Odd problem: JeeNode RF12B

Jee Labs Talk » Networking & wireless

Odd problem: JeeNode RF12B

(12 posts) (3 voices)

Tags:

No tags yet.

hacker007
Hi, I have several JeeNodes all power-supplyed via ATX power supply. On the new year's eve and thereafter all nodes started to work oddly. The ATX works great, there was no power oscillation I suppose, only fire crackers were exploding in amount... I use no RTC so there cannot be a new-year bug. Units worked great for months.
RF12B modules seem to hang on loop:
```
    rf12recvDone();
    while (!rf12canSend()) {
      rf12recvDone();
      delay(25);
    }
```
Seems like the unit never comes ready... I had to add a counter inside the loop and when it reaches certain number I reset the RF12B with RF12config(). So it works now.
Did anyone occur the same problem? Any idea what curropted the units?
Posted 02/01/2011 10:42:03
jcw

Why the 25 ms delay? Short packets can come in within 1 or 2 ms.

Posted 03/01/2011 10:38:24
hacker007

No special reason. Just to make it sure. I changed to 2ms. Now my code runs the above loop for 250 times (with 2ms delay) then I reset the RF via Rfconfig(). I have to do it all the time...
Another odd thing is, the response time to a new RF message from my PC was 1 second. Now, after new year's eve the response time is 30 seconds. Seems like the RF12B are running very slow... I did not change the RF12 library at all!!!

Posted 03/01/2011 16:54:41
jcw

Just a check - did you get a new RF12 subversion update recently? (you said you didn't, so I assume that's not it)
I checked in some half-done changes recently. I don't think it broke anything, but then again...

Posted 03/01/2011 19:34:17
hacker007

My Jee libs are from mid december. No recent update. No firmware changes were made when the unit stopped working. I even tried disconnecting units one by one until only one stood, but the problem persists. It does not come out of rf12recvdone() loop unless I add the counter as stated above and exit the loop and software-wise reset the RF12 unit. After the reset I am able to transmit promptly... until the next (transmit) loop. Inbetween I am able to receive messages... Any idea?

Posted 04/01/2011 21:20:38
hacker007

Now when things got worse, I put RF12demo.6 on one of the strange Jeenodes. It starts, but when I send a packet to it via Jeelink (leds on it flash immediately) I have to wait 10 minutes to get ACK back from Jeenode.
Any idea what could curropt all of the unit at the same time? Overvoltage? (I have 10V house supply from PC ATX unit and each of Jeenodes has its own LM7805 to prevent overheating)

Posted 09/01/2011 17:53:50
miker

I have been having a similar problem with a new Jeenode v5. I haven't tracked down the delay to the depth that hacker007 has. I am running the RadioSendConfirm example. The units are hanging on one of the "theRadio.write(0, received)" calls.
I thought I must have made a mistake constructing the new unit so I tested another. Same problem. I went back to one of my v4 units. No problem. Hmm.
Could this be a problem with the RF12 modules in this batch of Jeenodes?
All three of my units are running the same code (I will post a small sample tonight).

Posted 10/01/2011 20:45:07
hacker007

Hi. I have mixed units version 5 and version 4. The problem started when I installed version 5 units. Maybe this is it...

Posted 11/01/2011 16:11:30
miker

I am still trying to come up with simplified example. I have whittled my code down in an effort to demonstrate the problem only to find the problem to have disappeared. This makes me think it might be a memory problem, but I would be surprised because I don't think my code takes up much memory at the moment. Are there any tools to determine how much SRAM is being used?
Still can't reproduce on my JN4.

Posted 11/01/2011 19:11:29
hacker007

I use Memoryfree from arduino site.

Posted 11/01/2011 19:21:07
miker

I ran MemoryFree and I have over 1000 bytes free, so that isn't the problem. It seems like a hardware issue. The radio takes a long time to send a message following a button press (my hardware design includes a button which triggers a specific message when pushed). When I trigger the message through a serial command, I don't have any problems. This makes me think that my pull-down resistor on my button is not large enough. My knowledge of electronics is limited so this is pure speculation.

Posted 12/01/2011 14:32:49
miker

I am pretty sure that I figured out my problem. Perhaps the solution that I used can help.
My problem looks like a hardware problem.
To give you some background, I have a push button wired to my JeeNode which, when pressed, tells the JN to send a sync message to other JNs. Shortly after pressing the button, I was seeing the execution of my code "pause" for 40 seconds or more. When I debugged the code, I found that the execution was off in the Radio object in the write method.
My push button was wired with a 2k pull down resistor. As I began to research pull down and pull up resistors, I discovered that the Atmega328 has internal pull up resistors, so I decided to try that. Sure enough, that fixed my problem.
Not sure exactly why this fixed my problem but I suspect it has something to do with the higher resistance in the internal pull up.

Posted 13/01/2011 03:08:33