Not receiving? Add a Reset button

Jee Labs Talk » How can I fix ... ?

Not receiving? Add a Reset button

(20 posts) (7 voices)

Tags:

No tags yet.

petermac

Sharing this solution, in case someone else has the same problem...

I've got two JeeNodes. One is cyclically transmitting a RemotePort "on" for a few seconds, then "off". The other listens, and when it receives the "on", lights a signal LED, and also cycles a group of 16 LEDs through a 5940 driver. (The code is a revised version of the blinkxmit Ports demo).

Worked fine, BUT ONLY if the receiving JeeNode was on USB (a USB-Bub), and the Arduino Serial Monitor had been opened. Whaa? In other words, just powering-up the receiver by plugging in the USB was not enuf to have the radio hear anything (or that's my theory). The signal LED would not light. All the other LEDs are cycling fine on powerup, so the code isn't locked-up. Of course, that also meant I couldn't run the receiver off a battery.

Thanks to some comments in the forum, I've now added a physical reset button between the FTDI pin 6 and ground. After powerup, I just push it, and the program restarts, with the radio responding as expected.

This isnt an issue for the way I plan to use these for now, but eventually I want this to "just work" on powerup. Gotta do more research...

Posted 24/02/2010 22:52:36
jcw

Hm... so the radio is not reliably coming out of reset on power up? I must admit that I think I've seen this happen as well.

Maybe the RF12M needs more time after power has been applied, i.e. add a delay before the rf12initialize or rf12config call? If that's the case, I could insert the delay in rf12initialize...

Posted 24/02/2010 23:48:01
matajumotors
I also had this problem - not initializing rf12 module after power on, i dont now why it is happening, i tried to add different delays before and after initializing with no luck, but i solved it by addding extra "rf12config();" (2x intit); like this:

----------------------------------
void setup() {
```
    Serial.begin(57600);

    Serial.print("\nrooms;

    rf12config();

    rf12config();

    rf12easyInit(5);

    ...

----------------------------------
and now the devices start every time.
```
Posted 25/02/2010 04:40:41
jcw

Interesting. Maybe we need a delay somewhere in the middle of the rf12initialize() code, perhaps a delay between SPI config and the rest, or a quick pulse on the RFM12B enable line?

Please get in touch with me be email if you have a setup which repeatedly fails to power up properly, so we can test some variations.

Posted 25/02/2010 10:55:36
matajumotors

Strange is that the rf12demo code starts every time. Here is simple receiving node code published somewhere in forum. It is not receiving when using 1x rf12config();

http://kaspars.id.lv/rf12b/rfm12bnodeReceiver.pde

node parameters: I i9 g27 @ 868 MHz
I was using Atmega8 and Atmega168, same problem on both.

Posted 25/02/2010 13:01:05
jcw

Hm. Have you tested recently? (your example will need an extra #include "Ports.h" to compile. Maybe I messed something up in recent versions?

Posted 25/02/2010 14:05:08
matajumotors

Yes, it must be there, tested it today, but maby published wrong .pde . But as i have tested - the only "every time" working code from power on was rf12demo. Tomorrow i'll check some more.

Posted 25/02/2010 16:52:42
petermac

I can confirm matajumotors' fix works for me. Thanks.

I'm also putting the double rf12config(); at the very end of my setup(), which puts it about 5 secs after power-up (letting my LED driver initialize first).

rf12initialize followed by rf12config also works.

JC, I should also say this is using the RF12 lib from Dec 9th, on Arduino 17, if that matters at all.

Thanks for this great hardware/software!

Posted 25/02/2010 18:24:28
matajumotors

I just checked, the file was correct, it compiled to me, because i had port and rf12 library files in both "Ports" and "RF12" library folders ;). But still the "no reset" effect is there. I updated the link.

