0
\$\begingroup\$

I am designing a receive signal path for Ultrasonic Transducer.

The circuit needs to measure time of flight of ultrasound wave pulses with frequency of 1MHz(about 16 periods - 16us).

Below is the waveform measured. Yellow is Excitation Pulses to transmitting transducer, and Blue is signal received with receiving transducer.

enter image description here

The problem is that the amplitude of the received signal may change due to transducer spacing, or change in transducer mounting position, temperature, and other various reasons i don't know. And due to mechanical property of the transducer, the received signal amplitude naturally ramps up gradually, then ramps down.

I am thinking of applying Automatic Gain Control Circuit after 1MHz bandpass filter to feed to comparator.

I've looked at few options for AGC, but can't decide how to implement.

Unlike RF signals or sound signal , ultrasonic pulses don't have continuous input level. So, the AGC has to respond fast, adjusting within 1 period(1us).

1) Will traditional AGC configuration using amplifiers such as MC1350 or MAX9814 work in this particular application? which configuration would you recommend?

I have to keep the output signal of AGC at constant voltage amplitude(if there is signal), with DC offset of 2.5V. The jitter in Time of flight measurement is critical(1ns jitter may result up to 30% error in some application)

\$\endgroup\$
  • \$\begingroup\$ If there is no information in the amplitude itself, why not use a high(ish) gain combined with a limiter? \$\endgroup\$ – jippie Feb 24 '16 at 6:06
  • 1
    \$\begingroup\$ @jippie Wouldn't gain too large amplify noises also? \$\endgroup\$ – Steve Feb 24 '16 at 6:08
  • \$\begingroup\$ I would think AGC would make things even worse: it would over-amplify your noise. You seem to have a pretty clean signal, if you run it through an envelope filter and compare it against a threshold based on the signal average, it should pick out the return pretty well. \$\endgroup\$ – uint128_t Feb 24 '16 at 6:30
  • \$\begingroup\$ How could AGC "overamplify" the noise any worse than constant high gain + limiter? Also, what you're suggesting is a kind of AGC. \$\endgroup\$ – biggvsdiccvs Feb 24 '16 at 7:10
  • \$\begingroup\$ From your waveforms it seems to me that the front edge is ill-defined (any threshold you set will be amplifcation-dependent), but the back-edge is well defined: it is the peak of your received signal, where the derivate of the evelope goes from positive to negative. I am not an analog specialist, but it seems to me that you could catch that in a circuit. And it would be independent of any (slow) AGC. \$\endgroup\$ – Wouter van Ooijen Feb 24 '16 at 10:55
2
\$\begingroup\$

Implementing an approximation of an ideal exponential time-variable gain (TVG) amplifier is certainly a viable approach in ranging applications. The fall-off of response pulse (or chirp) with time is both enormous and predictable and can be thus profitably compensated for.

Polaroid's ultrasonic ranging system (marketed in the mid-1970s, and used for such commercial applications as door openers as well as the original application of autofocus cameras) used such a method- using an ASIC. You should be able to find the patent.

enter image description here

I've designed a few such systems used for such disparate applications as extruder control, film making and vending machines.

If you don't need a huge range, consider varying other parameters with time rather than trying to use a general purpose AGC. There are, no doubt, far more sophisticated methods that could be applied, implemented with DSP technology, chirp transmit signals, and the knowledge of the expected drop-off of return signal strength.

Resonance (and Q) of the transducer(s) can indeed be a major factor, depending on what your actual requirements are.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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