Antenna Tuning nRF9160

[Risto]

Hi there,

currently I'am designing a circuit for nRF9160 inspired by Thingy:91.

About Thingy's antenna tuning (Thingy:91 User Guide page 16+17) I have some basic noob questions:

- Can I use other RF-Switches than the QM12038, like Skyworks SKY13418-485LF or QPC6082TR7?
  Reason: Availabilty at Distributors.

- Are all RF-Switches drive RF Signals in both directions? are they bidirectional?
  SKY13418-485LF has an ANT pin while QM12038 has an RFC pin which confuses me.

- Is the matching circuit for each Frequency range an impedance matching that reduces return losses?
  Can I do the matching as follows?:
  1.) Configure the two routing RF switches for an frequency range I like to do the matching for.
  2.) Connect a MEGIQ Vector Network Analyser to RFC/ANT input of the first rf-swich (using a SWF connector)
  3.) Use MEGIQ's Measurements to find passive tuning network for the desired frequency range. Impedance matching / reducing return losses?   
  4.) Switch to the next routing path and do the tuning for the next band and so on.
 
- Does the Thingy:91 antenna matching represent an aperture antenna tuning? or is this something total different?

- When does it make sense to use several switchable tuning networks and when not? Does it always improve antenna performance?
  E.g. on nRF9160-DK there is only one tuning network for multibands, but long ground plane.

- Are the limited Thingy91 ground plane length + multiband application the main reason for the switchable tuning architecture?
 

Christof

[Andreas]

Hi,

 

1. Yes, these switches can be used. QPC6082 has slightly higher switching speed, but it will be switched >~50µs before modem activity so it is still plenty low.
2. Switches at least from from reputable manufacturers like Qorvo and Skyworks are unidirectional. There might be some parts that have unidirectionality, or some sort of coupling feature, but that should be specified. The ANT vs RFC is just a difference in nomenclature from different companies unfortunately, there is no standard way of describing pins in RF parts.
3. The matching network maximizes return loss (which is good) by ensuring there is minimal power reflected from the load as it has the same input impedance as the source. There are similar common ways to quantize this, e.g. insertion loss (lower is better), VSWR (lower is better), S11 (larger negative is better).
   1. Yes, each matching network has a limited bandwidth. Depending on how wide this is and which frequency it covers, you might need several to match the antenna for the bands that you want to support. The two switches then switch the RF path through the matching network that is optimal for the band/frequency that is being used at a given time.
   2. Not sure which switch you mean. The VNA should measure into the antenna as seen by the modem:
      
      I would not use a SWF connector though, rather just solder an open-ended coax directly on the PCB.
   3. Correct.
   4. Correct.
4. No, the Thingy91 does is not aperture antenna tuning, this is different. Technically this could be possible, but from what I know this is a quite advanced topic.
5. It only makes sense to use switching tuning networks when the antenna is not wideband enough to cover all the bands you want to support with only one. Antennas are typically made more narrow-banded by reducing the size of the ground plane, so in small boards (relative to the wavelength of the bands used) like Thingy91 it might be needed. You might not want to support all the bands that the Thingy91 supports, you might then be able to do without the switches. Fractus offers excellent support and will assist you with suggested matching network etc. if you pick their antenna for your product.
6. To some extent, yes, as mentioned above. Thingy91 is actually quite small to get good efficiency in the low bands.

 

Best regards,

Andreas

[Risto]

Hi Andreas,

thank you for your explanations.