Getting poor range with CC1101

I have been experimenting with the 900mhz Anaren module recently.   One thing I noticed is that it appears the low-pass filter on there seems to be pretty agressive.  The output power appears to start dropping off around 915Mhz and continues to drop up to 928Mhz.  I would suggest trying the lower frequencies for a fixed-frequency application.  Note that these observations are not from conducted measurements and only from 2 samples.  I could be wrong about the filter.  Taking a look at the FCC test report for these modules might reveal something for low-mid-high output power measurements.

Another thing you might be interested in trying is a frequency hopping algorithm to give you frequency diversity and interference immunity.  I will be releasing an update to the OpenFHSS project with support for Anaren modules connected to the TI Stellaris parts in the near future.   This allows for fast frequency hopping with 50 channels evenly spaced from 902 to 928mhz with a 5 mS channel dwell time.

Check here for more information and future updates:

http://sourceforge.net/projects/openfhss/

Very interesting Todd, I hadn't thought about the filtering.  Since I have the built-in antenna, that makes it tough to do imperical testing, but I will have to see if I can figure anything out this weekend.  I was digging around on the FCC site a bit and I found the filings for this module (though a quick glance didn't turn anything up, but that Test Report is lengthy and I might have missed something; i've never looked at the FCC stuff before and I am more of a digital background).  The docs are here: https://fjallfoss.fcc.gov/oetcf/eas/reports/ViewExhibitReport.cfm?mode=Exhibits&RequestTimeout=500&calledFromFrame=N&application_id=220423&fcc_id=%27X7J-A10040601%27

I will have to look into the hopping stuff, but I am a little stuck in my current design for now.  Could potentially get me some gain (though I feel like something is missing to begin with that should be there already)!

Ok,  looks like I was wrong about the low-pass filter, but right about the power output for the module with integrated pcb antenna.   Their module with external antenna does not show the problem.   Looks like the pcb antenna is the culprit.

I downloaded the test report.   There is a difference of almost 9dB between low and high channels in peak measurements for the module with pcb antenna.   You will more than double your communications range by using the low channel vs high assuming all other things (interference / rx sensitivity / etc) are equal.

Summary of peak power measurements from the test report:

Anaren module peak field strength measurements with 2dBi pcb antenna

907.778Mhz low channel fundamental peak horiz    91.34 dBuV
907.778Mhz low channel fundamental peak vert     81.21 dBuV

914.827Mhz mid channel  fundamental peak horiz   88.34 dBuV
914.827Mhz mid channel  fundamental peak vert    77.07 dBuV

922.046Mhz high channel  fundamental peak horiz  82.48 dBuV
922.046Mhz high channel  fundamental peak vert   75.10 dBuV

note:

The 9dB spread is only across 14.26 Mhz of the 26Mhz ISM band.   Operating this module from 902 to 928 may result in an even larger difference between low and high channel.

[edit]  swapped high/low channel descriptions due to typo.   Also added vertical measurements from test report

Wow, nice find (I probably would have glossed over those points, which was obvious since I spent 15 minutes and didn't see it)! 

One point I am missing though, what do you mean by low channel versus high channel (I saw similar comments in the test report).  I would imagine it means that low end and high end of the band this operates in, but that obviously isn't right since the high chan is at 907 and the low is at 922. 

Since I was just using the Anaren values for setting up the registers for 1.2kbaud 2FSK, I assume I need to go in and make the adjustment to the low channel values you quote above?  I guess modifying the FREQ0 1 and 2 register and leave the rest alone?

I have low and high swapped in my previous post.   I'll correct that.   Thanks for catching that and preventing someone else from getting confused.   I'm not familiar with Anaren's firmware, but it appears that you want to use the lowest frequency possible in a fixed-frequency application  (assuming that interference is not a problem).  

I should also point out that even with the uneven power output, I really like these modules.   Good deal for < $15.  You get a pre-certified module and the ability to connect your MCU of choice.   Can't beat that when testing costs around $10k for FCC.   I did some range testing for a project using the frequency hopping library I mentioned with the Anaren pcb antenna module and compared it to a 2.4GHz Zibbit module that is in a current design.   The Anaren module definitely won the test and we will be integrating it into the next round of hardware development.

No problem, though I didn't "catch" anything, I just thought I didn't understand it!

But now I am still confused, you say that we will want to use the lowest freq possible, but the low freq (which is the low channel) is the one putting out 9dBuV less, which means it is putting out 9dBm less, right?  So if that is the case, I would want to be using the highest freqs for the best gain, right?

I like them too.  They just worked out of the box, and once you get some basic code running to talk to them, it is pretty straightforward (and I was using a lowly PIC).  I liked the FCC-preapproval as well, though it does appear goign with the built in loop antenna might not have been my best choice.

