Archived Forum

We have never noticed a difference in range. I think due to physics the greater the frequency the better the penetration. All devices on the ISM bands should only transmit for a few mS. We have not had much experience using the OEM system in areas with busy RF's. Most of our deployments so far have been quite rural. 

Hi.

I've been searching here and I couldn't find any order of magnitude for the range of the RF link between emonTX and emonBase, be it here, or there, for instance.

What kind of range can you expect ? Could this be a concern in big buildings, or large houses with thick concrete walls ?

Thanks for any feedback.

Take a look here: http://jeelabs.net/projects/cafe/wiki/FAQ

Note also that you should be able to improve the range by reducing the bit rate. There's been a recent discussion mention about this, sorry but I can't find it just now. here: http://openenergymonitor.org/emon/node/1535#comment-8237

For the specific example we are talking about, lower frequency equals better penetration. See page 43 here:

http://www.telit.com/module/infopool/download.php?id=3636

Not sure you would notice the difference in your average house though. Might depend more on what the neighbours are using. The variables are many from what I can tell.

CAUTION: That data sheet is NOT for the RFM12B module!

Though the table of losses will apply, it is still necessary to take antenna performance into account to get any sort of estimate of the reliable range - and I cannot recollect seeing that for the RFM12B.

Hi.

Thank you guys for these answers. I was concerned about the transmission in a house through several walls and ceilings. I guess it's generally fine.

Besides, lowering the bitrate should not be an issue when monitoring a house : most measured values are quite low-freq and you don't need thousands of samples by second (unless you are really interested in electricity peak consumption, maybe).

I discovered the range improvement with lower bit rate after replacing a Maplin monitor with emon. Both systems use a version of the RFM12 module but the Maplin system had far greater range so I disassembled the PIC code in the Maplin transmitter and found that the only difference was the bit rate. That's how I ended up with the odd value of 3.918kb/s, it was the value used in the Maplin unit. I didn't take this further to see if lowering the bit rate even more would increase the range since this was enough for my house.

Does slowing down transmission also improve the reliability (so ACK's won't be necessary) ?
Maybe interesting to do some more range/reliability tests with this. OEM packets are so small transmission speed isn't an issue here (according to me).

Is it 4 kilobytes/s or 4 kilobits/s?

bits

Just checked the regs for the 433MHz band and they allow a 10% transmit duty cycle so no legal issue there.

The 868MHz rules are much more complicated but seem to be either 0.1% or 1%

document is here if anyone wants to check for themselves....

http://stakeholders.ofcom.org.uk/binaries/spectrum/spectrum-policy-area/...

B/s = bytes per second, b/s = bits/second.  (B is bigger than b !). No, I don't like it either, but that's the way it seems to be done.

@MartinR:

Lowering the bitrate will in essence reduce the required RF-signal bandwidth .

The range improvement is due to the fact that available  RF-power is divided on the bandwidth. (A narrow signal will travel further than a broader one, if the power is constant.)

If you need (or want) both power and bandwidth to be untouched, you'll have to change antenna.

Hi all.

I'm experiencing range issues.

I get a nice connection inside my house, even if the signal has to cross brick walls. However, when it is in or close to the gas meter box outside the house, the base (Raspberry Pi) doesn't get its signal, although it's only a wooden garage door, a caravan, and a not so dense hedge to cross (Unbelievably, it even works just before I cross the hedge... actually it works pretty much everywhere, except where I need it to...). It can cross a longer distance if moving aside to avoid the caravan and the hedge (but moving the caravan or the meter box is not an option).

I suspect the cold affects the battery and I'm working on this (in another thread), but in the meantime, I'm following other leads. Besides, I'm not sure only the temperature matters because it also fails even if it's been moved out of the house a minute ago.

It used to work, though, and I don't know what changed since, apart from the temperature. And it sometimes work but with a log of missed frames. I don't think I can quantify this any better, but obviously, I'm reaching the limits of the range.

Antenna

I remember that I broke the end of the RF cable on the RFM2Pi board or on the TX. I had to cut a few millimeters and solder it back on. Does that matter ? I mean does the antenna length precision matter that much ?

Bitrate

Back to this thread's topic. I'd like to try lowering the bitrate.

