SparkFun Forums

[rcooke]

Hi Folks,

We need to make a decision on which of the two frequencies we will use in a new product. Does anybody have any words of wisdom on which band you would pick?

We need a worldwide frequency as we want to export the product. We'd like at least a 100 foot range (indoor). It is somewhat cost sensitive (consumer product). We are only controlling remote units - not sending gobs of data. I'd like low power and 3V battery power.

I have read that the 433MHz band is quite crowded in the EU but I thought if we used the nRF905 we could use FHSS.

Any ideas?

[mzoran]

If the remote will be on for more then very brief periods of time you can not use 433MHz in the US. In the US, that band is reserved for brief periodic data transfer such as found in garge door openers, alarms, and weatherstations. The FCC imposes a limit on the percentage of time that the unit can be transmitting. You generally need to use 900MHz or 2.4GHz in these cases and most equiptment that supports changing the channel as required for FHSS is usually 900MHz or 2.4GHz. The difference between the two is that 900MHz has a longer range and is less sensitive to design messups but the frequency range is more crowed and can only support smaller bandwidths. The antenna is larger for 900Mhz.

[rcooke]

Thanks. I believe that for our application - controlling a remote device - the ON time will be very brief. I looked up the 433MHz rules and I believe that it says no more than 10 minutes on for every hour.

The Nordic nRF905 can change channels in the 433 MHz band and I like the range.

Richard

[mzoran]

Actually, I think the rules for 433MHz are quite complex. Apparently, the FCC has a specific interpretation that isn't clearly spelled out in the rules( a.k.a the REAL rules). Here is a guide written by linx who makes cheap 433MHz modules to help customers make it through FCC testing for 433MHz products.

http://www.linxtechnologies.com/documents/AN-00125.pdf

The only other brainstorming thought I have to offer is that the power limits for 433MHz are very small especially if the device falls under part E. As far as I know, unlike 2.4GHz, 433MHz does not have a provision for more power if FHSS is used. The range that is advertised with modules is usually bullshit and only obtainable in very ideal conditions if obtainable at all. I haven't done the field strength to power calculations but I suspect the full 10dBm of the nordic will not be permitted as most 433MHz modules I've seen are more like 0-1dBm.

I've played with the 433MHz modules from spark fun which are a few dBm and had some trouble getting things to work through a few walls especially if something with alot of metal or dense materal such as kitchens, bathrooms, or utility closets were in the path. I tried to use the linx 433MHz modules which are about 1dBm for a telemetry project and the usable range was < 30 feet. The distance of the transmitter and receiver from the ground makes a big difference in the range and published ranges almost always assume at least a 3 foot clearence from the ground. I suspect these reasons are why most newer products with the exception of garage doors openers, alarms, keyless entry systems, etc. use 2.4GHz.

That said, 433MHz does have many advantages such as longer range and the waves don't travel in straight lines as much for comparable powers of 2.4GHz. I've heard 433MHz is much less sensity to bad designs such as poor trace routing, inadequate ground planes, etc so it may indeed be a better match for your product.

BTW, thanks for letting me vent some of the information I've colected over time. Good luck!

[rcooke]

Thanks for the detailed reply. It's funny - I've received responses that say 433MHz is better and others that say 2.4GHz is a better fit. I always felt that RF design was a black art and so far I haven't found anything that changes that.

The one thing that 433MHz has going for it is a competitor uses that band. I think I'll start at 433MHz and if that doesn't work out I'll jumper to 2.4GHz. Most of the firmware will be usable on both so I won't be losing that much time.

Thanks,

Richard

[stevech]

I'd suggest that your use case and desire for low/high NRE should drive this decision. The 433MHz band is intended for super low duty cycle transmissions and very low speeds. And essentially unidirectional data.

If your use case is other than that, I'd go with a 60mW IEEE 802.15.4 radio. Use optional ZigBee only if there is a need for meshing. The 1mW radios are OK for a few tens of ft. indoors. The 60mW with chip (tiny) antennas work for 100+ ft. in an office building with walled offices. Can't beat the ease of use and price of these modules, e.g., Maxstream XBee and many others. Jennic is sold by SFE, but it's not the best value IMO.

[rcooke]

I've looked at the Xbee modules but at 50ma RX current that's a deal breaker for us. The device has to be run on 2 to 3 AA batteries (AAA would be better). They are also a bit expensive.

Our application is very low duty cycle but it is asynchronous so we can't cycle the RX on/off to save power. We only need to send approximately 60 - 70 bits (preamble + address + payload + CRC) and then the receiver sends back a confirmation also of 60 - 70 bits.

[stevech]

Normally, 802.15.4 end nodes are operated in the standard's sleeping modes - - where an end node wakes up every x mSec to see if its parent has queued data link layer frames for it. Same idea as in the power saving mode of 802.11. Done with flag bits in a beacon sent by the parent node - and it is lines-powered. Or there's an option for the non-beacon mode where the end node much less frequently wakes up and sends a poll message to its parent.

But yes, if the data volume is very low and infrequent, then other radios are better suited.

[ke7eha]

for small data throughput, I would look to TI's CC series transciever chips. They are frequency agile, and may operate on the three major ISM bands.

My personal choice, the cc1100 can operate on the three major ISM bands, has a small amount of support passives for impedance matching, decoupling, and the such. It draws low power, increasing battery life, and has a programmable gain up to 10 dBm. The cc1100 communicates through SPI, and has Wake on Radio capabilities.

The CC1100 is a 20 QLP [a quad flat no-lead package] and it is 4mm x 4mm. A datasheet, as well as more detaioled information, may be found at TI's website, conveniently linked here:
http://focus.ti.com/docs/prod/folders/print/cc1100.html

The CC1100 is available for purchase on digikey, for around $5 a unit. IN quantity, it costs approximately $2 a unit.

*edit:

I forgot to mention, the CC1100 is sub-1GHz. This band is more suitable for long-range communications, as you are less limited by line of sight propogation. You won't get sky-wave propogation or ground wave propogation, but the lower frequnecy can bend around obstructions better than the 2.4 GHz communications can. TI also offers 2.4 GHz communications options.

Also, there is the CC1110, a CC1100 with an 8051 core on-die with the transciever.

[brennen]

Just my $0.02, but the nRF24L01 chips from Nordic are extremely low power and operate in the 2.4 GHz band. I have quite a bit of experience with them and they're not very difficult to use as long as you're using a microcontroller with a built-in SPI port. At full power and data rate they use 11.3 mA in TX mode and 12.3 mA in RX mode. In power down mode, they only use 900 nA.