Author Topic: PCB manufacture - antenna options  (Read 1253 times)

DRCO

  • NewMember
  • *
  • Posts: 19
PCB manufacture - antenna options
« on: October 19, 2017, 11:05:25 PM »
Hi Forum,
I have recently created a board that marries up the Particle Photon / Electron and an ATMEGA 328p (DIL) chip with the RFM69HW. In breadboard testing I used a Moteino (433Mhz) with a simple monofilament antenna as supplied with the Moteino and soldered to the ANT through-hole. I also modified the code for operation at 1200 BPS.
In the breadboard configuration I could get around 150 metres reliable range with the RFM69_ListenMode firmware – and at least 700 metres in normal (Gateway / Node) type configuration. This included some brick walls and metal structure.
Unfortunately, with my custom built board, the range is about 70 metres (tops)  in ListenMode and 300 or so in normal operation.
OK – so something not great about the boards but I was wondering if anyone in the forum could offer some advice on the following:
I use a uFL SMD connector situated immediately above the Radio module ANT pin, it pigtails to a SMA connector and I’ve tried various antennas – and made some myself using a right angle SMA connector and 75ohm Coax Cable, which I’ve trimmed and checked RSSI on each modification. After a lot of effort there I found the best antenna performance was from these cheap Helical Antenna I got off eBay.
https://www.ebay.com.au/itm/5pcs-433MHZ-Helical-Antenna-for-Arduino-Remote-Control-/192080346150?hash=item2cb8e17c26:g:~ukAAOSw44BYfYbG

So a couple of questions...
Can the Helical Antenna be better than a longer wire filament? I trimmed from about 20 centimetres down to around 8 cm in 5 mm steps – I found the best performance was at around 9.8 cm to 1.12 cm – this is considerably different to the 17.3 cm antenna supplied with the Moteino – but when I build an antenna that length performance was terrible. The best i got trimming an antenna was still -5db worse than the helical.
Does the uFL connector ground connection to the PCB ground plane (which is relatively big, around 25cm2) affect range performance? I noticed better performance if the “ground plane” (PCB) is horizontal rather than vertical (tested while maintaining the antenna perpendicular) .
Would I get any benefit from a chip antenna?
Greatly appreciate any feedback or suggestions.
Chris

john4444

  • Jr. Member
  • **
  • Posts: 71
  • Country: us
Re: PCB manufacture - antenna options
« Reply #1 on: October 20, 2017, 09:02:41 AM »
Hi DRCO,
Quote
Can the Helical Antenna be better than a longer wire filament?
I presume you mean a ¼ or ½ wave length wire.
Then no, the wire will provide longer range & better S/N compared to a helical.
Quote
Does the uFL connector ground connection to the PCB ground plane... affect range performance?
Absolutely. Ground connections need to be extremely low resistance.
Quote
Would I get any benefit from a chip antenna?
No. Small antennas require a lot of trade-offs to provide reasonable performance.

I noted that you used 75-ohm coax. That can be a significant issue. Try 50-ohm coax instead.
Every connection will reduce performance and the longer the coax the more the losses add up.
It is not clear in your description but connecting a radio to an antenna through a breadboard will result in very poor performance.
Check out John K2OX's forum comments for good info on antenna performance.
https://lowpowerlab.com/forum/rf-range-antennas-rfm69-library/antenna-tutorial-or-antennas-in-a-mote/

Good Luck,
John AE5HQ
John AE5HQ

DRCO

  • NewMember
  • *
  • Posts: 19
Re: PCB manufacture - antenna options
« Reply #2 on: October 23, 2017, 05:31:21 PM »
Hi John,
Thanks for your response and sorry about the late reply. Just to clarify a few points...
Regarding the helical vs single wire (1/4 wave monopole) antenna, all that I’ve read suggests you are right but I simply can’t seem to trim the Gateway and Node antennas to a length that outperforms the helical. I’ve had a few goes now – and although the technique is not “scientific” I am measuring RSSI with every trim (of approximately 3mm) and the best I get still underperforms compared to the helical.

To be clear, I’m using a right angle SMA connector with a crimp cable side. I remove the crimp sheath (leaving a hole through to the centre pin) and heat / gun solder paste to connect to the centre pin. I am VERY careful to insulate against the ground. The antenna wire is the copper core of a 75ohm coax cable. You mentioned that I should try 50 ohm, and I get it that the impedance of the radio is 50 ohm but we are just talking a piece of wire here – does the diameter of the wire affect the impedance? Everything I’ve read suggests not, the dielectric compound surrounding the inner core cable – and the shield composition has the most effect, although diameter of the inner core does affect impedance, it should be offset by trimming the wire. All connectors (the SMA, uFL nas pigtail are 50 ohm rated – as is the input to the RFM69HW.

Regarding the grounding – I’m pretty confident this is extremely low resistance. The uFL pads are within a mm of the radio pads and all are contact make – with over solder. Probably as good as Felix’s, Moteino – although his construction techniques would no doubt be better than mine.

Regarding the breadboard assembly, I rigged a RFM69HW to some veroboard and soldered an antenna exactly the same as the ones provided by Felix with the Moteino directly into the Antenna pad trough-hole on the RFM69HW. The SPI wires to the ATMEGA chip were jumpers – the regulator etc... were rigged pretty much in the same way.

One thing I note is that I’m running my config at 1200Bps and the specs for the radio suggest I can get comms at -120 db. With the breadboard assembly, I note I’m getting RSI around -110 at 150 metres before I lose communications (I’ve detailed in this post...
https://lowpowerlab.com/forum/low-power-techniques/low-power-listenmode-with-lower-baud-rates/).
With my current fabricated PCB I never get a connection with anything above 100 dB in ListenMode – and the range is around 80 metres at best.