Re: ATS-4 build questions, output power
- I found the Sokal WA1HQC QEX article at www.classeradio.com/sokal2corrected.pdf
www.classeradio.com has some great "Big Iron" Class E Amp projects.
Also Paul NA5N has some great stuff too:
Scroll down to Tutorials & Articles section for his "Handyman" guides.
73 Trevor AB5EU G3WQO
P.S. I sent this before but then realised I was'nt at my authorised yahoo email so hopefully this wo'nt posted twice.
- I took some measurements of the transmitter this afternoon with my
oscilloscope. Photos are here:
I took four measurements on each band. First, at the output of the DDS
(SIGNB), probed at R22. These were all pretty clean square waves, 3v
peak-to-peak. I labeled these "stage 1" in the photographs.
Next I probed at pin 2 of the 74AC02 (the output of the "b" gate).
These were 5v square waves, some of which had significant ringing. The
ringing looks unhealthy to me but perhaps it is not bad enough to
cause a problem. These are labeled "stage 2".
"stage 3" is measured at the gate of the BS170 MOSFETs. It looks very
similar to stage 2: 5v, square, with significant ringing.
"stage 4" is measured at D3 (the 47v zener). This is where things get
a little bit interesting. As one would expect, I suppose, there is a
relatively flat period while (I assume) the FETs are sinking current
through L8. Then the FETs switch off, causing a steep rise in voltage,
which reaches a peak somewhere around 25 or 30v, then drops off as the
power flows to the antenna. Or at least that's how it looks on all
bands except 20m. On 20m, the signal looks like this:
Two peaks, the second greater than the first. I don't know enough to
tell if this is normal or not, but it does seem to be different than
the behavior on the other bands, so I have to assume this unusual
shape is somehow related to the low output on 20m. Here are the graphs
on the other bands for comparison:
In an attempt to isolate the source of the problem, I removed C49, on
the theory that perhaps Q1 was somehow absorbing or passing some of
the power. But my waveforms and power output were basically unchanged.
I will try digging into the class E design articles that folks
suggested tonight, and I may hack on the circuit a bit more too. I
have some fresh BS170's that I could throw in. But if anyone has any
insights or suggestions based on this latest data, I would love to
hear them. At this point I only know enough about RF circuits to be
dangerous. But I'm having fun!
Thanks and 72/73,
Look at your 80m-4 picture. You see, that the drain voltage falls down, then it nearly reaches zero and shortly goes up. The ideal curve would be that the lowest point touches zero at exactly the same time the BS170 gate voltage is switched off. Read Sokal's articles, you will understand why. The FET shall be switched off when its drain voltage and drain current reach zero. Or at least close to.
Your 30-4 curve does not look correct. The minimum voltage is far from zero and it reaches quite high when the BS170 gate voltage is switched off. BTW, follow the gate voltage on the second channel. Your 15m, 40m and 80m seem reasonable to me. Your 20m curve certainly does not look good to me. It shall not have two peaks, or at least the second peak shall be negligible compared to the first one.
I just hope I am not misinterpreting your pictures and not confusing you even more, hi.
Thanks. I'm working my way right now through NA5N's articles, then
onto WA1HQC. But I skimmed the waveform one and I recognized the same
bump you saw in 80-4. However my guess is that this is this is a
minor optimization problem unrelated to the major problem I'm having
As for the 30-4 photo: I had the scope set for AC coupling, so I
assume that the bottom of the curve in this photo (and the other
stage-4 photos) is in fact at 0v, although I don't know that for sure.
I should have set the scope to DC mode.
In fact, this is a 4-channel scope so I should really solder some test
leads into the circuit and hook up all 4 probes at once. That would
be a lot more clear and concise.
Do the harmonics look right to you in the 30-4 photo?
