Hello - newcomer here. I'm trying to set up some JeeNode USBs (v2) to measure river water heights each side of a sluice gate. It turns out that my preferred loations are at the limits of the RF range back to my house, so I am looking to increase RF range. Has anyone experimented with using half or full wavelength antennae to see what increase can be achieved? I also wish to hide my JeeNodes a little to avoid theft - has anyone managed to site the antennae a few metres away from the JeeNode via some kind of coax cable?
Thanks
Nick
RF range
(24 posts) (12 voices)-
Posted 07/01/2010 14:46:16
-
This is where mesh communication would pay off, which the RF12 driver doesn't do.
I'm hitting a similar limitation in a different context: can't get a signal from the garage, which is two reinforced-concrete floors away from the JeeLink in my office. One plan is to explore a simple relay placed halfway: receive packets in one network group, and resend them in another network group. Don't know yet how to handle acks, but maybe for simple broadcasting sensors with central acks it's all doable. Needs more thought.
Another option is try improving all the antenna's. I'm out of my league here, but there are directional antenna's and tuned antenna's. Maybe someone with more RF design experience can shed some light on how to go about impedance matching, connecting cables, and optimal matching of antennas... I admit that I stopped tinkering once the wire antenna worked, but surely there is room for improvement.
Posted 07/01/2010 15:30:15 -
An interesting test would be to use a PCB edge mounted SMA connector and see how it compares to the wire antenna. You can see pics of someone using it here: http://blog.strobotics.com.au/2009/05/15/strobit-wireless-widget-photos/
Posted 10/01/2010 20:39:15 -
NickR,
have you looked at directional antennas for the frequency your RFM12s operate at?.With "coax cable", one tends to limit the cable length to a minimum to avoid signal loss, though the loss can be compensated for by gains in the sending and receiving antennas.
Posted 14/01/2010 08:08:20 -
My plan is to first test 1/4 vs full wavelength antennae, and then if still unhappy to try to build a "cantenna" (although a cantenna would be low priority compared to getting one site working fully, ie parallax ping -> jeenode -> jeenode -> web server (via php & mysql)). I hope to try the antenna length tomorrow, but no promises; however whenever I do I shall report the results here.
(The good news is that since my last post I have managed to get a parallax ping ultrasonic rangemeter working, and relaying over rf between nodes - not much to brag about, but fantastic progress for me! )
Nick
Posted 16/01/2010 22:20:59 -
OK - I've done a simple test. A jeenode USB v2 with an 85mm antenna competing with a jeenode v4 with a 345mm antenna. Both running the same sketch, both powered by 2 brand new AA batteries out of the same packet, both radioing a test signal back to a jeelink every 5 seconds.
And the result is that when I took them to a distance where the signal was intermittent (unfortunately roughly where I need to site one of my water level sensors) there was NO DISCERNABLE DIFFERENCE in signal quality between the two.It may be worth noting that before I got to this point I made a lot of mistakes with my experimental setup, and as a result I suspect two other factors are important:
1) I think I got a little more signal with 3AA batteries (as opposed to two);
2) I think I got a better signal when all the antennae were absolutely straight and vertical.
But both these points need more experimentation. Also did I read somewhere that we can control the signal strength of the RF12? And of course the "cantenna" needs trying at some point.For now I aim to move my receiver a bit closer to the transmitter, and get on with php and mysql and getting some kind of basic server up and running (the river has been up and down like a yoyo these last 24 hours and I had to get out of bed at 2am to check it, so I could really do with getting at least one station fully working asap).
All comments/suggestions gratefully received,
Nick
Posted 17/01/2010 12:14:28 -
Quick experiment to get max range with what you have might be to add a ground plane: straight wire pointing up from the receiver, with entire JeeNode placed on a conductive ground plane (without shorting out any pins of course).
In ASCII art form: [code]|[/code]
Should get the most out of a simple quarter-wave wire antenna.
PS. indeed, I'd use 3 AA's, so that the voltage regulator does produce 3.3V, not less.
Posted 17/01/2010 15:27:21 -
Thanks JC; I'm not following you about the detals of the ground plane - do I simply stand the jeeNode on a sheet of polythene and that on a sheet of metal? Do I connect the metal to a ground pin on the jeenode? Does the sheet have to be any particular size?
NickPosted 17/01/2010 16:24:48 -
I'm not sure either :) - but yes, that's what I would try. Grounded to the JeeNode ground. Perhaps A4 sized, the larger the better as I understand it (at least half a wavelength on each side would be my guess). Take a baking plate or something, just to see whether it has a significant effect - you can always tweak and come up with a more elegant solution later. I'd try that on both ends, in fact.
Posted 17/01/2010 16:45:15 -
Here's a link with antenna design notes for the RFM12's:
http://www.hoperf.com/upfile/antennasmodule.pdfAlso this one:
http://www.hoperf.com/upfile/ANTENNAS1212b.pdfLower baud rates get longer range according to HopeRF. Not sure if this always works out on FSK, though. Maybe you can slow down the RFM12B baud and extend your reach. Maybe you already tried this, though.
RFM12BP is a higher power module which might help. Bit late now, maybe.
A repeater sounds like a goode idea.
And I really like the idea of a mesh. Any plans for that, JC?
Posted 17/01/2010 21:22:04 -
@jcw - I was also thinking of using a simple relay from one group to the other for those hard-to-reach corners. However I don't understand how this can be achieved. The group code is put in the synchron pattern register of the RF12B. Is it correct that the RF12B then filters out messages for which the group code matches the synchron pattern ? For a group relay function one would like to be able to receive messages from 2 groups.
