• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 30 subscribers
  • Views 15546 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

IWR1443BOOST: Sub-millimeter accuracy distance measurement on a short range (from 5 to 30cm).

Viktor Varjas
Prodigy 80 points
Part Number: IWR1443BOOST

Hi,

In our medical device development project we would like to use the IWR1443BOOST for accurate short range (from 5cm to 30cm) distance measurement with a resolution around 0.1 mm.

As I understand at this point, just by relying on the frequency shift of the IF-signal using the 4GHz max Bandwidth of the device the max resolution is only around 5 cm (as observed with the drone altitude demo and calculated with the "mmWave Sensing Estimator"). However using the phase of the IF-signal it should be possible to reach sub-millimeter accuracy (vital signs demo). On the other hand, in all examples I found that the phase-based distance measurement was only used in the context of vibrating/periodically moving objects (water surface, engine, heart beat, etc).

In our case we would like to measure an object moving on the mentioned very short range (25cm) with high (0.1mm) resolution. We would like to have this fine resolution on the full range. As an example: the object can be 70.4mm away from the sensor and it can move to a position which is 125.7mm away from the radar and just stay there. So the motion of the object is not periodic at at!

Is it possible to configure the device to achieve the above mentioned goal? If yes how to do it? (Basically we want to substitute a laser distance gauge with this radar).

Thank you in advance,

Viktor Varjas

  • 0
    TI__Mastermind 30405 points

    Hi Viktor,

    TI's mmWave devices can indeed achieve measurement accuracy of 0.1mm.

    You are right that with standard FMCW processing, the best range resolution you can achieve is 4-5cm (with the maximum bandwidth), however it is important to distinguish between Range resolution and Range accuracy and you can achieve sub-millimeter accuracy using a 2 stage algorithm, where you first get an initial range estimate using standard processing and then perform a zoom FFT around the range of interest to get finer accuracy. In fact, we are developing a high accuracy demo to demonstrate this capability, and it will be released on TI.com.

    Difference between Range Resolution and Range Accuracy:

    In conventional measurements, Accuracy cannot be better than the resolution simply because Resolution is the finest your scale can measure. However, in Radar terminology, Resolution (Range Resolution) is always referred to in context of two (or more) objects and is defined as the smallest distance between two objects that allows them to be detected as separate objects. This is a physical limit (Range Res = c/2*B, where c is the speed of light and B is the sweep bandwidth) and depends on the chirp sweep bandwidth that the radar sensor can provide. The larger the sweep bandwidth, the better the range resolution.

    So fundamentally, with a maximum bandwidth = 4GHz, the Range resolution is about 4cm.

    However, Range accuracy which depends upon Range resolution can be better than range resolution:

    Range Accuracy

    This is often defined as a rule of thumb formula for the variance of the range estimation of a single point target as a function of the SNR. Essentially, this is a measure of how well we can determine the true range of the target. Range accuracy is a function of Range Resolution and SNR.

    *Note: This is the theoretical maximum accuracy and additional post processing may be needed in software to achieve this value. The theoretical max range accuracy is a function of Range resolution and SNR. It is defined as Range Accuracy = Range Resolution / Sqrt (2 x SNR). However, the achievable accuracy also depends upon (limited by) the size of the FFT as it defines the FFT resolution (aka range inter-bin resolution). Intuitively, the FFT granularity needs to be fine enough to measure the desired range increment and so the application may need to first get an initial coarse range measurement using the provided chirp configuration and then perform a zoom FFT on the range of interest, using additional processing in software with a higher order FFT.

    Please refer to the following threads on considerations related to high precision measurements.

    https://e2e.ti.com/support/sensor/mmwave_sensors/f/1023/t/597434

    https://e2e.ti.com/support/sensor/mmwave_sensors/f/1023/t/612730

    Regards

    -Nitin 

  • 0 in reply to Nitin Sakhuja
    Prodigy 80 points
    Thank you Nitin, this answer was very helpful! Especially the note regarding the object being stationary. Our object is moving quite fast sometimes and distance has to be sampled with a high frequency (~100Hz), so there is no time to wait for seconds before an accurate measurement.
  • 0 in reply to Viktor Varjas
    TI__Mastermind 30405 points
    Hi Viktor,

    I think I need to elaborate on the stationary part. The fundamental requirement is not that the object be stationary for high accuracy measurement but to achieve a good SNR (about 20 dB). Depending upon your object, if you can get the required SNR, you can achieve the desired accuracy of 0.1mm at 100 Hz. In fact, the high accuracy demo we are working on right now does not require the object to be stationary. We will have the demo available on TI.com in the next couple of months.

    The thread I referred to you earlier is related to extremely high precision distance measurement (50 microns) which is why the object is kept stationary for a couple of seconds to get a good SNR by accumulating more chirps.

    Can you provide an idea of the object being detected in your scenario?

    Regards
    -Nitin
  • 0 in reply to Nitin Sakhuja
    Prodigy 80 points

    Hi Nitin,

    The object is a plastic circle with 12cm^2 area (thickness 1-2mm). Would that work out? What would be the best material to use?

    The surface normal of the circle is not orthogonal to the radar-plain!

    Thank you in advance,

    Viktor

  • 0 in reply to Viktor Varjas
    TI__Mastermind 30405 points
    Hi Viktor,

    If you have the flexibility to choose the material, metal would be the best choice as a radar target. Even better if the target is shaped like a corner reflector. Given that the target is relatively small with a radius of 1.95 cm, you may need to use a narrow radar beam to reduce clutter and improve SNR. This can be done through the use of a lens (made of dielectric material) or antenna design.

    We haven't tried this scenario so it would be hard to give a definite answer on the achievable accuracy in this setup but with a metallic target, it might be achievable. The best way to figure this out would be to try to detect the object in it's desired orientation in stationary state first and see what kind of accuracy you can get.

    Regards
    -Nitin