wall-plug power (including class H modulator operating at carrier, and losses associated with the power supplies): 99 watts

The output inductor and the cores of my 1:1 4-core output transformer are just barely warm to the touch. Probably around 100F v. 75F room temperature. Then again, I'm only running 39 watts of output power.

For a simple starter rig... using standard #12 electrical wire seems to be an acceptable compromise. On my 200 watt (at carrier) balanced 2-FET x 2-FET rig, I use #10 stranded, insulated wire (close wound) that I purchased at the local NAPA auto parts store. It does not feel warm when I operate key down for extended old buzzard transmissions.

Best regards
Stu

P.S. ... David: A nice screen shot of the output spectrum of your rig!
Stu
Bernardsville NJ
AB2EZ
Member
 
Posts: 16
Joined: Sat Apr 26, 2003 9:15 pm
Location: Bernardsville, NJ

Postby ke1gf » Mon Aug 01, 2005 1:22 pm

RF Output power: 39 watts
Input power (to the RF stage): 40.5 volts x 1.21 amps => 49 watts
Efficiency (output/input) of the rf stage: 80%


Stu your efficency seems quite low, can you attribute it to anything like IxIxR losses during the switch on cycle or anything like that? Maybe you're not driving the FET's hard enough, I run about 30v P-t-P on the gates of my 11N90's and I'm getting 90% efficiency.


Regards,
-Bill 'GF[/img][/quote]
User avatar
ke1gf
Young Brainiac
 
Posts: 229
Joined: Mon Feb 03, 2003 11:38 am

Efficiency question

Postby AB2EZ » Mon Aug 01, 2005 7:17 pm

Bill

The efficiency is definitely lower than I would like (mostly as a matter of pride, rather than necessity).

I tried increasing the voltage on the drain of the QFET (at carrier) from 35 volts to 42 volts (by adjusting the "variac")... but that didn't improve the efficiency.

I changed the output transformer back to 1:2 to reduce the current that flows in the output coil at full output power. Now there is no noticible temperature rise at all (it was small before, but I could feel some temperature rise with a 1:1 transofrmer) in any of the output components (coil, transformer)... but still no significant measurable increase in efficiency.

The drive voltage to the gate of the QFET is quite low (~5 volts as measured at the QFET leads)... so that is probably the problem. The rig works fine, but as you said.. not enough drive. For the same input circuit design, the drive voltage is much higher on the gate of the driver QFET of my 2 x 2 rig... so I need to figure out what the problem is.

Another thing I need to check is whether or not I have too large a doorknob capacitor. Steve says that using too large a doorknob capacitor will reduce the efficiency. Right now I'm using 500pF, but that may be too much for a single FET output stage. I may try 250 pF.

Stu
Stu
Bernardsville NJ
AB2EZ
Member
 
Posts: 16
Joined: Sat Apr 26, 2003 9:15 pm
Location: Bernardsville, NJ

Postby frank carcia » Tue Aug 02, 2005 11:21 am

Stu,
Look at your turn off drain voltage. It should be about 3.5 times VCC at the peak. Lower shunt cap will increase the voltage and larger will reduce it. I started at 78% with my final. My problems were 1.shunt cap too small
2. 1:4 transformer wanted to be 1:2. 3. Q was too high for 1:2 transformer so reduced inductor value and added C. 4. Inductor wire size too small. Fixing 1 through 3 got me to about 88% at 1 KW out. I was running about 24 V P-P drive so increasing that may help. At 750 watts out I was over 90%
sheetrocking and dreaming of radio projects. Wild life on the beach helps though.
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Where are my 4.2 missing watts?

Postby AB2EZ » Thu Aug 04, 2005 3:17 pm

Well...

In search of 4.2 watts (90% v. 80%) when running with 42 watts DC power input to my 1-FET rf module...

I tried all sorts of things that didn't have any favorable effects on the efficiency, or anything else:

e.g., several different designs for the output transformer, removing the doorknob capacitor, etc. So I didn't stick with any of those changes.

With the 500pf doorknob capacitor in place, the peak drain voltage is 120 volts when the DC drain voltage is 35 volts. So that is pretty close to the target of 3.5x the DC voltage.

What did help was increasing the drive voltage (as suggested by Bill). I hooked the class E rig up to my main station rig... set at ~4 watts output... instead of my QRP rig running at ~1 watt output. The voltage on the gate is now a reasonably clean sinousoid at ~ 15 volts peak. Prior to that it was a fairly distorted sinoisoid with a positive peak value of about 10 volts. The quoted input power levels are approximate... so the ratio of the voltages is not 2:1 as one might expect with a 4x increase in drive power.

With that change, the class E rig seems to be even more stable (e.g. insensitive to the loading capacitor value), it produces an even nicer class E waveform (on the drain of the FET), and it is more efficient:

Current results:

At 35 volts DC drain voltage and 1.2 DC amps of drain current (at carrier):

Input power: 42 watts
Output power:38 watts
Peak power (on positive modulatiuon peaks): 220 watts
Efficiency: 90%

At 41.5 volts DC drain voltage and 1.4 amps of drain current (at carrier)

Input power: 58 watts
Output power: 50 watts
Peak power (on modulation peaks): 300 watts
Efficiency: 86%

To some extent, this is just an intellectual exercise. Obviously, the extra power to run my FT-1000 Mk-V as a driver (v. my FT-817ND QRP rig) offsets the increased output power from the class E rig by more than 10:1 ... but it was interesting to see the effect of having a drive signal of the proper level.