Posted 26/02/2010 06:43:45
stef007

Hi, I can also confirm matajumotors's fix of putting a second call to rf12initialize() after some led blinking (and delay) in the setup() routine. Without this it looks like the RF part is not started.
At first I thought the AVR did not start at all until I found this thread and did some tests.
I must remark that the RF part did not fail to initialize everytime I booted up the jeenode. It did work sometimes. But I could not find a specific pattern.
BTW this was my first post. I was playing around with my jeenode starter kit. Everything went very smooth except for this small hickup. Very impressed and happy with your jeenodes;-)

Posted 05/04/2010 18:56:42
marvings

Oh, i wish i had seen this earlier! I thought i messed up something with the LCD or inside my sketch when fighting with some problems over the easter days!
My nodes worked fine when connected to the computer after starting the serial monitor once but without the computer on external power i had huge problems. Sometimes the nodes would not come up at all, sometimes they did but without receiving RF data. Since the sending node is down two floors inside my mailbox i searched for the reason there - distance too far? Batt empty?
After too many frustrating tries i put everything up in one room and a third node lust listening to RF i zeroed the problem in on the receiving node, my supposed "central" with LCD. As modding the LCD-Libs to not powering on the backlight at startup did not help i found out that after a reset with the button everything worked fine.
Now i know why and later will alter my sketches to include the double init...
jcw, could you please put that into the documentation until resolved so no more people run into this problem?

Posted 06/04/2010 14:12:31
jcw

Done - here and [here](http://cafe.jeelabs.net/sw/library-rf12/) for now, because the docs are currrently being maintained in two places.
Thanks. I really need to dive into this.
In fact I've seen it happen here too, but I never took the time to investigate further when it did.

Posted 06/04/2010 16:25:46
marvings

Perfect, thanks!
Since i have a nice setup to test this here maybe i can help investigating.
We will see later, right now some clothes need my help to get out of the washing machine ;)

Posted 06/04/2010 17:30:20
alanmoore

I haven't had a chance to really test this yet but I suspect that the eepromreadbyte code in the rf12config is failing the crc check (perhaps because of simple startup timing) - and if that happens rf12initialize is never called. Checking the return value of that function (it should not be zero) would show it up, but of course you'd have to flash an LED or something to show the error. The RF12Demo code reinitializes the configuration under these circumstances.
But thanks for pointing this out - I spent most of yesterday evening checking this out - worrying around with a multimeter thinking I wasn't providing enough power with my batteries... :-)
-Alan

Posted 06/04/2010 20:13:56
jcw

That's a good place to start, thx.
If this is the cause, then rf12initialize() ought to always work. And RF12demo always working properly might be caused by the fact that it does a Serial.print() call first, which takes a little time.
Can someone confirm that the problem is gone if you do a serial print (and begin) first? And that the problem is limited to config?
It sounds very plausible. I tend to insert serial prints at the start of every setup(), because I use that in JeeMon, even just for identifying the current code when the sketch isn't going to use the serial port. Would explain why I "almost" never run into it.
Next step will be to find out exactly what the problem is? Delay? Pin I/O not inited yet? Some clock pulses needed on SPI before it works properly?
Hah, this sounds solvable, at last!

Posted 06/04/2010 20:34:56
jcw

SOLVED! I added a rf12xfer() call in rf12initialize. The SPI bus isn't always working properly on the first transfer. I've been able to reproduce the problem, even with rf12initialize(). It wasn't a startup timing problem, as far as I can tell.
Also explains why RF12demo works: it does another set of SPI request to check for the dataflash memory on the JeeLink. Which is enough to make things run properly, even on JeeNodes.
The RF12 driver has been updated in subversion and in the zip file, see http://news.jeelabs.org/code/
I'm not updating the RF12demo flashed onto the ATmega, because it works. But all other sketches should be recompiled to be sure this problem goes away. Please let me know right away if you still run into the RFM12B not initializing properly.
And thanks everyone for patiently adding info to this thread until the solution came in sight!

Posted 06/04/2010 21:01:21
alanmoore

Great stuff....!

Posted 06/04/2010 21:11:11
marvings

> It sounds very plausible. I tend to insert serial prints at the start of every setup(), because I use that in JeeMon, even just for identifying the current code when the sketch isn't going to use the serial port. Would explain why I "almost" never run into it.
And that would also explain why i (now) always ran into it! Since the problem did vanish after starting the serial monitor i removed all serial output from my sketches!
Perfect, that must be the solution!

Posted 07/04/2010 00:49:51
alanmoore

Yes, that fix worked for me also. Nice one.

Posted 07/04/2010 01:35:59
ian

This sounds very like the problem I had with the 'RemoteBlink' example.

Posted 07/04/2010 11:58:31