Page 1 of 1

2200m Amp - Does This Like Class-E?

PostPosted: Sun Aug 08, 2004 8:59 pm
by VE7SL
I'm wondering how these waveforms look regarding Class-e

Gate:
Image

Drain:
Image

LPF Output:
Image

They are from a homebrew IRF640amp running about 105W output on 137kHz.

Steve / VE7SL

2200m Tx

PostPosted: Mon Aug 09, 2004 3:52 pm
by VE7SL
Could someone comment....do these waveforms look like I'm close to Class-e with this amp?

PostPosted: Mon Aug 09, 2004 9:35 pm
by ke1gf
Hrmn, looks like you could use more shunt C, I'm seeing the wave slamming 0v and I see it comming back up and then the switch turns on. It should gently land at zero and then the switch should turn on.

This will paper will tell you how to fix it exactly:

http://www.classeradio.com/sokal2corrected.pdf

It's a good read...

Except for that, it looks close.

Best,
-Bill 'GF

PostPosted: Mon Aug 09, 2004 11:14 pm
by VE7SL
Bill - thanks a million for the look-see and the advice. I'll try and tweek the shunt C a bit and see what happens. This could explain my 'almost' class-e efficiency...still not quite there yet.

Steve / VE7SL

PostPosted: Mon Aug 09, 2004 11:42 pm
by steve_qix
The waveforms look very class E like. The drain waveform, as Bill pointed out, could be a little "fatter" and not as high. More shunt capacitance should correct this. Also, appears as if the coupling is too light (too little loading capacitance), as the falling edge of the drain waveform comes down very fast, and does not "roll off" before hitting the baseline.

Of course, I'm being very picky.

Looks pretty good overall.

Regards,

Steve

PostPosted: Tue Aug 10, 2004 10:32 am
by frank carcia
I agree with Bill you need a bit more shunt C for 2 reasons 1 slope 2 duty cycle. When the shunt C is low the off pulse returns to zero quicker.
I don't know your VCC but the off pulse should go to about 3.5 times VCC. A low shunt C will make the off pulse a bit higher in peak voltage.
As Steve says it does look close.
A quick test would be to reduce power a bit and see if efficiency jumps up. Efficiency should go down as power is increased but if you see a big change with a small power reduction that is another clue. The ripple during the on pulse makes me think it is loaded a bit heavy for the shunt C value. fc

PostPosted: Tue Aug 10, 2004 11:55 pm
by Tom WA3KLR
The FET appears to be on about 61 % of the time. It should be on 50 % of the time. With the SPICE simulations I did in the past, I found that the Class E stage's operation and power level was changed dramatically by small changes in duty cycle. Lowering the duty cycle appeared to be a way of reducing power. But you don't want to run higher than 50 %.

A good way to get a 50/50 waveform is to generate a signal at 2f then drive a flip-flop. The following drivers have to not add duty cycle bias.

PostPosted: Wed Aug 11, 2004 6:41 am
by frank carcia
Tom,
Remember the drain is at zero when the fet turns on so you can't tell
by looking at drain voltage. I think the fet is turning on at that little ripple
after the darin voltage arrived at zero.

PostPosted: Wed Aug 11, 2004 11:40 am
by Tom WA3KLR
You're right Frank. So much for my posting at 1 in the morning!

Going back the drive signal. that appears to be about 50/50. We don't know the amplitude of the drive signal, but taking the gate threshold voltage into account, in actually the FET on-time is probably just a hair under 50 %. So everything look good.