Using attenuator on nRF52 for finer control over the TX power

[Siva]

Hi All,

We have build a BLE beacon device based on the nrf52 chip. The device works fine but we would like to get a finer control over the TX power since it is used for proximity approximation. To achieve this we are planning to add a variable attenuator to our design (We haven't finalized on the attenuator chip to be used).

The idea is to control the TX power from the nrf52 based on the needs via the interface that the attenuator exposes (SPI/I2C). I don't see any issues with the software part of it but I am really concerned about getting the exact control on the Tx power even after adding the attenuator and also any side effects of doing this.

I would like to know from you guys about any similar experiences that you had or any possible challenges/issues that you feel I would face when doing this.

Thanks in advance.

[Andreas]

Hi,

 

We do not have any experience with doing this as far as I know, maybe someone else has some experiences to share?

 

In any case, what you describe should be doable in theory and from what I am able to see on distributor sites there are at least a fair number of parts available that might be suitable. Still, an attenuator with finite steps (e.g. 0.25 dB, 0.5 dB, 1dB etc) will still leave some variation in the output power. Similarly, depending on how you plan to do the calibration, there might be some variation coming from this as well, e.g. probe wear.

 

I am a bit worried that what you are trying to do will not make much difference in practice, even if you are able to get the total variation down to about 0.25 dB or less, there will qutie likely still be significant fluctuations to signal strength due to fading. These attenuators are expensive, so before making up my mind I would compare boards with improved TX power control with boards that do not have this, to see if there is any difference and possibly being able to shave a quite significant item of your BOM. Remember that even if the TX power control setting in the nRF52 is 4 dB, the actual distribution is much less. Although not a side-effect, you will have to figure out how to calibrate this in your production line, which might be a little costly for all I know. Apart from that I can not think of any drawbacks, I assume a little loss in RX will not be a large factor for a beacon.

 

Best regards,

Andreas

[Siva]

Thanks Andreas. To give you a glimpse of the use case, the customer wants the beacon to sent an advertising packet when the PIR's connected to it (GPIO of nrf52) is triggered by some movements. The requirement enforces that this packet has to be only received by someone who is to the close proximity of the beacon (let say within a meter). What we have observed is that the -40 dBm is to feeble to achieve this and -20 dBm is way to stronger causing the advertisement packets to be transmitted to someone few meters away which is against the requirement.

I guess the -30dBm option would have fixed this problem to an extend but this seems to be deprecated in the nrf52. That is the reason why we thought to adding an attenuator to control the Tx power.

As you have mentioned the calibration of the attenuator would be the tricky part but I believe we dont have an other option to reply upon.

Thanks a lot for your inputs. Any further suggestions or comments are welcomed.

[Andreas]

Hi,

 

This use case makes more sense.

 

You are correct, the -30 dBm output power simply yields -40 dBm. I still think that a variable attenuator is a little overkill, and an extra cost that should not be necessary. Of course this is up to you, but I would look into deciding for a ~10 dB fixed attenuation and use the -20 dBm setting. The exact value can then be tuned and tested for a number of samples. Alternatively you can probably achieve this by mistuning the nRF matching network, or antenna. The last alternative will let you omit the attenuator entirely but it is a more complicated effort, requiring instruments and knowledge for RF testing and tuning.

 

Best regards,
Andreas

[Siva]

Hi Andreas,

Thanks for your response. I am sorry I didn't quite get how to do this. Could you please elaborate?

Andreas said:

The last alternative will let you omit the attenuator entirely but it is a more complicated effort, requiring instruments and knowledge for RF testing and tuning.

Andreas said:

Alternatively you can probably achieve this by mistuning the nRF matching network, or antenna

[Andreas]

Hi,

 

Consider the reference design, there are two components (L1 and C3) that match your load (antenna typically) to the radio, while suppressing harmonics. Usually these have to be 'tuned' to optimize the performance for a specific layout, small layout changes can have a relatively large impact on performance, but you can try to tune L1 and C3 to reduce the output power. You need a spectrum analyzer to know how much a given combination of L1 + C3 will do to the output power, but it is of course possible to just modify a board and measure the range directly, it will just be a little less accurate and probably take some more time.

 

Depending on your antenna, you might also be able to mismatch it (similar to L1+C3) to reduce the energy transferred, in essence attenuating your signal. Doing this it is not as easy to determine how much the signal is attenuated.

 

Best regards,

Andreas

[Siva]

Thanks Andreas.

These are good options which are worth trying. Thanks for pointing out.

I'll get back to you on this later.

Cheers