RF Transmitter only works when I touch antenna

I’m working on a low power ATmega328p based sensor with on-board RFM69 transmitter/receiver.
I made 3 versions of PCB, and first two had no issues with RF, they even worked without antenna in close proximity.
So I made some minor tweaks to last working version (moved antenna hole a little farther from transceiver case it was blocked by it, rearranged some parts, added footprint for large RFM69 module (RFM69W) and rerouted).
Now unit is not transmitting or receiving unless physically touch antenna or part of the PCB (i.e. coin cell battery mount) with my bare fingers. I tried adjusting antenna lengths (it’s working at 915Mhz) shifting it to different positions, etc.
I even soldered second PCB from same batch (thinking that maybe RFM69C module I have was bad), and still same issue…
What am I missing?
I’m attaching schematic from previous version (R0) that works, and new one (R1) that has to be touched in order to work…

What kind of antenna is being used? Is it a wire antenna or an etched pcb antenna. At 915 Mhz there is no such thing as a wire and tracks on a PCB become reactive components which can detune whatever antenna you have ,or even worse place a dead short across output port of the transmitter.

It's a wire 1/4 wavelength...

mauried has the answer… Poor RF design of your board. RF layout is an art-form and involves some serious understanding of HF engineering.

Suggest you read up on RF design… An old ARRL handbook is a great resource, or check it out online.

Ray

You really need to solder the 1/4 wave antenna directly to the Antenna pin of the Transciever chip. Trying to use connecting tracks on the PCB wont work unless you know some of the characteristics of the PCB. Namely the board thickness and the dielectric constant of the fibreglass the board is made of. A 1/4 wave antenna has a feed impedance of somewhere between 20 and 40 ohms, which depends a lot on how good the ground plane is , and if you want to use a strip line connection , which is what a track on the PCB is then the characteristic impedance (Zo) of the strip line is determined by its width and the above 2 factors. The lower the Zo, the wider the transmission line becomes. In short , its a pain unless you are really good at RF design. If you want to have a go here are the formulas. http://www.eeweb.com/toolbox/microstrip-impedance

Hi, in the second diagram you say 2 footprints, so you have placed two footprints of the RF units on the board. Cut the antenna tracks away from the footprints and solder your antenna wire directly to the rf unit output.

See if that improves it, how long is your antenna? Can you post a picture of your project so we can see your layout, please.

Tom.... :)

Mauried thanks a lot of the information. I'll read up on the subject. I think it might be a "dead short" or something else that greatly interfering with signal.

TomGeorge: Hi, in the second diagram you say 2 footprints, so you have placed two footprints of the RF units on the board. Cut the antenna tracks away from the footprints and solder your antenna wire directly to the rf unit output.

See if that improves it, how long is your antenna? Can you post a picture of your project so we can see your layout, please.

Tom.... :)

Thanks Tom. I will cut the tracks as suggested! Track from one of the pads to antenna hole is about 4mm. I tried wire antenna 78mm and 85mm, made no difference... I'm also going to remove footprint for larger RFM module, maybe it will improve things...

Here are links of images, etc.: Eagle Board File PCB all layers PCB Top Layer PCB Bottom Layer PCB 3D rendering Top PCB 3D rendering Bottom Photo of top (assembled) Photo of bottom (assembled) Photo of bottom (bare PCB).

Ok I cut tracks and attached antenna directly to RFM chip. It didn't help :( It sill only works if I touch antenna with my finger (anywhere from middle to far end). Even tho receiver is just inches away from transmitter...

I might have found issue, and it probably had nothing to do with my PCB design :) I have strong suspicion that RFM chips I got were mislabeled, and actually tuned to 433Mhz instead of 915Mhz. Everything works when I change setting to 433Mhz, no need to touch antenna or anything... I also had a know 915Mhz chip from previous generation (RFM12b) and after I soldered it and changed code to accommodate different IC it also works.

I have s similar design and a similar problem here. RFM69 is on the bottom side, antenna on the opposite side. When I bend the antenna out of the PCB to 90 degree and then in parallel to the PCB it isn't working. When I do not bend the antenna and it goes straight to top it is working.

By calculating the antenna length how is the PCB trace handled? I have about 3 mm from the RFMxx module to the antenna hole? Can this be a factor to keep in calculation (antenna length = trace + antenna) or do I have to count the length only by wire?

In principle it must work, eg the Moteinos are designed this way, but I don't know the edges I have to tweak in case it is not working as expected.

Clemens: By calculating the antenna length how is the PCB trace handled? I have about 3 mm from the RFMxx module to the antenna hole? Can this be a factor to keep in calculation (antenna length = trace + antenna) or do I have to count the length only by wire.

I'm pretty sure you have to factor in trace length. Anything between your RFM pin and antenna hole adds to the antenna length.

BTW, I finally got new modules (this time RFM69W) and everything works on the same board where it wasn't working before. Even without any antenna at close range! :)