What kind of ranges were you getting in your testing?  Like I said, I was a little dissapointed that I seemed to drop out at roughly 500ft.  I assume if I can get something close to 6dBm gain though, I might be able to double the distance (though not counting on it).

I updated previous post with typo fixes and also add vertical measurements from the test report.   The difference between low and high channels with vertical orientation is 6.1dB,  so not as bad as horizontal orientation, but low channel will still have double the potential range.

toozie21 said:

But now I am still confused, you say that we will want to use the lowest freq possible, but the low freq (which is the low channel) is the one putting out 9dBuV less, which means it is putting out 9dBm less, right?  So if that is the case, I would want to be using the highest freqs for the best gain, right?

I had that table messed up more than just swapping high and low.   I went back and carefully pulled values from the test report this time,  hopefully I got it right this time.

toozie21 said:

What kind of ranges were you getting in your testing?  Like I said, I was a little dissapointed that I seemed to drop out at roughly 500ft.  I assume if I can get something close to 6dBm gain though, I might be able to double the distance (though not counting on it).

Our application utilizing the Anaren module is not dependent on a long communications range, so I wasn't doing absolute distance measurements.   These were just some real-world type tests to evaluate the relative performance between the two systems.   If you need long range,  then you really should look at getting the modules with a connector and look for a 1/2 Wave dipole.  One like this should perform really well:

http://www.mouser.com/ProductDetail/Laird-Technologies/0600-00048/?qs=50ljmCSb6%252bVAIXmr14XWUg%3D%3D

I'm going to dig through the test report in the morning and make sure I understand where you got those numbers from (not doubting you, just want to make sure I learn it properly), seems like a good lesson.

Good point on the external antenna, that would probably solve a lot of issues.  You don't happen to know what crystal they are using on these modules, do you?  I am assuming 26MHz, but I can't find it anywhere for sure.  If it is 26MHz, then it seems like instead of running at 915MHz, like they claim, the module is really running at 902.5MHz which puts it way down in the low channel land.

EDIT: Actually, if that is true, then it should be putting out the most it can.  But if the crystal is really something like 26.36MHz, I can suddenly be back up at 915MHz like they claim I am.

toozie21 said:

I'm going to dig through the test report in the morning and make sure I understand where you got those numbers from (not doubting you, just want to make sure I learn it properly), seems like a good lesson.

Good point on the external antenna, that would probably solve a lot of issues.  You don't happen to know what crystal they are using on these modules, do you?  I am assuming 26MHz, but I can't find it anywhere for sure.  If it is 26MHz, then it seems like instead of running at 915MHz, like they claim, the module is really running at 902.5MHz which puts it way down in the low channel land.

Please do double check the numbers.   I just typed it wrong before I went back to double check, but the result is the same.  6 to 9 dB difference between high and low frequencies depending on orientation of the pcb antenna.  The low frequency (907.xx) will perform much better than the high frequency  (922.xx).   I knew this from observing the signal on a spectrum analyzer while operating the module (pcb antenna) in a 50 channel frequency hopping mode before looking at the test report.   I am glad to see that this is only characteristic of the pcb antenna and not the low-pass filter.  

I haven't actually pulled the shield off of the module to look at the crystal, but I am quite confident that it is 26Mhz.   I am running the Anaren module with the same firmware as my high-end product with a 26Mhz TCXO and it is completely compatible.

There are so many factors when trying to figure out communications range that it is really hard to just give a number on range.   Some of the bigger factors are quality of antenna (definitely recommend a half-wave dipole vs 1/4 wave),  data rate  (IF filter band width),  multipath (frequency hopping can help here),  and one really BIG one:  height above the surface.   http://en.wikipedia.org/wiki/Fresnel_zone

Frequency hopping can also help in other ways.   If the environment is changing or the system is mobile, fast frequency hopping can help by "finding holes" in the spectrum that are better in a dynamic fashion.   This could be due to multipath (orientation), interference, etc.   So, while it doesn't improve power output, it can improve communications range and certainly makes the system more robust.

For example,  with the long-range radio currently part of the OpenFHSS project range could vary from 1/4 mile in really bad conditions with lots of trees and buildings between and being close to the surface.  The range can improve to 10 miles or more with one end on a moderate hilltop.  With both ends on hilltops, 20 miles or greater is definitely possible.   For reference, the best I have achieved with 1W 902-928Mhz frequency hopper link is 55 miles between two hilltops.

Those are thoughts from my experience anyway.   Good luck!   Hope you figure out how to get the range that you need out of it.

Todd