Lime
SDR Transmit Experiments
|
|
As a receiver.. the Lime SDR-USB,
like most SDRs in general has lots of problems, in part no doubt
generally resulting from square waves used within the processing
circuitry, which result in less than ideal reception. There are
also problems from extraneous noise sources from the computer
needed to work the SDR, its display, power supplies, and other
local noise from electric lights etc etc. All this can often
be ignored in practice. For example you can tune to a really
weak radio beacon and hear it perfectly well.. being completely
oblivious to the zillions of interference spikes and receiver
spurii occupying the spectrum outside the frequency to which
you are tuned. Alas.. although the extraneous noise spikes can
be deemed irrelevant in receive mode, you cannot ignore SDR-generated
spurii when transmitting. For example you may inadvertently transmit
a signal on 121.5MHz (the aviation distress channel) or completely
block an important airband channel used by your local airport.
Below is a picture of my Lime
SDR with the top cover removed. The circuit board has two sets
of receiver inputs, Low, High and Wideband together with two
sets of transmitter outputs Low and High. Careful scrutiny of
the spec of the board (even with factory modifications incorporated)
will reveal that it precludes its use as an HF receiver unless
an upconverter is used, but apparently allows its use as an HF
transmitter, and conveniently, most amateur radio transmit modes
are covered. |
 |
|
The first test was to
set the SDR frequency to exactly 100MHz and look at the output.
In AM and FM the RF output popped up at exactly 100MHz and, using
the drive control provided in the program (SDR
Console Version 3.0.7 produced by Simon Brown G4ELI), I could
set the RF level from -66dBm to +9.9dBm. However, what appeared
on the spectrum analyser screen was not one, but several RF signals.
Looking at the spectrum immediately around 100MHz I found that
besides the correct 100MHz signal I found additional signals
every 250KHz up and down. These were not as strong as the desired
signal and varied in their relative strengths depending on the
drive setting, for example, some 20dB down at each progression
ie. -20dBm @ 100MHz, -40dBm @ 100.25MHz, -60dBm @ 100.50MHz and
-80dBm @ 100.75MHz etc.
These spurii are clearly visible
using a spectrum analyser, but would not be quite as obvious
if one was to use a wattmeter. Many years ago I discovered this
when tweaking the power output from a Microwave
Modules 2m linear. A wattmeter adds up all the RF signals
within its remit and indicates the total power. Effectively,
each individual signal has RF energy which can be converted to
heat and each spurious signal will add its own amount of heat
to that produced by the desired output so a typical wattmeter
displays for example not just the power in a 144MHz signal, but
in addition the power of any signals at 288MHz and 432MHz. Twiddling
a linear might result in 10% more power, but that extra 10% might
be due to power in harmonics. I found that I could readily generate
100 watts from the MM linear but only 80 watts when tuning only
the 144MHz signal. The extra 20 watts resulted from RF output
at 288 and 432MHz. I recall the MM designers warning of this
when discussing their linear back in the 1980s.
I connected my HP431C
wattmeter to the Lime Tx aerial socket and read in excess of
10dBm although the true output should have been 9.9dBm.
In fact, if for example the
spurii from the Lime are each 20dB down on the adjacent lower
frequency, the result of the amount of power in each spurious
signal isn't that great and may be considered as unimportant
except under two circumstances... firstly.. consider what a sensitive
receiver would display on its S-meter. At VHF S9 is reckoned
to be -93dBm and given 6dB per S point which means that if sufficient
attenuation is provided between the Lime TX output connector
and a VHF receiver to produce an S9 reading for our 100MHz signal,
the 100.25MHz spurious output will display a little under S6.
This is not really good enough for a self-respecting radio ham.
