Trying to understand Listen Mode

[aussieW]

Hi

I am hoping to develop some nodes that use Listen Mode for an application. So far, I am having mixed results. Sometimes nodes wake reliably, but then they go through a phase when they are very difficult to wake. I decided to have a look at what was going on from an RF perspective, so I broke out my HackRF One to capture the signals. So far I am just using the Listen mode examples (slightly modified) from the ListenMode branch. When I send a normal message the signal is centered around 915MHz with approx 100kHz between the upper and lower signals (see screenshot below). When I send a Listen Mode burst it seems to be centered around 919MHz with approx 220kHz between the upper and lower signals (see screenshot below). In addition to the signal at 919MHz there is also a strong signal at 915MHz which appears to have a much wider gap between the upper and lower frequencies (sorry, I didn't measure them). I understand that the ListenMode burst has been offset to minimise impact on nodes communicating in "normal" mode.

So my questions are:
1. Does this look normal for a Listen Mode burst?
2. Could this strong signal at 915MHz be having an impact on the reliability of the acknowledgment being sent from the Node back to the master i.e. is it swamping the return signal?
3. Does the presence of a signal at 915MHz negate the benefits of offsetting the burst?

Looking forwards to your thoughts on this.

[Felix]

That same section contains other radio settings that influence how the signal looks like.
I think you'd want a burst to be offset from the main channel.

Anyway:

Not sure if you're aware but in this thread the Listen Mode is discussed at length and also this thread and the consensus was that there is a bug in hardware causing Listen Mode to sometimes get locked in RX and not wake the node, hence draining the battery.
So you might want to reconsider using ListenMode, or otherwise implement it in a different way - user joelucid also posted a manual implementation (front end code only).

BTW the ListenMode frequency offset is done by this code:

Code: [Select]

  writeReg(REG_FRFMSB, readReg(REG_FRFMSB) + 1);
  writeReg(REG_FRFLSB, readReg(REG_FRFLSB));      // MUST write to LSB to affect change!

[aussieW]

Felix, thanks for your reply.

Yes, I have worked my way through the thread you mentioned, though I must admit I did not absorb it all and don't recall seeing a specific conclusion of a hardware fault but there is a good chance I missed it. Anyway I have bought quite a few of these modules so I would like to continue to experiment if for nothing else other than a learning exercise.

I understand that the listen mode burst is offset as per the code you referenced.  Just to be clear, I am using the current RFM69 library without any modification. My question relates to the fact that the module seems to be transmitting at both the offset frequency of 919MHz as well as the normal frequency of 915MHz at the same time (as per the screen shot in my original post). Now this may be caused harmonics or reflections or bad antenna design or proximity between transmitter and receiver or whatever rather than by the module itself, but I have a very limited understanding of RF so I can't really tell. I just thought it looked a bit strange and may have been causing some of the reliability issues I have been experiencing.

I am not ready to give up on Listen Mode yet as I can see it would be handy in my application so I will continue to experiment/play.

[damonb]

Forgive me if I misunderstand your post, but I suspect what you describe as "transmitting on two frequencies at the same time" is just how FSK modulation appears to a spectrum analyser. Frequency Shift Keying means sending the zeros and ones by transmitting at a defined offset above and below from the nominal centre frequency. Because this happens at the configured bit transmission rate, it is too fast for the spectrum analyser to resolve and display.
In other words, it looks like a normal FSK signal.

[aussieW]

Well I don't profess to be an expert in this subject but here is how I am interpreting what I am seeing.

The attached screenshot called NormalMessage_1.jpg shows a normal message sent at 915MHz. I have marked what I believe to be the upper and lower frequencies of the FSK signal.

The screenshot called ListenModeBurst_1.jpg shows when a burst message is sent. I have circled two signals. I believe the signal circled in blue to be the burst mode message (which seems reasonable given the code Felix referenced in his post). Again I have marked what I believe to be the upper and lower frequencies of the FSK signal. The leaves the signal which I have circled in purple. It appears to be at the base frequency of 915MHz, but much wider and occurs for a longer duration than a normal message. In fact the duration which I have circled in green appears to be the same for the normal message and the burst message. Now if I assume the burst message goes for a second, then so does this mystery signal. My questions are:
1. what is this mystery signal and what is causing it?
2. is it impairing the operation of Listen mode, thus making it less reliable?
3. am I completely wrong about what I am observing?