Could you give some details of AGCBACKOFFGAIN?

[Zhang Xin]

Hi Nordic,

It mentions that "Gain will be lowered by the specified number of gain steps at the start of CTE. Note: First LNAGAIN gain drops, then MIXGAIN, then AAFGAIN" in nRF52833's PS v1.3, however, there is no explaination about the step, LNAGAIN, MIXGAIN or AAFGAIN. 

Besides, what are the ordinary value/range of the I/Q samples (based on some AGCBACKOFFGAIN value maybe, e.g., when AGCBACKOFFGAIN = 0)? I usually get 100~200. Is it too small for the 12bit ADC? Is it possible that the samples value would impact other samples nearby? For example, if a sample's value is very large, the next sample would be larger than it should be?

Thanks.

[Andreas]

Hi,

 

LNAGAIN, MIXGAIN and AAFGAIN are internal gains that are not documented externally. Details are not needed for this purpose though, as the total gain is what is important. Higher AGCBACKOFFGAIN = lower total gain, you should experiment with what works best for your application.

 

AGCBACKOFFGAIN is a 4 bit value, default is 0. If and how much you need to back off depends on the amount of phase noise due to multipath in your setup. If the receiver saturates it will give I and Q samples that are -32768, in this case it should help to increase AGCBACKOFFGAIN.

 

Best regards,
Andreas

[Zhang Xin]

Hi,

Thank you for the reply. Due to the lack of detailed explanation, I wonder if you could answer the following questions:

Is the AGCBACKOFFGAIN step linear or logarithmic?

I thought the IQ samples’ amplitude of the reference antenna should be around the same values for different situation as the AGC should compensate the change to keep the IQ samples’ amplitude constant, however, I found these samples' amplitude changed notably sometimes when RSSI changed (due to polarization or distance changing). Is this phenomenon normal?

I thought the IQ sample range should be -2048~+2047 (12bit ADC), and I don't know why it is "-32768" as you mentioned.

Thanks again.

BR

Zhang Xin

[Dmitry]

Hi,

The same behavior on my nrf52811 - the range is between -200..+200 most of the time.

Zhang Xin said:

I thought the IQ samples’ amplitude of the reference antenna should be around the same values for different situation as the AGC should compensate the change to keep the IQ samples’ amplitude constant, however, I found these samples' amplitude changed notably sometimes when RSSI changed

Are you using the same antenna for packet data and reference? I see an amplitude almost constant.

Zhang Xin said:

I thought the IQ sample range should be -2048~+2047 (12bit ADC), and I don't know why it is "-32768" as you mentioned.

-32768 means saturation (here it's explained by Kenneth)

[Zhang Xin]

Hi Dmitry,

Thanks a lot.

I use the same antenna for data and reference. When I rotate my tag, the RSSI of the locator would vary notably (not sure if it's due to the polarization or the directivity of the tag/locator's antenna). When the RSSI is low, the IQ amplitude of the reference antenna may be different. For example, the switching pattern is 0,1,0,2,0,3,0,4..., 0 is the reference channel, the amplitudes of the 0 channels are not around the same value, and they are stable (similar values pattern such as 100 120 125 95 (should be around 100 for example), depends on the rotation of the tag). Seems that IQ samples would be influenced by their neighbor sometimes.

The “-32768” is not found in the PS of nRF52833 or nRF52811, I didn’t pay attention if I got this value before. Is it true for 52833 and 52811?

BR

Zhang Xin

[Andreas]

Hi Zhang,

 

The AGC is frozen during the CTE part of the packet. As the signal propagation might be different between different antennas, the gain might have to be backed off a bit to avoid saturating the downstream receiver.

-32768 is valid for nRF52833 and nRF52811, it's just a documentation bug that it is missing and it will be added in the next respective releases.

 

Best regards,
Andreas

[Zhang Xin]

Hi Andreas,

Thank you for the clarification. I will try to see if the problem I encountered is related to the saturation.

BR

Zhang Xin