Stu
Stu
Bernardsville NJ
AB2EZ
Member
 
Posts: 16
Joined: Sat Apr 26, 2003 9:15 pm
Location: Bernardsville, NJ

Postby frank carcia » Thu Aug 04, 2005 4:03 pm

Interesting Stu. I run about 24 V P-P so maybe I could hit it a little harder.
I also found your gate signal is cleanest when the rig is properly tuned.
I suppose that is when the least amount of crud gets reflected back at the gate.
When we tested my rig at QIX pad his rice box had a problem driving my rig with a high VSWR. It was a bit unstable with poor drive. This is not the case when I use my TR7 to drive it with a VSWR of under 2:1.

Presently my plan is to use a BB 100 watt amplifier strip from an old Raytheon marine rig that has 3 stages and takes about 10 mw of drive.
The hP8640B could be the source or the HP3325 for thos SBE nights.
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Postby ke1gf » Fri Aug 05, 2005 2:42 pm

Yeah stu, so driving the gates a bit harder helped eh? From what you said you weren't even driving them out of the triode region or just barely.

I get my efficiency and stability from drive, many a time I have thumped the "switchmode drive" bible to Tom Vu and Tina while they were building their big rigs. I like to have a stable rig that fires up each time I key the thing no matter what, it gets imbarrasing when you have a crap out.

It's very important!

And I don't use transzorbs, they can create more problems than they solve most of the time.

I use this formula from IXYS's MOSFET and IGBT driver bible to compute how much drive power will be needed

Work = (Gate Capacitance)x[((P-t-P Voltage) x .707)^2]x(frequency in hz)

or you can do it this way, it's the same thing

Work = (Total Gate Charge)x(Vcc)x(frequency in hz)

sorry but you can't lick the laws of physics, at least not yet.

Uhh, so you probably say what advantage does 30V p-t-p have over 24V p-t-p, other than making more heat?

it's the duty cycle, it goes a little something like this

lets say that the FET acts as a switch at +10V G-S

so if I run 30V p-t-p which is 15V pk = A (sinewave)

10V = (15V)sin(2xPIxfxt)

so at 4 Mhz, we know that t = 250ns they are hand in hand

but, we want to know how much of the period is the FET acting like a switch...

make sure your calculator is in radians, hehe!!!

arcsin( 10V / 15V ) / ( 2 x PI x 4000000 ) ~= 29ns after zero crossing

this is where the miller plateu occurs, the little notch in your sinewave.

and we can subtract back from 125ns at the zero crossing and get

125ns - 29ns - 29ns = 67ns of ON time per period

so the %DC is 67ns / 250ns ~= %27, you can recompute it for different amplitudes...

this is at 4 Mhz, at 8 Mhz though the %DC increases, and as the frequency gets higher the %DC gets closer and closer to %50 on...

No magic here, but if you go back to my equations the other day I was using %50 DC to do my figuring.

I'm to the point now that I design drivers for the E rig and I use sinewave drive at the final transistors because of the timing aspects that you get for free, no more rice box.

I'm really not interested in fiddling for a week with a tapped tank so my rice box is happy, and then when I "YAAAAAAY" into the mic watch my SWR foldback kick in... SNAP BANG POOF!

Cheers
-Bill 'GF
User avatar
ke1gf
Young Brainiac
 
Posts: 229
Joined: Mon Feb 03, 2003 11:38 am

Postby frank carcia » Sun Aug 07, 2005 9:13 pm

Bill,
Transorbs are great in their proper application which does not include
being a drive limiter. They are great for short pulses and transients. Beyond that they are fairly useless. The zener voltage takes off as soon as it starts flowing current. Yes, you have to deal with extra C as a parallel load on the gate drive. I use a single 1.5 KW part on each phase.
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Postby AB2EZ » Thu Nov 24, 2005 12:44 pm

Hi!

As an update...

I put up a 160 meter inverted vee, and I decided to move this 1-FET rig to the top band. It took about an hour to wind and put in a new output coil, to add an additional 500 pf doorknob capacitor across the output of the FET, and to make a few other minor tweaks. It works like a charm! I'm going to post a picture of the 160 version on my web site. See you on or about 1885 kHz!

Best regards
Stu
Stu
Bernardsville NJ
AB2EZ
Member
 
Posts: 16
Joined: Sat Apr 26, 2003 9:15 pm
Location: Bernardsville, NJ

Postby AB2EZ » Thu Nov 22, 2007 12:32 pm

Happy Thanksgiving (2007) to all...

As an update... I obtained a couple of IXYS IXDD414 driver IC's from Steve (WA1QIX), and I used one in my single-FET Class E rig to substitute for the input tuned circuit I was using up to now.

It works like a charm. I actually mounted the IXYS device right on top of the Q-FET, using the same 4-40 screw. Remember the the QFET's heat sink is insulated from the main heat sink, but the IXYS device's heat sink is grounded. On 160 meters, there is no significant heating of the IXYS device that I can detect.

I'm relying on the IXYS device's internal "enable" pull up resistor to keep the enable at Vcc. Therefore, I have only 4 leads to connect: Vcc (12 volts), ground (directly to the QFET source lead), input, and output (directly to the QFET gate lead). I'm using a .01uF (10,000 pF) capacitor from Vcc to ground (very close to the device leads)... and that appears to be adequate for this application.

I'm driving the IXYS device with a combination of 1.25 volts of DC bias (applied only when transmitting) plus a 3 volt peak 1.885 MHz rf signal.

Note: for this particular sample of the IXYS device, the switching theshold appears to be only about 1.5 volts. This is not a problem... I simply adjusted the DC bias to be below the threshold of the actual sample I have.

Again, it works great.

Best regards
Stu
AB2EZ
Stu
Bernardsville NJ
AB2EZ
Member
 
Posts: 16
Joined: Sat Apr 26, 2003 9:15 pm
Location: Bernardsville, NJ

Previous

Return to Your Photos and Articles

Who is online

Users browsing this forum: No registered users and 1 guest

cron