Anyone else experiencing this?

[Yveaux]

Hey there!

I run a simple setup: ethernet gateway (Uno, ethernet shield, nRF24), powered 24/7 and a 2xAA battery-powered (no step-up) sensor node (Pro-mini, 3V3 @ 8MHz, Si7020, LDR, nRF24), both running MySensors 1.4b @ 1Mbps.

The gateway I virtually never powercycle and has a static location.

The sensor node I pickup on a daily basis, move it to my work desk, power it, do some tests, power it down and store it again. It runs a simple sketch which switches on a LED, measures 3 sensors, sends them to the gateway, switched off the LED and goes to sleep until woken by the watchdog. Then it repeats. The values are sent to the gatewey with nRF24-ack's enabled and the sensor waits for a reply from the gateway. If that doesn't come it resends its message a number of time befire giving up.

Ok, so far the introduction ;-)

This setup runs fine, generally. The LED indicating wake/transmission blinks for a fraction of a second and then, after the sleep time has elapsed it blinks again.

On some days (!) the LED stays on for a long time (seconds) before going to sleep again. Sometimes it even stays on for many seconds... The nRF24 wireless sniffer I'm working on reveals a lot of CRC errors in the nRF24 packets, which explain the long on-time of the LED. When many CRC errors occur, the sensor will have to resend many times until succesful or even giving up.

The next day this behaviour is normally gone and everything is fine again...

I really do not have a clue what's causing this. I'm starting to suspect the weather/phase of the moon/Russians/... (fill in who/whatever you find suspicious)

Anyone who has a clue what could be causing this or experienced similar behaviour?

[marceltrapman]

No clue what happens to you but this is what I see.

I have a setup of a gateway (nano, radio, serial connection, 24/7 power), a repeater (nano, radio, 24/7 power) and a temperature sensor (pro mini, radio, dallas, 3v batteries).

I am testing the battery level and move the sensor while testing it.
I see a lot of fails when I have it next to me on the coach while testing (gw direct).
I see no fails when I lift it about 30 cm (gw direct).
I see no fails when I have it on my workbench (repeater).
When I disconnect it and bring it back to its normal location the sensor sometimes connects and sometimes (like right now) refuses to connect again.
I have no clue what is going on (because I am waiting for your sniffer @Yveaux :)) except that I know that the current level is ok.

[hek]

Your neighbor is a frequent microwave oven user?

Maybe you could log something to your eeprom to see how frequent it is.

[Yveaux]

@marceltrapman said:

because I am waiting for your sniffer @Yveaux

Just get it from my GitHub then ;-) (https://github.com/Yveaux/NRF24_Sniffer)
You have to figure out how it works by yourself until I have some time to blog about it...

Anyway, the nRF24 seems to be very sensitive to location/orientation then?! It's true that I just 'throw' it on the table an the orientation will be different eveyday.
Is that the right conclusion? Really?

@hek I'm sure they love their microwave but to use it one day for the whole evening and the other day not at all seems very unlikely to me...

[Damme]

@Yveaux I tied your sniffer, I got so far I was going to specify user DLT user 147, but I couln't figure out what plugin to use to what protocol and sizes.. If you could help with that I would be greatful!

[Yveaux]

@Damme I will I will... Just find some ime to blog about it...
Today I made radio parameters configurable, which is an absolute must IMHO.
The code is almost in a state I dare releasing it to all of you, so It won't be long hopefully...

[Damme]

And btw, I experience myself wierd transmission loss, it can work 100% with no packetloss for 30-40h straight, and then not working for ~1-120min and then work fine again.. I ´can't figure out why.. I havent looked into the raw data yet.

@Yveaux just post a screenshot on the DLT user dialog.. :hamburger:

[Yveaux]

@Damme You got quite far actually! Make sure DLT user 0 (DLT=147) payload protocol is set to nrf24, header size 0, trailer size 0 and copy the right MySensors dissector (mysensors1.dll for v1.3 or mysensors2.dll for v1.4) into your plugins directory (next to nrf24.dll). Better to have only one mysensors plugin as the dissectors cannot reliably determine the protocol version.

[Damme]

@Yveaux Aha! I was almost there, I had both mysensor dll loaded.. :) Thanks dude!

[Yveaux]

@Yveaux said:

as the dissectors cannot reliably determine the protocol version

@hek shuffled the header bytes around going from 1.3 to 1.4 and that killed my heuristic dissectors ;-(

[Yveaux]

@Damme You seem to be an experineced Wireshark user/developer, right? This stuff is far from trivial...

[Yveaux]

@Damme said:

work 100% with no packetloss for 30-40h straight, and then not working for ~1-120min and then work fine again

Could be the same issue, though. My hobbying evenings are not that long nowadays...

[Damme]

@yveayx continue here..:) http://forum.mysensors.org/topic/229/your-workshop/17

[Zeph]

@Yveaux

Anyway, the nRF24 seems to be very sensitive to location/orientation then?! It's true that I just 'throw' it on the table an the orientation will be different eveyday.
Is that the right conclusion? Really?

See the antenna patterns @a-lurker found in an ap note for a similar meandering inverted F antenna: http://www.ti.com/lit/an/swra117d/swra117d.pdf (scroll down to the polar charts).

There are 6 charts with two each (horiz and vertical polarization) for each of the three planes. Our nRF is probably different in detail but similar overall.

So just rotating the nRF 20 degree along some axis might well significantly reduce or enhance its antenna gain. This is without any question of reflections or interference.