AWR2944: AWR2944: Dependence on RF Profiles and The Calibration Error During Calibration

Hi,

1. Are all the questions related to phase shifter calibration only or all the calibrations?

2. Do the questions refer to factory calibration procedure? Procedure mentioned in https://www.ti.com/lit/an/spracv2/spracv2.pdf .

I hope the customer is referring to the application note: https://www.ti.com/lit/an/spracf4c/spracf4c.pdf.

Regards,

Swati

Hi, 

1. The above problems have these problems for phase shifter calibration and antenna imbalance calibration,
2. All questions are about the calibrations mentioned in the spracv2 documentation

Hi Libo,

Please find the answers below:

1. Is constant frequency calibration required during calibration? – You can use a chirp signal for calibration. Constant frequency (0 slope) is not necessary.
2. How does the choice of center frequency during calibration affect PS INL? – As we move farther from the calibration frequency, the accuracy of the phase shifter calibration degrades. Hence, the user should always perform phase shifter calibration with the user in-field profile frequency.
3. As the calibration itself also has the problem of jitter in the measurement data, does the calibration operation require N times of calibration to obtain the expectation? – Doing the calibration N times would ideally lead to less measurement inaccuracy and noise. It would be recommended to check whether the IF tone for phase shifter calibration measurement has a good SNR (>40dB), and if that is the case, then even 1 time calibration is good enough.
4. If N operations are required, what is the value range of N?- Answered above
5. Is this measurement noise Gaussian, and can we use the mean instead of the expectation?- Can you elaborate on the question?
6. During the calibration/restore process of the TX phase shifter, the document does not require us to perform INL correction on the measured PS Cal Result Array DegreeTXm (0 to 63) value, but only requires us to save and restore it to the AWR device . So how does the AWR device correct the INL caused by the phase shift? According to this process, the AWR device can completely complete this process automatically. Why do we need to implement this process manually? –
   1. Based on the restored PS Cal Result Array DegreeTXm (0 to 63) value, the device internally computes the required correction settings
   2. Can you specify the process that you are mentioning?

Regards,

Swati

Hi Swati,

Thanks for your answers!

1. The noise mentioned in point 5 above refers to the measurement noise during the calibration process.
2. For PS Cal Result Array DegreeTXm (0 to 63) in point 6, for example: what I get is ([0, 1.6, 3.18, ..., 62.5] << 10), should I adjust it be ([0, 1, 2, ..., 63] << 10 + Temperature Correction LUT (Tpresent)) and then restore it to the AWR?

Regards,

Libo

Hi Libo,

1. I am still not clear about the question. However, we have not plotted the noise vs calibration so we cannot say for sure if the noise would be Gaussian.
2. Keeping these values would add to non linearity error. However, for  PS Cal Result Array DegreeTXm (0 to 63), how are you getting the values of 62.5? Can you share your calculation?

Regards,

Swati

Hi Swati,

1. I see
2. 62.5 is just my assumption. According to the statement in the document, I understand that the PS Cal Result Arraytxn (0-63) is a vector containing 64 elements. Each element in the vector is a Q10 decimal. This vector is saved by the AWR device returned The step -length measurement value of the phase shift, its denominator is 360 °. PS Cal Result Array Degreetxm (0-63) is to convert it to an angle value. This value should not be as big as 62.5, he should be the decimal between [0, 1]. Is my understanding right? Regarding it, is the correction method I mentioned in the previous question is right?

Regards,

Libo

Hey Libo,

The value should be a Q10 decimal and the value should be between [0,1023]. Let's suppose, for (0, 1, .. ,63), tx0 phase shifter setting is [0, 5.625, .., 354.375] and your measured phase in degrees is [0, 5 , ..., 356], then the 10 bit quantized LUT measured phase (to be input in the API) is [0, 14, ...] (measured phase * 1024/360 (degrees)]. 

Regards,

Swati

Hi Swati,

In the antenna calibration stage: we need to manually align the corner reflector to Rx1, how is this process achieved? Because, if we want to extract 0° by radar, we need to solve the angle first, but the antenna has not been calibrated at this time, and the data is wrong; and if we do not apply radar measurement, how do we ensure the corner reflector‘s effective point is at 0° of Rx1?

Regards,

Libo

Hey Libo,

Your understanding is correct. Radar data is wrong and cannot be used to do the calibration. The corner reflector has to be aligned externally (manually) to 0 degrees of all TX channels. The value measured vs the actual value (at 0 degrees aligned manually) will give the final calibrated value.  

Regards,
Swati

Hi Swati,

What I want to know is the method of manually aligning the corner reflector with the Rx, because our products need to be mass-produced, so in the production process, we need a standardized method to achieve it. Do you have any good method?

Regards,

Libo

Hey Libo,

You have to align the boresight of the corner reflector to the origin of the RX array at a recommended distance of 6m in a controlled environment. You can have a test fixture to do this, or use laser distance meter to do this.

Regards,

Swati 

Hi Swati,

Thank you for your answer
But this method is what we use in the lab. Our products need to be mass-produced, and this method cannot be implemented in batches in factories. Do you have an operation method that can be applied to mass production in the factory?

Regards,

Libo