RFM12b Antennas + Metal Enclosures

Added by fire5ign over 5 years ago

I’ve been using a handful of devices using the Jeenode libraries, all feeding a JeeLink attached to an RPi with great success for the past 3 or 4 months. I get fairly good communication through the house, such that all of my devices can be positioned in such a way as to be readable by the JeeLink. The devices are all either un-enclosed or boxed in plastic enclosures. I’m using the 433 MHz band and the antennas are trimmed to 17cm (quarter wave).

My latest project was to build a receive-only unit with a 4-line LCD display. I tested it for a week before setting it inside an enclosure, which happens to be an aluminum box, and it worked fine until I started attaching it to the metal. I had naively figured that I could just drill a small hole and feed the antenna wire through it, but reception is abysmal. I tried keeping it attached to the bottom of the box, but taking off the cover (see photo), and reception still did not improve, I tried attaching the 0v side of the circuit to the box, hoping that would improve things, but it did not.

I know about trimming the length of the antenna to match the frequency, but I have a feeling that I need to
1/ match the impedance of the radio to the antenna and/or
2/ use an SMA connector

but both of those issues seem to raise more questions than answers, like: how do you match the impedance? and: what exactly is an SMA connector and how do I use it?

Or, should I just drop the metal box and find good ’old plastic boxes?

Unit with the cover off

DSC_2659.jpg (64.6 KB) DSC_2659.jpg Unit with the cover off

Replies (7)

RE: RFM12b Antennas + Metal Enclosures - Added by martynj over 5 years ago

fire5ign, it would be a shame to drop the diecast box - looks neat.

The problem is that the antenna is active along its whole length, from the connection to the matching network on the RFM12B module, through the connecting traces and out along the approx. quarter wavelength of wire. It is working against its reflection in the groundplane of the module/JeeNode (as imperfect as that is).

By feeding the wire through another ground plane, the quarter wavelength/reflection is drastically altered, giving the output signal (and to an extent the RFM12B transmitter section) a hard time.

The solution is to extend the matched termination point away from the module as far as where the box is penetrated. Use a length of co-axial cable (ideally 50Ω, but 72/75Ω will work). Strip back the outer PVC and create a ‘pigtail’ from the woven sheath. Connect the inner to the ANT pad and pigtail to the adjacent GND pad (it’s easier to insert pins here and trim after soldering) with the minimum of white inner insulator showing.
Pass the entire coax through a hole in the box and strip back a quarter wavelength + 10% to expose the bare inner. This is now the radiating portion of the antenna. Leave enough of the outer sheath to make a similar pigtail. You can experiment with connecting that pigtail to the box adjacent to the penetrating hole, or do the first test without (the reflection happens at that surface anyway.

Why the extra 10%? If you have access to a field strength meter (or configure a spare JeeNode to use NRfMON ), then the exposed antenna length can be trimmed for maximum output.
For a neater effect, buy a quarter-wave molded antenna with connector and mount a mating connector in place of the hole through the box.

RE: RFM12b Antennas + Metal Enclosures - Added by fire5ign over 5 years ago

Thanks very much for your helpful reply. This gives me lots to go on. One further question: is attaching the 0 volts side of the circuit to the metal enclosure a good idea?

RE: RFM12b Antennas + Metal Enclosures - Added by martynj over 5 years ago

fire5ign, a reasonable question, but a fuzzy answer ;)
It is actually a similar issue as connecting the second pigtail to the box or not (that’s really the right way to do it, rather than a random length of wire from the PCB to the box).
The box has several useful electrical functions:
* Reduces the penetration inside of general EMI outside the box
* Reduces the stray radiation outside of EMI generated by rapid current transitions inside the box
* With the right geometry, acts as a reflector under the orthogonal ’whip’ antenna
It can do all of these without reference to the DC 0v of the PCB. Can’t in general say if bonding is going to be better or not
usually it does no harm (and if you go the connector/commercial whip aerial route, is done automatically at that connector).
RF at these frequencies can play tricks though - watch out for some aerial detuning when the box is touched for example (even with non-conductive paint).

RE: RFM12b Antennas + Metal Enclosures - Added by fire5ign over 5 years ago

So, here’s what I’ve got now: I found some very light gauge coaxial cable, taken from an old iMac’s wifi antenna lead. I don’t know what impedance it is, so already I’ve got a variable I can’t account for (is there an easy way to determine its impedance?). But I thought I’d give it a try. I’ve done what I think you described earlier, and drawn a sketch (attached). I’m not getting much better performance with this, but I may have misunderstood your instructions. I’ve ordered some 50 ohm cable on ebay, but since the problem isn’t solved yet, I may try loading up NRfMON to see if I can tinker with it for better reception.

Is this correct?

RE: RFM12b Antennas + Metal Enclosures - Added by fire5ign over 5 years ago

Just fired up nRFmon, but could not get it to work. When trying to connect, it complained this 4 times: ‘error reading file18ea788 I/O error’, then gave up the ghost with a black box and “No hw ld”.

RE: RFM12b Antennas + Metal Enclosures - Added by martynj over 5 years ago

fire5ign, I expect the salvaged co-ax was a soldering challenge - that size usually relies on mechanically crimped connectors!
It won’t look so neat, but the try using the untrimmed coax to get through the metal, then do the conductor fan out on the “outside”. This may not seem like much of a change, but with the considerable thickness of the metal and presumably a circular hole, this has created a short section of another coaxial ‘cable’ with mismatching impedance.
Note that a coax connector, doing the same pass-through task, has critical dimensions tightly controlled for adequate performance at (just) sub-GHz. Just look at the prices of ‘N’ types !

RE: RFM12b Antennas + Metal Enclosures - Added by fire5ign over 5 years ago

I think reception is improved somewhat, thanks.

BTW, the problem with nRFmon is solved. Apparently one of my ATMEGA328 chips is bootloaded with a funny fuse setting, which was doubling the serial rate. Instead of downloading the sketch using the Lilypad USB (it’s not a LilyPad but that was the one that worked), I used the Duemilanove 328 setting. No more problem. I’m looking further into the issue with nRFMon.