I'm wondering if this is a normal phenomenon, or if it is related to
the ringing I see in the earlier stages, e.g. 30-3:
Thanks again, and 72/73,
- Here are some photos showing 4-channel measurements:
ch1 (yellow) is the input to U6c (pin 2)
ch2 (light blue) is the gate drive signal (gate of Q2, Q3, Q4)
ch3 (purple) is the drain of Q2, Q3, Q4
ch4 (green) is the antenna output
I haven't looked at these long enough to make heads or tails of them,
but maybe this will shed some light on the situation. All channels are
72/73, Dave W2VV
- After studying these some more, I have a theory. You guys can tell me
if this makes sense:
I think the problem is the ringing in the gate drive signal. On 20m,
the first "ring" peaks at about 2.5v. According to the BS170 data
sheet, Vgs(th) is 2.1v typical, 0.8v min. So Q2-Q4 are turning on
again, allowing them to sink current through L8. This explains both
the plateau and the delayed peak in the drain voltage. It also
explains why these FETs get noticeably warmer when transmitting on 20m
versus other bands.
What I don't understand is why I have all this ringing. Any ideas? Has
anyone else seen this effect in their ATS-4?
72/73, Dave W2VV
Unfortunately my scope is on another continent right now as I will move from USA back to Europe soon, otherwise I would do couple of measurements on my ATS-3b.
The FETs are tricky in a sense that they have pretty high gate to drain capacitance, which also changes with the drain to source voltage. As the gate voltage crosses the threshold, the drain voltage drops and the cumulative gate capacitance changes, which exhibits as a dent in the gate voltage curve.
Also some of your scope pictures show high frequency oscillation, which may be an effect of a bounce in your scope leads. How long are hot and ground connections to your scope probe?
The Class-E PA is sensitive to the length of leads between the PA and filters. Steve KD1JV optimized the board layout to keep the upper band filters close to the PA and lower band filters further.
> I think the problem is the ringing in the gate drive signal. On 20m,How long is the lead from the PA driver chip towards the FET gates? I suppose the lead is pretty short.
> the first "ring" peaks at about 2.5v. According to the BS170 data
> sheet, Vgs(th) is 2.1v typical, 0.8v min. So Q2-Q4 are turning on
> again, allowing them to sink current through L8. This explains both
> the plateau and the delayed peak in the drain voltage. It also
> explains why these FETs get noticeably warmer when transmitting on 20m
> versus other bands.
I don't have the ATS-4, only the ATS-3b. There is a chance that the board has insufficient grounding (ground plane is broken into multiple parts) or the power supply for the PA driver or PA or both is not sufficiently blocked. In first case it helps to unite the ground plane with multiple bridges, see the siver and blue jumpers.
I did those changes to minimize controller noise in receiver during tuning.
In case of insufficient power blocking, try to add some small value (5uF or so) low ESR capacitors across the power leads of the driver chip.
I hope I did not cause more harm and confusion then good with my amateur advices. My profession is software, very far from analog design.
73, Vojtech OK1IAK
- Thanks everyone for the help and suggestions. I got in touch with
Steve KD1JV and he suggested that I re-wind L15 with an extra turn.
That significantly improved my output power and efficiency on 20m. I
also scrunched up the windings on L16 and L18 which gave me a small
additional boost. Here are my latest figures:
15m: 3.9w output power, 520ma input current @12v (63% efficiency)
20m: 5.0w, 840ma (51%)
30m: 4.7w, 610ma (64%)
40m: 4.7w, 600ma (65%)
80m: 5.5w, 660ma (69%)
Scope traces and FFT spectrographs of the output on each band are on
my flickr site now:
I never was able to do anything about the ringing on the PA gates.
After trading e-mails with Steve I think this was just a red herring.
Efficiency on 20m is still strangely a bit lower than the other bands.
I don't know why, but it doesn't seem worth worrying about. However,
if anyone else experiences this and comes up with a way to improve it,
please let me/us know.
So thanks again everyone for all the help, both on and off list. I'm
really looking forward to putting the rig on the air soon, maybe even
this weekend. Let's hope the weather cooperates.
72/73, Dave W2VV