RF12 library default settings - increasing range

Jee Labs Talk » Networking & wireless

RF12 library default settings - increasing range

(12 posts) (6 voices)

Tags:

No tags yet.

edwardhorsford

Hi JCW,
I'm currently testing range on a pair of Jeenodes with the aim of getting maximum range.
Whilst browsing the RF12 library, I see several settings selected, and wonder if there was a reason for their specific selection, or whether you just picked one:
rf12xfer(0x94A2); // VDI,FAST,134kHz,0dBm,-91dBm
You've selected:
134kHz out of a range of 67 - 400kHz
0dBm out of a range of 0 - -20dBm
-91dBm out of a range of -73 - -103dBm
Do you have any comments on these selections, or suggestions on what might increase range (I will separately test baud rate too).
---
On a related note, I see you now have a function to send a direct command to the RFM12 -great! Would it be possible to have a companion function which changes the RF12 library default (potentially saving to eeprom)?. Thus rather than editing the library for baud rate, i could save it there.
Perhaps another option would be to make an expanded version of rf12initialize(); which takes baud rate, and other parameters as input.

Posted 07/07/2010 13:54:53
jcw

Most settings were copied from examples I had seen on the web etc. My understanding is that narrowing bandwidth increases range but also limits the baudrate - so it's a trade-off, as most things in life.
I'm hesitant to make these settings too easy to change, because every change makes the transmission incompatible with units running with other settings. I have the impression that the current settings cover the most common straightforward WSN-type use.
My suggestion would be to look in a slightly different way at what you're after: what if you/me/someone creates a sketch which lets you experiment with settings from the serial port / command line, and which can "dump" the current settings back to the serial port, as hex words perhaps. Then all you need is store these words in EEPROM and replay them to recover those settings, or even generate the C calls needed to do so, and then simply embed those calls in your sketches.
There are no doubt more optimized settings (somewhat depending on what you're after, of course). But what you ask is easily accomplished with the current API - the baudrate, for example is a matter of one extra call to rf12control() after calling rf12initialize().

Posted 07/07/2010 14:18:16
mars

I can remember that someone already tested this, ie extended the range by limiting bandwidth and baudrate (down to 4.8k if I recall) so he has now receiption in his house from top to bottom, where this wasn't possible with the default settings.

Posted 07/07/2010 15:48:07
jeroen

Yes Mars, that was me. In have full reception in my house and garden. (with 4.8k) Works fine no problems at all (only thing is after a svn update I have to change the baudrate. Baudrate Must be the same on all units otherwise you don't receive anything) (cost me about 4 hours last week with a new sketch to discover the mistake)

Posted 08/07/2010 07:45:42
jcw

Jeroen: there is now a ](http://cafe.jeelabs.net/man/rf12/rf12control/) function, which you could call right after rf12initialize() to set the baudrate you want. Then RF12 library updates won't affect you.
You do sort of reinforce my point: once you start changing such parameters, it adds to the burden of maintaining everything everywhere... (you also illustrate that there can be good reasons to do so).

Posted 08/07/2010 08:15:00
jeroen

Jean Claude, thanks for the reply. I didn't know that feature. So it something like. word rf12control(rf12xfer 0xC647);

Posted 08/07/2010 08:29:16
jcw
Just add this line after your rf12initialize() or rf12config() call:
```
    rf12control(0xC647);
```
It was added on June 20th - see the change log.
(the SYNOPSIS in the manual page shows the declaration, not the way the function should be called)
Posted 08/07/2010 10:38:04
andras

For those using battery operated rooms nodes, it may be interesting to know how much extra range you get at what cost. I think it would be something like this:
Increasing range by lowering the bitrate 10x like above means the radio has to remain on 10x longer while sending and listening, so that would cost 10x more tranmission energy. In this post: JCW showed that only 20uA of the 90uA total average current draw of a rooms node is due to sending and receiving. At the 10x lower bitrate, the total average current draw would become 270 uA, which is 3x more than the 90uA before.
The receiver sensitivity will become about 8 dBm better at the lower bitrate (also see post ), so you can loose an extra 8dBm of signal power and still have about the same bit error rate. The theoretical range (when not dealing with obstructions) doubles for every 6 dB, so we have a "range improvement" of a factor 2.5 here.
So when you take into account the total power consumption, it's not a bad result I guess. Still you have to change batteries 3x more often ofcourse.
It is also interesting to note that the most "economic gain" is achieved when lowering the bitrate at the higher bitrates. For example when going from 115K2 to 57K6, the receiver sensitivity increases with 6 dBm, so twice the range for twice the transmission energy. When lowering from 2K4 to 1K2, we get less than 1 dBm extra receiver sensitivity; that's only 10% more range at twice the cost.
Ofcourse, this is just theoretical stuff. In real life your mileage may vary due to a lot of other factors. Still, it gives an idea of cost/benefit proportions.

Posted 12/07/2010 22:40:40
ericvrp

It sounds like what we really want is dynamic speed selection based on signal strength or packets loss percentage.
Is there a way to determine signal strength when packets are received? I could use that information for a simple tracking device for my son who likes to run in random directions. :-)

Posted 05/03/2011 10:51:28
jeroen

@eric give the boy a phone with google.com/latitude. Or something less complicated, a chain. ;-)

Posted 05/03/2011 18:03:55
andras

@eric
You can readout the digit RSSI to get an indication of the received signal strength. You have to set the treshold first. When receiving data, the RSSI bit in the status register will theen indicate wether the received signal is above or below the treshold you have set. Here is a link to a post from jcw where you can see how to set the RSSI treshold level and read out the RSSI:

Posted 12/03/2011 20:36:41
ericvrp

@jeroen, those are both good solutions but having a data line open when on holiday will get expensive real soon and the chain will not get through customs. ;-)
@andras, thanks!

Posted 12/03/2011 20:43:39