On the TX side.

I modify the jeelib library. In RF12.cpp, in function rf12_initialize, I modify line

rf12_xfer(0xC606); // approx 49.2 Kbps, i.e. 10000/29/(1+6) Kbps

into

rf12_xfer(RF12_DATA_RATE_3); // Approx   4800 bps

(BTW, I don't know why they're not using the constants defined in RF12.h.)

Now I don't get anything, but I think it is because both the TX and the base need to use the same bitrate. Seems obvious, but if anyone can confirm...

On the emonBase side.

I'm using RFM2Piv1, so it seems I'd need to get myself an ISP programmer to reflash the ATTiny... I guess I'd be better off buying an RFM2Piv2 that can be reprogrammed through the Pi.

Power

Apparently, the output power is already set to its maximum:

rf12_xfer(0x9850); // !mp,90kHz,MAX OUT

For anyone interested, I got the datasheet and programming guide here: http://www.hoperf.com/rf/fsk_module/RFM12B.htm

Is anyone else experiencing range issues ? My range is surprisingly bad. There must be a reason. It's very frustrating.

Did anybody try to lower the bitrate already ?

it's not necessary (or advisable) to change the jeelib library, you can set the baud, or any other parameter, within your sketch like this....

rf12_initialize(nodeID, freq, networkGroup);                          // initialize RF
rf12_control(0xc657);  // approx. 3.918kps, i.e. 10000/29(1+0x57) kps

but, as you say, you obviously need to change both the transmitter and the receiver.

Great, thanks Martin for the tip.

Perhaps could you consider adding this as a parameter to the RFM2Pi board. Just like you have b4 for the band, you could have, say, r3 for bit rate = RF_DATA_RATE_3.

I believe the new RFM2Pi is less of a pain to program, but it's even better not to reprogram at all.

Or perhaps is it useless and everyone is happy with the default.

Perhaps could you consider adding this as a parameter to the RFM2Pi board. Just like you have b4 for the band, you could have, say, r3 for bit rate = RF_DATA_RATE_3.

Wrong Martin Jérôme, the RFM2Pi board is Martin Harizanov's project, I'm Martin Roberts. I am however running an RFM12Piv1 (and all my other nodes) at 3918 baud as it's the only way I can get the range I need.

I've attached the ATTiny hex file in case you want to give it a try.

In general (and contrary to the first old response) penetration / distance decreases with increasing frequency.

Lowering the bitrate also increases penetration, as does using an antenna with higher gain, or sending with more power.

Antenna length is not very critical, but its orientation is. You may try rotating the wire or device with its antenna pointing to the receiver or to the sky, and do the same at the receiving side.

You should also see if connecting the ground to your gas meter helps. Any large metal object in the vicinity of sender/receiver can have adverse (or positive) effects. And pointing the antenna the opposite direction of the gas meter. And moving the sender 1/4 or 1/2 wavelength from the gas meter.

Another alternative is to replace the "wire antenna" with a better one. 1/4" whip with a magnetic base could help. Some are delivered with a 3m coax cable, so it could help you to find a "better" spot for the antenna.

I personally had quite a poor experience at 868MHz (but at 38kHz, maybe I should reduce it). One concrete wall/one floor is OK, but 2 floors becomes problematic. Maybe because my floor heating is using a metal grid (Faraday cage)...

Disclaimer : My HW is a bit different, I use panstamps, not RFM12b.

Jerome: I remember that I broke the end of the RF cable on the RFM2Pi board or on the TX. I had to cut a few millimeters and solder it back on. Does that matter ? I mean does the antenna length precision matter that much ?

That same question was asked a while ago here.  I think the outcome was that having the correct length of wire for the aerial is definitely a good idea, but the degree of improvement is difficult to predict. 

Using a proprietory aerial would ensure that the effective length is correct, although there is still the difficulty of making low-loss connections at the RF chip and bulkhead, as has been well discussed.  An effective ground plane can also make a significant improvement to the range that can be obtained.

Right Martin, I was confused with the other Martin.

I should be able to borrow an ISP programmer and hopefully try a lower baud rate soon.

I can also try to "orient" the antenna. It used to work in the same position, so I assume the antenna must have moved a bit.

Thanks for those informations.