Unexpectedly high RSSI samples

Hi,

I wonder if you can help with this query.. I'm using the nRF52811 in a design that uses a proprietary protocol. The system is currently using 8 channels, hopping and keying up every 4.17 ms and sending a short packet in 125k coded mode. I'm currently using another board to perform RSSI measurements on all channels from 2401 to 2483 MHz with 1 MHz spacing. The boards are both synchronised to GPS 1 PPS and the RSSI Task is scheduled to occur smack in the middle of the burst sent by the first board. I just grab one sample per channel. The transmitting board is set to -16 dBm, and I am connecting it to the second board via a 20 dB attenuator, so expect to see approx -36 dBm reported, which I do. But I am seeing surprisingly high numbers 1 MHz and 2 MHz above and below the active channel, especially given I am using a 125k mode. Below are some max-hold captures generated by a PC application I wrote to plot the RSSI samples. The scale is -110 dBm to -36 dBm. The first board is using 2407, 2415, 2435, 2439, 2443, 2447, 2467 and 2475 MHz. Each bar is a sample from a channel, so they are 1 MHz apart. What's also alarming is that if I replace the first board with a clean, unmodulated IFR/Marconi bench source, the two channels above are always reported only ~35 dB down (middle plot). Something is not quite right.

Any useful feedback welcome !

Cheers

Pete

Parents

0 ketiljo over 4 years ago

Remember that coded phy is running with 1 MBps so the carrier is ~1 MHz wide. The RSSI detector will measure the energy in the channel but some energy from the adjacent channel will also we part of this. The unmodulated carrier doesn't have a bandwidth so very little adjacent channel power.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 user79956 over 4 years ago in reply to ketiljo

Thanks for the reply,

Yep, I realised after I posted that the coded phy would still be 1 Mbps.

But the fact that I am seeing energy only ~30 dB down on the channel 2 MHz higher when I use an unmodulated calibrated source is still puzzling me. And it's not symmetrical. I've checked the signal on a spectrum analyser and it is very clean and within 120 Hz of 2441 MHz. Another plot at -40 dBm:
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel

Reply

0 user79956 over 4 years ago in reply to ketiljo

Thanks for the reply,

Yep, I realised after I posted that the coded phy would still be 1 Mbps.

But the fact that I am seeing energy only ~30 dB down on the channel 2 MHz higher when I use an unmodulated calibrated source is still puzzling me. And it's not symmetrical. I've checked the signal on a spectrum analyser and it is very clean and within 120 Hz of 2441 MHz. Another plot at -40 dBm:
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel

Children

0 ketiljo over 4 years ago in reply to user79956

What you see at +2 MHz is the image frequency. This is at fRX - 2 MHz in 1 Mbps mode and fRF - 4 MHz in 2 Mbps mode. The reason you see it at +2 MHz is because fRX is moving and the carrier is static.

Co-channel rejection isn't as good at the image frequency as it's on the opposite side side of the carrier.

https://infocenter.nordicsemi.com/topic/ps_nrf52811/radio.html#unique_447008773
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 user79956 over 4 years ago in reply to ketiljo

Thanks for the reply.

So the IF is 1 MHz then, with fLO 1 MHz below the wanted? That would also explain the high level I see at f+1 MHz, which must be the response to LO leakage. Is the IF given in the datasheet? I couldn't see it. The figure for image rejection is a bit useless without knowing the IF, I think. Your comment about the co-channel rejection is a little confusing - co-channel is just that, an interferer on the same frequency as the wanted. It shouldn't be affected by the image response.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel