Directly connect SMA to RFM95W

Hi everyone,

I’m trying to make my first node by using an Arduino Pro Mini and the RFM95W. I’d like to avoid soldering the RFM95W to a board, so I was thinking of soldering the wires directly to the RFM95W, but I have a few questions about it:

  1. I have seen most PCBs designed for the RFM95W include capacitors. Is there any downside to not including them?
  2. What would be the best way to solder an SMA antenna directly to the RFM95W? I basically see two options: using an SMA connector and soldering one gnd terminal and the signal terminal to the board (would that fit dimension-wise? is it a problem to only solder one of the two side pins to gnd?), or using an SMA to u.Fl IPX cable, such as , and soldering it (would there be any relevant signal loss/impedence mismatch problems related to this?)
    I also looked for PCBs with an smd SMA connector on one side, and throughhole signal and gnd pins on the other, but I could find anything already designed. Would this make sense too?

Well, that’s about it for now. I’m really excited for this technology, and hope I can make some interesting stuff.

Thanks for any help you can give! It’s really appreciated

1 Like
  • You can use a SMA/u.Fl, but why not just add a simple wire (=antenna) to the RFM95W, see the forum for the required length.
  • The forum also show examples of soldering the Arduino (3v3 version) and RFM on top of each other with wires.
  • Capacitors are useful for filtering noise and/or to avoid a dip in the power when an IC suddenly requires more (example: send data). So you can try without, and add a cap (between 3v3 and gnd close to the RFM95) when you experience issues, like a spontaneous reset.

Thanks for your reply!

I was thinking a proper SMA antenna would provide better performance, and I plan to use it for relatively long distances. Is this not true? Would a helical antenna be equal/better than these options?

Yes, thanks for the reference. That’s what I was planning on doing, I believe I saw that post and inspired me to do it.

About the capacitors, that definetly makes sense. Now that you mention it, I think I remember viewing a video by Andreas Spiess where he mentions the power issue.

Anyway, you basically solved my doubts, beside the little thing on the SMA antenna vs wire vs helical performance. So thanks for the help!

A ready made helical antenna wire 868Mhz is often used (and already the correct length).

1 Like

The almost best results I had so far were with relatively thick solid copper wire. 82mm long and 1,5mm in diameter, that’s what’s usually used for main current (230v). That was for diy nodes. The semtech node with the pcb antenna performs far better, but it works directional. So you’ve got a huge range looking ahead while you’ve got a much smaller range in the other directions.

This solid wire works far better than a helical antenna.

1 Like

No a directly connected helical antenna will not be better.
But mounting (somehow) an SMA connector and then using a poor antenna won’t help much either.
A chain is only as strong/good as it’s weakest link.

It all depends on what you are trying to achieve. Should your solution be as small as possible OR do you want to have the best reach/performance?

If you want to solder it directly to the RFM95W module there is actually only one single (good) solution: solder a cable with SMA connector on one end (or other antenna connector to your liking) directly (and properly) to the module.

You will not be able to solder an I-Pex (U.Fl) connector onto the module reliably (and the same actually holds for an SMA connector).

Why do you not want to use an adapter?

HopeRF has an RFM95W adapter/breakout board without any additional components so additional components are not explicitly required. I use them on breadboards but they can be used on prototyping PCB’s as well.

Example link for the RFM95W adapter board:

1 Like

Why 82mm when a quarter 868Mhz wave length is 86.3mm? (Antenna length for 868 and 433 MHz)

… solder a cable with SMA connector on one end (or other antenna connector to your liking) directly (and properly) to the module.

I’m looking for good performance, size is not much of an issue (within reason). Would soldering a cable like this work for what you are saying?

If not, what do you mean by “properly”? Any tips on how to do it?

Why do you not want to use an adapter?

Mainly for the price. An RFM95W module is about 5-6€, the board you linked is already more than that, and doesn’t include the module.
Basically, I wanted to know if it is possible to do it without the adapter without making the performance worse. Ultimately I would consider it if there is no viable alternative, but I like to consider all options, and learn something in the process.
Thanks for the insight and the help! It’s really appreciated.

The almost best results I had so far were with relatively thick solid copper wire. 82mm long and 1,5mm in diameter

Interesting, I will definetly be trying that.

Because of the velocity factor.

Velocity Factor

Varies with the diameter of the wire too…

1 Like

I have yet to read/study about the velocity factor. :slightly_smiling_face:

But also see this (although a different but also 1/4 wave antenna):

“Hooking it up to the spectrum analyser at work revealed that my build was actually better suited for 900MHz so I increased the length and tuned it until it was very close to 868MHz and surprise surprise the ideal length is 86mm.”

Do bear in mind that whilst 82mm might be a good length when there is an effective groundplane,
it may not be the optimum length with no effective ground plane.

Adding a counterpoise, the same length of the ‘antenna’ will turn it into a dipole in effect, which should imporove things.

Yes, my suggestion was a cable like that. I have used it for connecting an SMA antenna to a specific 433MHz tranceiver for an RFLink Gateway (I used a prototyping PCB shield for an Arduino Mega to build the gateway).
But I normally prefer ready made adapter boards with an SMA antenna connector.

With “properly” I meant “with craftsmanship”. :slightly_smiling_face:
This is RF so it should be done correctly, not some sloppy soldering and keep the shielding on the cable in tact as close to the RFM95W module as possible.

Yeah, €7 for the adapter is relatively expensive (therefore ‘example link’). I ordered 20 pieces two years ago somewhere in Europe for about €1 per piece (without: SMA connector, headers and without shipping costs).

I don’t know, but I did notice on multiple occasions even a mm more gives poorer results. Although I didn’t dig into it I do recall a radio amateur once said to me that the rule of thumb about 1/4 wavelength was in fact ‘almost 1/4 wavelength’.

edit: sorry, didn’t notice I had to scroll down when I replied on the notification. Those particular nodes indeed did have quite a good ground plane.


For optimum results, there are benefits in tuning antennas to suit.

There is a natural variation between LoRa modules due to manufacturing tollerances, what is the best length for one module may not be the best for another.

Do you know the ‘Doug LaRue’ boards? About 1 euro each.

Yes, the ground plane is of substantial influence.

If I understood correctly, the length of a quarter-wave mono-pole is based on having a proper groundplane because the ground plane in fact serves as the counterpoise of a half wave di-pole.
From this perspective, when having a proper ground plane the length of the mono-pole should be exactly one quarter-wave length (velocity factor not taken into account).

Here in the Antenna length for 868 and 433 MHz topic a general rule of thumb is mentioned to use 5% for the velocity factor.

86.3mm x 0.95 = 81,99mm which exactly matches your 82mm.

Hello bluejedi,
I have bought 2 Hoperf Rfm98w adapter boards and i have soldered it as you see on picture. But i couldn’t find the output pin names (right side pins) on adapter.
Do you know these pins ?
Thank you!

Hi @doganyldz,

The pins on the right side are +3.3V and GND pins for power.

As positioned on your picture the right row of pins (starting from the top) are connected as follows:

  • GND
  • GND
  • GND
  • 3.3V
  • GND
  • GND
  • GND
  • 3.3V
  • 3.3V
  • 3.3V

To power the module you only need to connect one of the GND pins and one of the 3.3V pins.


The adapter does not have a pin for DIO5. This normally is not an issue because for LoRaWAN only DIO0 and DIO1 are needed (when using a LMIC LoRaWAN library).


When using this adapter on a regular breadboard there will be only one ‘row’ of pins available for wiring. Use that to wire the left side (RST … DIO1).

The 3.3V and GND pins that are positioned on the right can each be wired using breadboard jumper wires that are (partially) positioned beneath the adapter. Place those jumper wires first and then place the adapter on the breadboard.
Take great care to wire the correct power pins. It’s easy to make a mistake because the wires are positioned below the adapter so you can’t see to which pins they are connected once the adapter is placed over them.

  • Wire one of the GND pins from beneath the adapter to a row outside the adapter with a breadboard jumper wire. Then connect that row to GND (power rail with another jumper wire).

  • Wire one of the 3.3V pins from beneath the adapter to a row outside the adapter with a breadboard jumper wire. Then connect that row to 3.3V (power rail with another jumper wire).

  • After the breadboard jumper wires are positioned, place the adapter on the breadboard in the correct position.

breadboard jumper wires
Breadboard jumper wires.

1 Like

Thank you so much ! i will open a new thread about this project.
Have a good day !

1 Like