[Amps] MOSFET amp filtering - was: auto-tune

[Jim Brown]

On Wed,12/14/2016 8:12 PM, Gary Schafer wrote:
> Inter modulation is just what the term implies. Two or more signals modulate each other in the signal path. For this to happen there needs to be a non linear device in the path to cause modulation to occur.
>
> A single signal by itself passing thru a non linear amp will generate
> harmonics but not IMD. If a second signal is present it will modulate the
> first signal and the first signal will modulate the second.
>
> Any time you want to transmit more than one signal thru an amplifier the
> amplifier must be linear. Of course SSB contains more than one signal
> (frequency) at any given time as does a two tone signal.

Exactly right. And a keyed CW signal DOES contain many frequencies 
because it is the 100% amplitude modulation of a continuous wave by a 
rectangular wave. Indeed, our use of the words "continuous wave" to 
describe Morse transmitted by keying a radio transmitter is another one 
of those wrong words used to describe things that cause us to 
misunderstand how things work.

I posted this link a few days ago, and I will do so again. Please study 
it. It illustrates many important points.

http://k9yc.com/P3_Spectrum_Measurements.pdf

Please look at slides 8-11, which are a REAL continuous carrier (that 
is, keydown). The sidebands are hum from the rig's power supply.  Now 
look at slides 14-16, which show the spectrum of a series of dits at 
about 30 WPM.

Slides 31-34 show distortion produced in the K3's output stage as a 
function of power supply voltage. Some rigs have this characteristic and 
some do not.

The series of slides beginning with #38 show keying bandwidth of some 
other rigs that were made available for me to test. Note that all have 
keyed bandwidth much greater than the K3 used in the first series of 
slides, but SSB bandwidth is as good or better than the K3. The 
difference in the CW bandwidth is primarily due to the rise and fall 
time and the shaping of the keying waveform, which in a K3 has the shape 
of a raised cosine.

Slides 51-58 show CW bandwidth of an FTDX5000, which was notorious for 
being very broad on CW, before and after application of a firmware 
modification that Yaesu issued to improve this issue.

Slides 69 and 70 compare an ACOM 1010 at its rated power to a Ten Tec 
425 Titan at rated power. Slide 70 can be compared with an earlier slide 
of the K3 barefoot -- bandwidth with the Titan is the same as from the 
K3 barefoot. The same was true with the KPA500 at rated output (500W). 
But the ACOM 1010 is considerably broader at 50 dB down.

What do these slides show?

1) ALL modulation produces sidebands. I learned this in a Communications 
Theory class in 1963 when I studied EE.

2) NO power amplifier is perfectly linear, but some are better than 
others. The KPA500 and Titan are sufficiently linear that they do not 
increasing the keying bandwidth of the rig with the narrowest keying 
bandwidth, while the ACOM 1010 increased the keying bandwidth.

Ah, you say, the difference is 50-60 dB down. Right. But if someone 
using that amp (or one like it) is putting a lot of signal into your RX 
by virtue of his QTH, band conditions, or antenna farm, he can make it 
difficult for you to work someone on an adjacent channel.

3) When we say that an amp is AB or B and that it has perfect linearity, 
we are assuming that it is an ideal amplifier. Almost nothing in our 
world is ideal. Ideal is a simplifying mathematical concept to aid in 
understanding the difference between one method and another.

4) The spectrum produced by a keying waveform is determined by the rise 
and fall times, and by keying speed. I haven't published the results, 
but a year or so ago I measured the keying bandwidth of a K3 with dits 
varied from 20 WPM to 60 WPM. Below 35 WPM, keying speed did not affect 
bandwidth. The increased bandwidth is clearly predicted by 
Communications Theory.

Does this matter on the air?  You betcha it does!  My QTH is 3 miles 
from K6XX, who runs a K3 to vintage tube amps which he carefully tunes 
to minimize the IMD they produce. Another active contester about 15 
miles away uses an IC 7600 and a power amp that is unknown to me. K6XX 
and I can work 500 Hz apart on CW and barely know the other is there. 
But that other contester chews up almost 10 kHz on CW, even though 
inverse square low causes his signal to be 14 dB weaker than K6XX. In a 
contest, within that 10 kHz there are typically 20-30 stations I'd like 
to work but can't.

The same thing happens on SSB. Roughly half of all signals I see 
ragchewing or working nets on 40M are splattering loud enough 5 kHz 
either side of their intentional bandwidth to prevent a QSO on the 
adjacent channel!  This is NOT receiver overload, it is dirty transmitters!

5) All rigs contain an RF output stage, and some are dirty. I've let 
several guys on 6M know that they're splattering, and they told me they 
were running a cheap Yaesu barefoot. My measurements of the PSK31 signal 
varying the power out of a K3 showed that it had far less IMD at 25W 
than at 100W.

These measurements require a spectrum analyzer with rather narrow 
frequency resolution. The resolution of the P3 can be set as low as a 
few Hz. Neither my HP 8590D nor my Rigol analyzer have resolution narrow 
enough to show it, but modern SDR receivers do. The spectrum analyzer 
used for these measurements is an Elecraft P3, looking at the 8 MHz IF 
of a K3 as a front end.  I recently bought several low cost SDRs for use 
as spectrum analyzers! Indeed, all that is required is to drive them so 
that they are within their own linear range.

73, Jim K9YC