I’m just posting this in case it might save someone else a bit of a headache.
These chips in particular, and apparently others that use similar output stages, are exceptionally sensitive to impedance matching on the output. With a moderate mismatch (1.5-1) you may see a third or fifth harmonic that exceeds the carrier level (!), as well as a significantly depressed carrier level.
As it happens, there exists a transformer specifically designed to act as a filter/balun for the Atmel parts. I can’t swear to it for other chips. Getting this little piece of information, and redesigning our antenna for a closer match resulted in a 10dB boost in radiated power along with the harmonics being below -45dBc instead of +15dBc.
Two resources that you should know about if you are in this or a similar hole, especially if you are within shouting distance of Sparkfun:
1: http://www.emitestlab.com/ Dennis King is SUPERB, and his pricing per half day is surprisingly reasonable. He uses a GTEM so scan time is fast and the results correlate well with our OATS.
2: http://www.ccn-i.com/Welcome.html If Don DeGroot can’t figure it out, you are in one deep and UGLY hole.
Both are in Longmont, and I’ve been working with them for over a year now.
No direct relationship, just a very satisfied customer.
Thanks for the info. I’m using the AT86RF212 (900MHz) in a design, and used the balun & filter in Atmel’s reference design. Will hopefully get some code going this weekend to try it out…
The particular balun I’m referring to here is third(!) on atmels list, and at no point do they mention that this particular part, unlike the first two on their list, is specifically designed to address this rather awful problem.
Johanson 2450FB15L0001 is the good part.
Not the first time that Atmel has casually mentioned something in a footnote on page 300 that IMHO should have been a “black bar” warning.
Thanks for that. Which document are you referring to? Just curious, since I couldn’t see that info anywhere.
The doc on their site that mentions the balun is here:
http://www.atmel.com/dyn/resources/prod … oc8111.pdf
We have some additional docs from Atmel that aren’t on the web site, that go into more detail. In one there is a list of baluns, and the Johanson filter balun is only peripherally mentioned, and in none of their docs is any mention made of how the output stage reacts to impedance mismatch. Fortunately this showed up on our EMI pre-scan, and we didn’t waste a day at the OATS ($$$$)
The antenna (sleeve dipole) was designed with the correct lengths for open air, but the dielectric constant of the heatshrink shifted the resonant point by some 200 MHz, which was somewhat farther than I would have expected. Also the length of the feedline was an issue since the antenna isn’t 50 ohms at resonance, and the reflected energy was arriving back at the TX out of phase. Correcting the dimensions of the dipole for the heatshrink, and optimizing the feedline length made a huge improvement. Without Don’s help we wouldn’t have been able to see what we were doing, and we would have wasted a lot of time and money fumbling around. An SA won’t help you here, you need to see the antenna impedance across a pretty wide swath of frequencies.
Frequency stability is another point that is very much soft-pedaled in the docs. You need to cal the chip as close as possible to 0 ppM error, since the crystals drift a few ppm/year and within a few years you can end up with devices that can’t hear each other. We have a Trimble Thunderbolt GPS disciplined oscillator that is giving us a few tens of ppT error as a reference. The on-board cal register defaults to 00h, which sets the oscillator to the high side of center, so be aware of this when selecting oscillator caps. 1pF is very significant, you want the center of your distribution to be on the high side with the trim at 0, (default) so that you can trim it in. You can only trim the frequency downward.
Dennis is an amazingly good resource for prescans, both technically, and pricewise. As the old saying goes, “If you can’t measure it, you can’t improve it.”
Thanks very much for the detailed response.
I’m using a pre-made 1/4 wave dipole antenna, so I’m hoping the impedance mismatch won’t be such an issue. However, the crystal drift is particularly concerning, and not something I had considered previously.
Lots of things to consider in my testing, which I hope to do soon…
Quarter-wave dipole? I think you meant half-wave, at least I hope so.
One other thing to know, tune for resonance, then match impedance. There is a widely held misperception out there that somehow all antennas should be tweaked to be 50 ohms (or 75). A given antenna geometry has a particular impedance at resonance, which may be affected by construction, but in a “pure” state, those numbers are only sometimes 50 ohms. A 1/4 wave monopole is somewhere around 19 ohms, but adjusting the shape of the “ground plane” (actually a 1/4 wave choke) twoard the shape of a 1/2 wave sleeve dipole will get you to a 50 ohm match. The sleeve dipole is around 75 ohms. A folded dipole ends up around 300 ohms.
Don’t mangle your antenna to get 50 ohms, build it to resonate at the appropriate frequency, then match impedances if you care. You can calculate loss due to impedance mismatch, it is smaller than you might suspect. A cable length which is an odd number of electrical half-waves can make that even less important.
dbvanhorn:
Quarter-wave dipole? I think you meant half-wave, at least I hope so.
My mistake - I should have said 1/4 wave monopole / whip. Initial testing will be with this antenna:
http://www.antennafactor.com/resources/ … -cw-qw.pdf
Ground plane will be close to that specified in the antenna datasheet, so impedance should be close to 50 ohms.