inconsistent radio operation

(14 posts) (6 voices)

Tags:

No tags yet.

flyingkick87

I purchased a jeenode v4 and a jeelink recently. I had the demo working with both of them and proceeded to slowly modify the code to tailor it to my purposes. A few iterations later the radios were still working properly then suddenly stopped. I went back to the demo sketch to start over and now it is not working anymore (the microcontroller is still functional on both sides but I receive no good data through the radio. the jeenode will receive bad data packets at random times but the jeelink won't receive anything) Even though I had not modified the hardware at all, i went back and double checked all the connections on the jeenode and everything looks good. I am running out of ideas to test for the problem and I am baffled because it seems like its something hardware related but I didn't change anything.
Do you have any ideas/tests that could help me narrow down the issue. Could the jeelink be faulty?

Posted 17/06/2010 00:06:08
jcw

> Could the jeelink be faulty?
Can't rule it out. I recently swapped one out which was bad. Best thing to do is go back to the RF12demo v3 code, re-check all the configs, and use the "t" command to send test packets both ways. Make sure you use the proper frequency band, and that the antennas are connected.

Posted 17/06/2010 00:33:22
aromaoftacoma

I've been struggling with a similar issue but maybe not as bad. I purchased a couple of the RF12B,915mhz Jeenodes from Moderndevice. The AVR seems to work fine (all test sketches I build run fine). The radio communication between the two nodes is just really inconsistent. Occasionally the two nodes will communicate like they should (true for RF12demo and pingpong) for a long while but if I touch anything it seems that I'm back to the drawing board. When I say touch I don't mean changing the sketches I mean powering the boards down to move them onto battery power or swapping which one is connected to my laptop, etc. What am I doing wrong or overlooking? It has to be something simple..... Any tips would be appreciated. JC you are doing an awesome job!

Posted 25/11/2010 05:18:51
tankslappa

If physically moving the boards upsets things then that sounds like a dry solder joint. I had fun with one of those on a JeeNode SMD.
I would redo the ones on the RFM12B and ATmega socket.
Good news is you don't need to do them all, if the board is placed with the RFM12B on the right you only need to do the first two on the left and the one on the far right along the bottom. Along the top you need all except the middle one.
On the ATmega you only need to do pins 4,16,17 and 18.
ATmega pin diagram here - http://www.flickr.com/photos/stevesnaps/4840782367
I'd probably redo the regulator and power smoothing capacitor connections too while I had the iron hot.

Posted 25/11/2010 15:44:15

bgreene

dunno if its any help but i had a similar problem with bad crc between 2 jeenodes and was about to bin the whole lot when i realised my problem was probably caused by elv heating controllers. so i changed the receiver to only act on good crc
if ((rf12crc != 0) || ((n != sizeof payload))){ memcpy(&payload, (void) rf12data, n); if (payload.secretId == FMySecretId) { // okay data is definitely valid and for me } // ignore everything else }
this now works purrfectly bill

Posted 25/11/2010 16:47:57

jcw

> if ((rf12crc != 0) || ((n != sizeof payload))){
Shouldn't that be?
if (rf12crc == 0 && n == sizeof payload) {
Posted 25/11/2010 18:06:23

aromaoftacoma Thanks for the responses. I went through and touched up all of the solder joints on the AVR and the RF12B boards initially but the behavior hasn't changed. This morning I re-downloaded the RF12demo to both boards and issued a "t" and got back an "ok". I then tried a couple of random characters with an "a" and got back "ok". I did this a few times with no issues. I moved my FTDI board to the other Jeenode to try the same thing and back to square one..... I can't get an "ok" from either board now. Here are my radio settings and the responses on a "t" and a "0a" issued from the Arduino serial terminal. Are there any know issues with powering down a node, etc? They always know who they are after I drop power and they alway hold onto their sketches but my radio comms just stop working. Current configuration: A i1 g1 @ 915 MHz > 0t -> 66 b > 0a -> 0 b Current configuration: B i2 g1 @ 915 MHz > 0t -> 66 b > 0a -> 0 b Posted 25/11/2010 18:24:38 aromaoftacoma Sorry I forgot a detail. I have checked the power regularly and I'm always at 3.3vdc(+) and over 4.3vdc(P). Posted 25/11/2010 18:28:59 jcw I'm at a loss to explain this. Could you take some pictures and email them? Close-up, whole setup, whatever might help. Posted 25/11/2010 22:16:43 aromaoftacoma Will do. I fired the two nodes up today running the pingpong sketch(I use your blink plug for the send/receive status). They worked perfectly. I left one of the nodes downstairs on my desk and carried the other one around the house to test range. It worked flawlessly and the range was awesome. I power cycled the node I was carrying down four or five times during my tour of the house and it picked the ping back up instantly. When I got back downstairs I powered the node that was sitting on my desk down to see if it would do the same and when I powered it back up they refuse to talk again. Now I at least feel like I have a suspect node. I'm going to throw the suspect node in the fridge for a bit and then try the test again. Maybe it is a connection issue. Posted 26/11/2010 17:56:53 aromaoftacoma I think I may either have a cracked trace or a bad connection on the radio itself. At this point I have retouched every connection on my suspect node twice with an iron, checked all connections point to point with a multimeter and swapped out the 328. If I leave the suspect node powered down for 30 minutes or toss it in the freezer for 20 seconds it fires up fine and they will talk until I power down the suspect node. No complaints about the product at all. Just a bad part I think. It happens..... For the rest of the weekend I'll just keep using my reset process (quick freezer visit) and follow up with Moderndevice next week. I appreciate the suggestions. Posted 26/11/2010 20:49:11 jcw Wow, neat trick! Yep, if freezing consistently makes a difference, then a cracked trace would be my guess too. Posted 26/11/2010 22:20:50 tankslappa It'll work fine outdoors in the UK at the moment then! Brrrrrrr! Well there are only 7 connections to the RFM12B, and they all come direct from the ATmega, so you could have a play with some patch wire. That would prove it once and for all. Posted 26/11/2010 22:53:11 mlucia This is exactly the problem I've seen, but have not tried the freezer yet. I would say that chilling the board should make a cracked trace open as copper shrinks quite a bit. I'd suspect a bad part over a bad trace. If you have some freeze spray you can isolate to a section or part pretty effectively. I will try this and report my results. THanks! Posted 09/02/2011 12:43:28