Smart v5 vs Blog v3 vs Nooelec Mini quick test resulst

[Greg Troxel]

Recently I asked for advice. I had already just about decided to get a
NooElec SMART v4, and the group talked me into also getting a Blog v3
(thank you, that was the right decision). When the dongles arrived the
NooElec SMART is labeled v5, but it seems to be pretty much as expected.
So I now have a Smart v5, Blog v3, an earlier SmarTEE v2 and a NooElec
Mini and Mini2. So this is 3 units that meet the current definition of
good (TCXO, metal case, low noise electronics) and 2 early-days units.

I ran some tests at 915 Mhz, which seems the hardest ISM band, trying to
hold as much constant as I could, while staying sane. This meant:

5 emitters, all Ecowitt. 2 air quality, 1 in, 1 out, and 3 outside
soil moisture sensors.

Same computer, same USB extension cable, same rtl_433 build and args.

Same antenna position, same emitter positions. Humans inside and
wildlife outside both wandering.

Same magmount antenna for the good units (that came with the Smart v5,
thicker coax than RG174). Using the magmount that came with each of
the lesser units, but swapping in the radiating element from the v5 as
happily they all have the same thread.

At least an hour of collection per dongle.

I intend to write up results with graphs as a web page, but the bottom
line is pretty simple and this is the first cut.

Signal strength patterns (snr, signal, noise) similar for the three
good units. Maybe SMART v5 has lower noise level. SMART v5 and Blog
v3 have similar signal levels, edge to Blog. And SmarTEE v2 slightly
lower signal levels on one outside moisture sensor, but higher signal
levels on another. Good decodes all around.

Markedly higher noise and lower signal levels for the Mini and Mini2,
but really it is mostly good enough for this.

Frequency *stability* was good for all of these.

Frequency accuracy was amazing good for the 3 good units and bad for
the other two, to the point where you really can't use them without
going through calibration.

I wrote code to average the frequencies of each emitter and then all
of them. Using the indoor emitter and the group mean got
essentially the same results.

I assumed that the mean LO frequency of the 3 good ones was correct,
which corresponds to my indoor emitter being at 914.995 MHz. This
is of course equivalent to assuming the emitter is right and the
mean LO frequency being 5 kHz high. (I might be able to actually
measure this but it's not easy.)

The deviations from "good group" mean were:
Smart v5: -0.5 ppm
Blog v3: +1.0 ppm
Smart v2: -0.6 ppm
and based on my radio experience that is stunningly good at the $30
price point.

The NooElec Mini reported the emitter at 914.928 MHz, which means
the LO is high, and the reported frequency is -73 ppm from correct
(correct being the average of the 3 good ones). With no
calibration, I was missing decodes on one of my emitters. I didn't
want to mess with the cal arg, so I just changed the frequency. I
think this should just move the passband and not change the reported
values, but it did achnge them. The reported low frequency was
reduced a little and the high was reduced more, maybe 1 kHz and 10
kHz. Decodes of my lowest-frequency emitter started working.

The NooElec Mini2 reported the emitter at 914.958 MHz, which means
the LO is high, and the reported frequency is -40 ppm from correct
(correct being the average of the 3 good ones), but the full group
average results said -51 instead. I did a non-recorded few-minute
run and picked a frequency by eyeball, to avoid missing data.

So, if you have any non-Blog units older than the Smart generation, you
may want to upgrade. But, if you don't need super low signal
performance, and you do a calibration run, the older ones are ok.

Not that the SNR etc. are self-reported and the metric that really
matters is decodes. There is more to understand here, especially AGC
behavior.

In the graphs:

emitters have colors, by a hard-coded lookup table from type/id to
codepoint, which is consistent between the two graphs. Green is
indoor air quality, red outside air quality, and blue, yellow, and
purple are three soil moisture monitors. The blue one is two years
old with the original battery and the other two were just installed.

Vertical lines denote dongle changes. The order is
Smart v5
Blog v3
SmarTEE v2
Mini (black plastic case)
Mini2 (blue plastic case)

Signal graph shows SNR (square, 20 to 40ish), signal (x, below 0), and
noise (-35ish).

Frequency graph shows freq1/freq2 and a synthetic center freq.
(Modulation is FSK so freq1/freq2 are reported.)

The Mini has a frequency jump because I stopped and then restarted
with -f 914.936M. I cannot explain why the jump is different across
the two FSK channels.

I cannot explain the mini2 startup transient in signal strength on the
purple emitter.

[Michael Cadiere]

Read a few post here and there that the v5 is able to do direct sampling unlike 

The v4