Hello,
how to calculate what frequency (and bandwidth) is required for an imaging FMCW radar to detect an object of 1mm size (from aluminium) at a distance of about 50cm?
Thank you
The range resolution of FMCW radar is proportional to the reciprocal of the bandwidth, i.e. delta x = c / (2 * delta f). You need to define this based on the tuning range of your radar's sweep oscillator. Range ambiguity is rarely a problem for such short range radars because the sweep repetition time is usually much longer than the propagation delay. For detecting a 1mm object, you will need to pay special attention to the SNR of the return signal. This means using good quality low phase noise sources in your transmitter and receiver conversion chains. Directional antennas and use of mm-wave (Ka-band and above for favorable radar cross-section/wavelength2 ratio) will help as well.
As George Slade pointed out, a high the carrier frequency is is important due to fluctuating RCS (when objects are smaller than wavelength). Furthermore, higher frequency bands allow you higher bandwidths, which are important for the range resolution (range resolution = c/(2*bandwidth), as George Slade mentioned).
You were talking about "imaging radar", so the question is which range resolution you are expecting? Detection might not be such a big problem (though difficult enough), but how about localization ("imaging")? Are 7.5cm range resolution (corresponds to ~2GHz bandwidth) sufficient for you? Do you need additional information besides range (angle or velocity)?
Regards,
Michael Ulrich
Thank you for the answers.
Range resolution is not critical. I am more wondering about the lateral resolution in XY (how small object can it detect).
Velocity is not required but angle is.
What do you think would be the lateral resolution of a 76 - 81 GHz, 4 GHz bandwidth FMCW radar, chip Ti AWR1642?
http://www.ti.com/product/AWR1642
The lateral resolution depends on your array design (antenna positions). With the AWR1642 chip, you have 4Rx and 2Tx, so you can achieve 8 virtual channels (MIMO).
The achievable resolution is the region of arrival/interest (DOA/DOI) divided by the number of channels. The resolution is constant in electrical angle, meaning that the resolution is roughly between 15° and 41°, depending on the angle, for 8 channels and DOA=DOI=180°.
This is probably more than you expected, but you can improve the angular resolution if you can ensure that the received signal is coming from particular directions only (e.g. by using a dielectric lens). For example, your angular resolution is only roughly 2° if you can limit the DOA/DOI to +-8°.
Furthermore, you can use high-resolution DOA estimation algorithms if you are only interested in detection and DOA estimation (not imaging). Using these algorithms, the resolution is rather limited by the SNR than by the number of channels.
Knowing more about your application, it would be easier to answer your question precisely. Do you really need imaging? What do you need angular resolution for? How many aluminum objects are present at the same time?
Regards,
Michael Ulrich
Thanks a lot for the answers.
The application is basically the following - see figure.
The position (angle) of an aluminium column should be detected, probably as a point cloud. The points will then be interpolated to calculate the angle.
The column is completely covered in vegetation (leafs). The columns are connected with wires. There will be likely more than few thousand columns to be detected, one after each other.
It is for an vineyard, the columns are holding the wires with the vine.
The column angle has to be measured in order to adjust the cutting, not to cut into the metallic structure (automatically). A similar machine to the one intended here is shown here https://www.youtube.com/watch?v=nXKH6O1S6As
I was wondering about the range and lateral resolution to estimate if the column and the wires can be "seen" by the radar.
Do you think the AWR1642 will be able to see the structure through the vegetation ( about 30 cm thick)?
Your application is very interesting!
In my opinion, you do not really need imaging. Point detections (range and angle) are probably sufficient. So you rather need angular accuracy (angular resolution is not so important).
The AWR1642 is only a radar chip, which does not include antennas and periphery. It is good for product design, but for a first investigation I rather recommend to use an off-the-shelf evaluation kit for a quick getting-started.
Your original question aims at detecting the wire directly through the vegetation? I think this is not possible because the wire will reflect the EM wave away from the radar and it will be impossible to measure a response (unless the radar is perfectly perpendicular to the wire). I think detecting the poles instead of the wire is more realistic.
Regarding the vegetation, I am not sure whether the radar will be able to measure through it. Generally, a radar with lower carrier frequency will suffer less losses due to the vegetation. I expect that at 77GHz, you will rather measure the distance of the vegetation. However, the poles would probably have a stronger reflection and can be distinguished from the vegetation this way. You will have to try it out, but I recommend using frequencies of ~10 or 24 GHz. Also, don't worry about angular resolution too much. As long as you can resolve multiple targets (poles) in range, you can perform a separate DOA estimation for each pole.
Also, I would think about using a range-only radar system. If distances alone were sufficient for you, the radar system would become significantly simpler (cheaper) and several development kits are already commercially available.
Regards,
Michael Ulrich
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication