Inquiry regarding the phenomenon where BLE TX power varies depending on the lithium-ion battery voltage

[sangyun.kim]

Device : wearable IOT device

Battery capacity : Max 4.2V (Typ. 3.7V)

BLE Tx Power : Tx power is highest when the battery capacity is 4.2V (measured in air, not conduction)

I wonder if this phenomenon is natural, and if so, what the reason is. 

[sangyun.kim]

This is the power connection of the nRF5340. Since VSYS, the power input to VDD, uses a 3.3V ldo output, there appears to be no power fluctuation, and when capturing the actual waveform, only a fluctuation of about 20mV is observed during BT Tx. 

The Tx power variation is as follows:
It was observed that the power variation in the advertising channel decreased by up to 10dB at 100% and 20% battery levels. The measurement was performed by placing an IoT device inside a shield box and measuring the power. 
BT power measurement is not conduction mode, but the radiated mode you mentioned. 

Additonal Test.
Actual battery in use = norminal 19mAh
The 20mV fluctuation mentioned above was measured by using a power supply to check the VSYS voltage drop. This time, we confirmed that when using the actual 19mAh battery used by the IoT device, the VSYS voltage drop dropped from 3.3V to 3.142V. The battery voltage at this time is 3.5V

[Jam Alan]

A Li-ion cell typically varies from about 4.2 V (fully charged) down to 3.0–3.3 V (near depletion). If the BLE radio's PA is powered directly from the battery (or from a regulator that does not tightly regulate the PA supply), the maximum RF output power often decreases as the supply voltage falls. speed stars

[Ressa]

sangyun.kim said:

The 20mV fluctuation mentioned above was measured by using a power supply to check the VSYS voltage drop. This time, we confirmed that when using the actual 19mAh battery used by the IoT device, the VSYS voltage drop dropped from 3.3V to 3.142V. The battery voltage at this time is 3.5V

What type of LDO is being used? Is it a Nordic PMIC device? Some LDOs require a higher battery voltage ( for example higher than 300 mV) than LDO output to operate safely. Also the 20 mV voltage drop will not cause 10 dB power loss. Can you try to use external power supply instead of battery and gradually decrease the power supply voltage (similar to battery voltage eat 100 % and at 20 % charge) to see if the power variation still exist? This is to role out the battery internal resistance effect on the power observation.  
Also what TX power is being used for advertising ? 

[sangyun.kim]

TPS7A0533PYKAR is used for this system. (https://www.ti.com/product/ko-kr/TPS7A05/part-details/TPS7A0533PYKAR)
Input is Vbatt.   output is 3.3V ldo output. 
Im using the maximum power of the BLE nRF5340, which is +3dBm. As mentioned above, I observed a phenomenon where the input voltage to the nRF5340's VDD drops from 3.3V to 3.142V, and I am wondering if this could affect the BLE Tx power. 

[Ressa]

Hello,

The PA in radio is using regulated internal voltage so as long as VDD is within recommended range, you should not notice power variation especially in range of 10 dBs. To find the root cause I would strongly recommend to have some conducted measurements (right before antenna) to first verify the power fluctuation and also see how much TX power is varying. 
Also as I suggested, try to use external supply instead of battery and manually change the external supply voltage to see if you get same observation or not.