Hi all,
I have been playing yesterday and today, measuring the signal purity of my
station under many different conditions. The test equipment used is the one I
described a few days ago: A DDS signal generator driving a mixer, which has a
small piece of wire to pick up enough signal, and feeds into the computer's
sound card, where I use FFT software (most of the time, Spectrum Lab by DL4YHF).
As a signal generator I'm using a small battery-powered MP3 player, with a short
piece of cable connecting its headphone output to the radio's mic input, via a
suitable voltage divider. I put several test signal files into the player, the
most useful of which is a dual tone signal with one tone at 734Hz and the other
at 1734HZ. So I get a spacing of 1kHz, but to tones are non-harmonic, so they
don't hide some kinds of spurs that harmonic tones would swamp. The test signal
files were generated in software too.
While I was characterizing the IMD performance of my TS-450 transceiver, before
even switching on the amplifier, I found something that looked strange at first:
In addition to the two tones, and the IMD products, I found additional spurs at
exactly the carrier frequency, 1kHz above the carrier, and 2kHz above the
carrier. The one 1kHz above the carrier was the strongest, only 16dB below the
main tones! The one at 2kHz was much weaker, and the one at the carrier
frequency was even weaker.
To make a long detective story short: The "spur" at the carrier frequency was
actually the carrier, suppressed about 50dB from the max output level. Not
brilliant, but within specs. And the other two spurs were generated in this way:
The two test tones, spaced 1kHz, generate a SSB envelope that is a severely
distorted 1kHz signal. The current consumption of the radio is largely a
function of the output envelope, so that the current from the power supply to
the radio had a strong 1kHz component on it, plus its harmonics. This signal was
feeding into the audio stages of the transmitter, and getting added to the two
test tones! So its 1kHz and 2kHz components appeared at the radio's output,
while the 3kHz and higher components were suppressed below detection level by
the SSB filter.
Most of this unwanted signal was getting in through my external DSP filter.
That's an MFJ-784B, which is powered by the same power supply feeding the radio,
and connects to the radio's speaker output, and to an external speaker. What
happened here is a classical ground loop: Some of the 1kHz-modulated supply
current to the radio was flowing over the ground wire from the power supply to
the DSP filter, then on through the audio cable's shield, directly into the
ground foil of the radio's IF board. Due to imperfect conductivity of that
ground foil, a significant 1kHz audio voltage built up there and fed into the
mic amp circuitry, which is on the same board.
At least that's what I thought at first.
When disconnecting the DSP filter from the power supply, the 1kHz and 2kHz
components in the output get about 20dB weaker, but don't disappear. The same
happens when unplugging the audio cable between the units. Obviously there is
some coupling internal to the radio, making audio components in the supply
current appear added to the modulation signal, although the main spur caused by
this internal coupling is about 50dB down from the carrier, and within the SSB
channel width, so it's not an issue.
But I want to keep using the DSP filter.
I tried using the front panel headphone output instead of the speaker output, to
drive the DSP. It's only slightly better. Installing a short and thick bonding
wire between the DSP's box and the radio's, produced only a slight improvement.
Opening the ground return of the audio cable barely improved the situation, and
made the speaker emit a strong 1kHz tone while transmitting. Cutting the
negative side of the direct wire between the power supply and the DSP, so the
DSP's negative return is only through the radio, didn't cause any change at all
- the offending signal comes in over the positive supply wire and through filter
caps just as readily as over the negative side!
Running the DSP from a separate, ungrounded power supply works fine, but I don't
like that. I hate wall warts! So it looks like I will have to find and install a
nice audio transformer between the radio and the DSP. Or just live with the
problem, anyway no fellow ham has ever reported it! ;-)
Then I switched on the amplifier (NCL-2000), and played a little with it.
Despite running the amplifier at very low idling current, much lower than the
recommended one, the IMD performance is slightly better than when running
without the amp. The reason: When driving the amp, I run the radio at only 30W
output, and at that level the radio is extremely clean (IMD3 around 60dB down),
and the amp adds a normal but not excessive amount of IMD. Instead when running
without the amplifier, I usually run the radio at the 100W level, and there it's
very dirty, with the IMD3 down only 24dB from each tone! And the higher IMD
products, while successively lower, are still significant even up to the 9th and
11th!
But now comes the most interesting discovery of the day, and the reason why I'm
writing all this: Until here I was doing all tests into a dummy load, with my
mixer's pickup antenna running along the coax to the load. I wanted to make sure
that I'm getting no RF from the antenna into any place where it doesn't belong,
so I switched to the antenna, found a clear spot in the upper portion of 40
meters, nestled amidst lots of strong broadcast signals, and made a test
transmission. Suprise! My signal had more spurs installed on it, than a
Christmas tree has stuff attached! And some of that stuff was even moving...
It turns out that the whole mess of strong signals that is in the air gets fed
from the antenna into the final stage, mixes with my transmitted signal, causing
thousands of strong spurs, and all these get re-radiated!
That's life, folks. Transmitter final stages are very big and powerful mixer
stages, connected to antennas through only broadband filters. Except while
driving a final stage deep into saturation, the trash generated by them from
mixing external signals onto new frequencies is FAR stronger than their
internally sourced IMD! It makes one wonder how much sense it makes to strive
for -40dB IMD products in a lab's perfect dummy-loaded environment, if this
won't cure the -20dB spurs caused by external signals as soon as a real antenna
is connected!
Comments, anyone? Do we have a case here for class A final stages, which should
be a lot less prone to mixing external signals? Unfortunately I don't have any
class-A radio at hand, to test this. It would be interesting if somebody who has
one, could do the test, and compare how much external signal mixing happens in
class A as compared to class B.
Manfred
========================
Visit my hobby homepage!
http://ludens.cl
========================
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|