Posted 24/03/2010 09:21:36 -
The RFM12B support sync on a single byte. I haven't added support for this in the RF12 driver, though. I'll look into this, but it'll probably be next week before I do.
Posted 24/03/2010 11:47:37 -
I did some range checks with my nodes these days and even managed to make one talk to me out of my mailbox quite good - and that's two floors and lots of concrete away!
I hat to solder a coax cable to the node and attach a homemade antenna to get the signal out of the metal mailbox but i had anticipated that the cable-antenna could not get out of that Faraday cage anyway ;)
While it works already i still want to explore how a reduced baudrate improves rf range. But i did not find how to set that baudrate except for
>> rf12xfer(0xC606); // 57.6Kbps (38.4: 8, 19.2: 11, 9.6: 23, 4.8: 47)
in void rf12initialize of RF12.cpp but that's no parameter of that void.
I guess i would have to reduce the baudrate of my whole network anyway so should i patch RF12 for this test? Or would this possibly cause problems somewhere else because the timing is different?
Did anyone else try out 9600 baud yet?
Posted 25/03/2010 11:36:28 -
Hi Marvin - I'm using a patched version of the RF12 library running at 115.7Kbps - seems to work well, though range is much reduced. I should note however that the "t" command of the rf12demo doesn't work, indicating something isn't too happy with this faster speed.
I've also taken it down to 4.8Kbps, though I don't recall range being much better than the default 57.kKbps. I should add that my tests were very informal, and I'm mostly interested in high speed communication rather than range.
I did try using a single sync byte, but seemed to start picking up a lot of junk that just wasn't worth it.
If Jean-Claude is going to look at this area again, I'd hope the two sync bytes are kept, and instead two additional bytes added to the packet. These could accommodate source AND destination addresses. If someone still wants to be able to 'listen' to different groups, they could always assign addresses in groups: ie group 1: 10-19; group 2: 20 -29, etc. Then each node just needs to mask for the group number in the address. This change might make the wireless protocols incompatible, but potentially wouldn't be too hard to integrate.
Posted 25/03/2010 13:17:32 -
My hunch is that lower baudrates will only give better range if you also lower the bandwidth used (it's one of the RFM12B config settings - haven't looked into it yet).
The change you describe for different groups could be accomplished with today's RF12 code: put everything into a single group, say 123, then add those extra bytes to the payload. Then it shouldn't be to hard to implement a relay/repeater which knows about these conventions and does the repeater thing. With as small drawback that all the nodes need to be "in" on this specific scheme (perhaps using a small C-level wrapper around rf12recvDone/rf12sendStart to handle the prefix).
I'll see if I can get some example code going next week.
Posted 25/03/2010 13:37:29 -
Okay, i will wait on that topic. Too many simulataneous open ideas anyway - need to close some ;)
With the external antenna everything will be fine for a start. I everything is working with real data i might get back to this and try a patched version with lower baudrate (and manybe bandwidth)...
Posted 25/03/2010 21:26:30 -
The RFM12B datasheet doesn't provide very much information on this, but the Si4421 datasheet (on which the RFM12B is supposedly based, see http://talk.jeelabs.net/topic/182#post-893 ) does have some interesting information on this.
If you look at page 37, you can see from the graph there that decreasing the bitrate from 57K6 to 1K2 will yield an increased sensitivity (if we accept BER=10E-3) of 11dBm. That's a significant improvement. So if a low datarate is no problem, then this is an easy way to get more range.
Also on this page a table if provided with the optimal bandwidth filter settings for a given datarate.
Posted 26/03/2010 10:01:17 -
Wow, that's very detailed Information missing in the Files of HopeRF! The Module is capable of diversity antennas which could also improve range even with printed antennas!
But that's too far from here and maybe not possible any more when using the Hope-Boards...
Since my (still to document...) external antenna works and i could possibly still squeeze out some more range by reducing the baudrate i am done for now but will get back to some measurements with different rates some time.
Posted 26/03/2010 12:05:00 -
I noticed another option if all optimizations don't seem to work. HopeRF seems to have a more powerful solution where they amplify the RF signal. It seems the rest of the device (tranceiver IC) is the same.
Here is the link:
http://www.hoperf.com/rffsk/rfm12bp.htm
Does anyone have experience with it ?.Posted 26/03/2010 16:25:43 -
After reviewing the specs, it looks like these boards are not pin compatible. Pity. It would require some rewiring.
Posted 26/03/2010 17:08:08 -
I've such a module and already did the necessary rewiring last weekend. I did build some kind of PowerRF-Plug (-,. But unfortunately I didn't find the time to test this setup. As soon as I got some spare time I'll do some tests and report about how it works.
Posted 26/03/2010 19:48:14 -
@MarvinGS: would love to know more about your antenna design. how is the coax cable incorporated? if not an image, then perhaps a brief written description?
Sincere Thanks.
Posted 29/12/2010 19:35:03 -
The spec for the RF says it support a diversity antenna. Its seems range could be extended alot with a sound antenna - but I cant figure out how to connect it. Ive seen 433mhz products going claiming 500 meter range (http://www.sparkfun.com/products/155) with 10mw output - and that must mostly be because of a sound antenna design.
Posted 15/01/2011 10:49:24 -
For lack of a better way, I'd try soldering a coax to the GND and antenna pins/pads, and then experiment with proper connectors and antenna's on the other end of the cable. I know little about antenna design though, so this might not be very good advice...
Posted 15/01/2011 11:45:01