Maybe a spurious signal 45dB down is OK (something like S2/S3
compared with S9)? The second instance is when a spurious signal
unintentionally interferes with another spectrum user especially
if it pops up in a sensitive area (eg anywhere in the air-band). |
 |
|
The pictures above and below may
look similar at first sight, but the one above shows the SDR
display immediately around 100MHz when "TX" is selected
at full power, and the second, below, shows signals seen across
the whole spectrum from zero to 1.5GHz. The indication is that
when transmitting a 100MHz signal at 7.82dBm many other signals
may also be transmitted. The chart below lists the harmonics
and its pretty clear that a square wave is involved because odd
harmonics are stronger than even. Note that the definition in
the picture below is not adquate to show the spurious signals
close to the fundamental, shown above. The danger comes when
an untuned power amplifier follows the SDR transmitter. |
 |
|
Table showing the strengths
of the 100MHz transmit signal and its harmonics |
|
Freq MHz |
100 |
200 |
300 |
400 |
500 |
600 |
700 |
800 |
900 |
1000 |
1100 |
1200 |
1300 |
|
Level dBm |
+7.8 |
-28 |
-17 |
-38 |
-16 |
-42 |
-18 |
-45 |
-17 |
-47 |
-22 |
-49 |
-29 |
|
|
|
This isn't the whole story
however, because if we recap and look closer at the transmitted
signal, we see some more spurious signals. In fact the fundamental
frequency plus its harmonics is each accompanied by a group of
signals apparently based on 250KHz.
Below: this signal sitting at
100.25MHz is present at the Lime output socket when either unmodulated
USB or LSB is transmitted at 100MHz.
Because we're using SSB the
main signal is suppressed making spurii more apparent. |
 |
|
Below: if AM or NFM is selected
one sees the picture below. In this example the drive setting
is reduced and shows the 100MHz output adjusted with the drive
setting to -12dBm accompanied by 100.25MHz @ -17dBm, 100.50MHz
@ -50dBm, 100.75MHz @-63dBm.
From my experiments with the
upconverter it almost appears that a 100.25MHz signal is used
to mix a processed RF signal of say 250KHz in a balanced mixer
and what we see above is leakage through the mixer appearing
at the output socket. Non-linearity is also then producing further
signals at 250KHz intervals. |
 |
|
Experiments continue...
next, what about transmitting at HF? |
|
I set the frequency to 10MHz and
pressed "Tx" but not much happened so I tried 30MHz
and saw a little RF output. Clearly the Lime is not going to
be much use as an HF transmitter unless it's used with a downconverter.
Maybe something similar to the upconverter although of course
for amateur radio use the transmit bands are tightly specified.
Perhaps I need to work out the optimum frequencies for amateur
radio use and then choose crystals to convert these down? As
far as the VHF bands are concerned, maybe the Lime will be OK
to drive a linear amp for the particular band. I found a few
very odd things , but I'll just recap on my earlier findings.
The picture below shows a signal of 1GHz from the Lime. It looks
really good. A nicely defined RF spike. I'm feeding the Lime
output through a 20dB attenuator so amplitude-wise it will be
about +2dBm. But is all what it seems? See the second picture
below this one... |
 |
 |
|
Above is a sweep of +/-
500KHz about 1GHz and you can see not a single clean spike of
RF, but three spikes. The two unwanted signals are precisely
+/- 250KHz from the main output signal. In fact the characteristics
of the Lime output don't really differ from the previous tests
at 100MHz except that the relative amplitudes of the spurious
signals differ slightly. Sometimes the upper signal is bigger
than the lower and sometimes smaller. |
|
As an aside... I carried
out additional testing using a newer version of the program.
This was Version 3.0.8 and installed on a second computer. The
results were weird and were due to a problem with the software.
Not in the operational coding, but in the installation. Maybe
a stray file left over from an earlier build? The symptom was
associated with a provision for offsetting the transmit frequency
but after uninstalling and re-installing the problem disappeared
and all was well. As a rule of thumb maybe it's worth re-installing
your program if strange results are apparent? |
|
continues.... |
|
|
|
|
|
|