I was trying to find out the radio parameters of a 433 MHz USB dongle.
I wrote some kind of scanner for a blue pill with CC1101 radio module and I got weird results:
When the base frequency is the same, I got smaller RSSI with wider receive filter bandwidth.
I expected the wider bandwidth to give at least the same RSSI as the narrower gave – tha narrower was included into the wider.
Or do the USB dongles have software frequency hopping or something?
Without the dongle I got some frequencies with RSSI about -60 dBm. With the dongle I suddenly got about 60 frequencies with RSSI between -49 and -60 dBm. The dongle was about 6.5 meters away from the CC1101 radio module, but there were no obstacles in between.
That would work provided you have got a $xxK spectrum analyzer and you are watching peaks of your signal with cursors on it, but those RSSIs inside the chips are fairly simple circuits giving you a rough indication of the signal power inside a channel.
My current understanding:
Those RSSIs usually have got the same structure as the popular AD8307 (a broadband logarithmic amplifier).
At the logarithmic amplifier’s output there is a usually a low pass filter with fairly long time constant (T) working as an “signal integrator” (therefore it takes some time to get max RSSI).
Thus the RSSI integrates log values of the input voltage (usually out of IF, or baseband filter) over a period T (an integral from 0 to T of log(u_input) ).
With narrow channels bandwidths (here we talking the bw of the filters at the logamp’s input) you are “closer” to the useful signal and your useful signal fills in the entire channel bandwidth (because the I/Q mixer places your useful signal close around the IF freq), therefore you will integrate bigger values (you get bigger RSSI readings), while with wider bandwidths your useful signal is proportionally smaller (in freq domain it occupies only a part of the channel filter bandwidth, the reminder is the noise which is usually much lower -> while the “peak amplitude” of your signal is still the same your signal’s envelope “is smaller in relation to set filter bandwidth”), and you will therefore integrate proportionally smaller signal with a smaller RSSI reading.
Mind the RSSI output depends on many factors, like antenna gain, LNA gain, I/Q mixer gain, IF gain, SNR, etc.
You may use the readings for relative comparison, like: at this specific chip’s radio setup, modulation scheme, BWs, channel n., antennas positions/radiation diagrams, gains, distances, etc., the transmitter T1 is -30 and the transmitter T2 is -20.
So, the signal from T2 is much stronger than from the T1. That is all..
For precise measurements you would need quite expensive equipment to calibrate your RSSI. The calibration must be done for each specific setup of the chip, frequencies, antenna positions, etc, btw.. A lot of work
[RogerClark – Fri Jul 21, 2017 3:34 am] –
Can I suggest you raise this topic on a TI forum about this device, as the would have specific knowledge on TI products
I did, but I didn’t, really, get an answer.
https://e2e.ti.com/support/wireless_con … 5/t/303845
https://e2e.ti.com/support/wireless_con … 5/t/229044
http://www.ti.com/lit/an/swra114d/swra114d.pdf
What I’m wondering is that very often when I get some data, the RSSI is pretty small. I get better RSSI with smaller RX filter bandwidth, but no data.
What I’d like to know, if there is the same transmission that fits within two different RX filter bandwidths, what happens to the RSSI if you switch from the narrower to the wider bandwidth. If the wider bandwidth is twice the narrower, should the RSSI stay the same or drop by about 3 dBm, or maybe they are totally not comparable?
If RSSI gives a measure of energy within a band, then widening the receive filter should produce higher RSSI,
because the band now contains the same energy as before, but also energy from background noise that didn’t fit within the narrower band.
So would it work if I divided the RSSI by the RXBW? It would then give me some kind of average energy density within the band? One would think that the transmission has higher energy than the background noise (at least in general) and a band closer to the actual transmission frequency band would give higher energy density.
Or maybe dBm is not very suitable measure for that. Maybe (a conversion to) plain (approximate) milliWatts needs to be used?
because the band now contains the same energy as before, but also energy from background noise that didn’t fit within the narrower band.
That would work when your RSSI gives a measure of Energy. But it does not.
RSSI measures voltage (like an oscilloscope) at its input. There is some voltage related to the signal, and some voltage related to the noise.
Imagine you have got an oscilloscope. Time base is set to 1ms. Your signal is 1kHz square wave, duty 50%, amplitude 5V (0..5V) There is some noise on the signal say 10mV. You do not sync, so the signals moves across the screen.
1) what do you mostly see with the 1ms time base? (==1kHz bandwidth). You see a square wave moving around, 1x5V and 1x 0V, the average on the screen is 2.5V. Your signal fully fits into your BW.
2) switch to 1usec time base (==bandwidth 1MHz). What do you see? Noise. Sometimes a small piece of something like a piece of a square wave flies around. The average of noise is zero. You get noise and a little bit of your signal sometimes and from that you see on the screen do make an average.. It will be almost zero.
3) switch to 100ms time base (==bandwidth 10Hz). What do you see? A continual box full of slim square waves (100x5V and 100x0V). The average from the screen is 2.5V.
When messing with radio-electronics you always have to think in both time-domain and frequency-domain and switch in between such it gives you the required result
As such I think it doesn’t matter whether it’s energy or something somehow dependent of the energy…
I don’t, however, understand, how the time domain fits here.
Is it because of the way digital filters work (basically difference equations)?
I’d consider filters more like spectrum analyzers than oscilloscopes. I guess I need to learn more.
I think I need to go through the oscilloscope analogy again.
About the background noise, I also took far-away signals outside the transmit bandwidth as background noise.
The more of those within the receive bandwidth, the higher RSSI?
Maybe I should have used a word like “disturbance” or “interference” instead of “noise”…?
