I am currently doing a project for bicycle overtaking safety in which an approaching vehicle or cycle from behind is found through ultrasonic sensor and indication is given to the rider.

I had a problem that the Ultrasonic sensor HC-SR04 is more divergent when the distance to be measured is more. The safe distance as per design is 3 m and using an Arduino microcontroller

The requirement is to obtained a narrow beam but without any reduction of signal intensity and without any reduction in sensing distance

Can anyone please tell me how to collimate the Ultrasonic Beam so that the angle is less but the sensing distance remains the same ?

Appreciate your help in advance.



2 Answers 2


There are several ways to focus ultrasound, but none of them are particularly easy, especially when the device has to be small, mobile, and take little power.

Something as simple as a exponential horn might be good enough. This has the advantage of providing a better impedance match between the ultrasound sensor and the air, but there is no way to avoid it taking space. I experimented with a exponential horn in a ultrasound project once and it did create a more narrow beam than the bare sensor, and sound was louder in the center of the beam than it was with the bare transducer.

Lensing (refraction) is possible, but pretty much inappropriate for your application. I once saw a demo of sound being focused by a lens made from thin plastic sheets and filled with CO2. The was a little pressure inside the lens so that the plastic sheet bulged out a bit on each side, forming a convex lens shape. The speed of sound is different in CO2 than air, so the whole thing acted like a convex lens and focused the sound in a surprisingly tight spot.

The best choice for your application is probably a phased array. Arrange several transducers on a plane perpendicular to the direction you want to transmit in, and fire them in unison. With the array spanning at least 1/2 wavelength in dimension, you get a focused beam, although it will have side lobes. The more transducers, the less power goes into the side lobes and the tighter the main lobe. For extra credit, arrange the tranducers in a Poisson disk pattern.

  • \$\begingroup\$ Parabolic dish is another approach, but it will be similar in size to the horn. \$\endgroup\$
    – Dave Tweed
    Dec 29, 2013 at 23:07
  • \$\begingroup\$ Olin Lathrop : Thank you for the idea and suitable explanation \$\endgroup\$
    – Anand M S
    Jan 5, 2014 at 19:49
  • \$\begingroup\$ Olin Lathrop : Thank you for the idea and suitable explanation. I can't go with phased array due to budget limitations in project henec thoght of implementing exponential horn. Can u please tell me or share the info about the construction mechanism and techniques involved me in making exponential horn? \$\endgroup\$
    – Anand M S
    Jan 5, 2014 at 19:59
  • \$\begingroup\$ @Anand: The time I experimented with one, I had the computer calculate the crossection outline and print it on paper at 1:1 scale. I took that to a machine shop that milled out the horn from a piece of round plastic (Delrin, I think) stock. \$\endgroup\$ Jan 5, 2014 at 20:52
  • \$\begingroup\$ @Olin Lathrop One question does ultrasonic give a good performance with Plastic material.. I read somewhere that it might lead to absorption and reduction in sensing distance.. Once again thank you so much for your input and help.. Really Appreciate it.. \$\endgroup\$
    – Anand M S
    Jan 5, 2014 at 23:13

I have had good luck using coin rollers (US dime, penny and nickel). The dime rollers just fit over the two shiny cylinders, but if your rollers have a "pre-rolled" ridge on one end, only the nickel (or larger) rollers are big enough.

Before using the coin rollers, I had many problems with "multi-path", where the signal was bouncing off obstacles that I was not interested in before striking my desired targets. With the coin rollers (one each over TX and RX cylinders), I get more results from my desired targets. YMMV, so try several...


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.