Converting the amplitude of a received ultrasound signal to a value

[DSG]

I would like to convert a received ultrasound signal to a value that is proportional to the amplitude (signal strength) of the received signal.

The transmitter will be either 237 KHz (or maybe 40 KHz) pulse. Of course, the received signal will be a 237 KHz (or 40 KHz) signal but increasing in strength until it reaches a peak value. I don't know enough about circuits, but one thought is to convert the signal to an analog signal that follows the peak values and use an A/D converter to read my values. Any suggestions?

DSG

 

[Analog]

You need a peak detector. Google circuits with that definition for a start.

 

[jpanhalt]

The transmitter will be either 237 KHz (or maybe 40 KHz) pulse. Of course, the received signal will be a 237 KHz (or 40 KHz) signal but increasing in strength until it reaches a peak value.

Not if the object is moving. Is this application for stationary mapping or detection of objects? John

 

[DSG]

I plan on transmitting about 10 cycles at the proper frequency and then monitoring the echo returns. There may be several objects from which to reflect the signal. I would like to then create a window around a baseline and monitor (alternately transmitting a few cycles and listening to the returns) for a significant change in the returns from the baseline.

Now that I am talking about it, I guess this would be a type of motion detector but motion was not what I was trying to detect. I'm looking for objects that change the "picture". I did want something a bit "smarter" than simple motion detection, hence the desire to analyze the return signal.

DSG

 

[DSG]

I looked at peak detectors and the problem is that I would like to follow each individual peak (i.e., when the amplitude of the signal goes back down) and not just the maximum peak. This is because I can have reflections from several targets. I suppose I can add a resistor to bleed off the voltage when waiting for the next reflection.

The problem with the motion detector is that a motion detector reacts when a threshold is passed and I need to pick up the object from amongst several reflections.

DSG

 

[Roff]

I have designed and simulated a circuit that has the output shown below. If you're interested, I'll clean the schematic up and post it.

Uploading is excruciatingly slow right now!

 

[DSG]

Roff,
Yes, that is the result I was looking for.
DSG

 

[Roff]

Can you use dual power supplies (+/-8V, or whatever you choose), or do you need to stick to a single supply? If the latter, be aware that your output will be elevated above zero volts, i.e., the output for zero signal input will be at half the supply voltage. Let me know what voltages you want to use, if you have any preferences.

 

[DSG]

Roff,
My input power is 24VDC. I was going to regulate it down to whatever I need. Another thought before I proceed... Even using an ultrasound signal at 40KHz, I will not be able to use a PIC to do an A/D sample fast enough to analyze the waveform. Is that correct? I may need to rethink my logic.
DSG

 

[Roff]

Roff,
My input power is 24VDC. I was going to regulate it down to whatever I need. Another thought before I proceed... Even using an ultrasound signal at 40KHz, I will not be able to use a PIC to do an A/D sample fast enough to analyze the waveform. Is that correct? I may need to rethink my logic.
DSG

I think you are correct about the PIC A/D conversion rate, but there are others (Nigel, Wm at his blue room, etc.), who are more qualified to answer that question.

 

[blueroomelectronics]

Problem is you've got to sample a waveform at a much higher rate than the signal. The 16F A/D is way too slow, some high end PICs could handle it though.

 

[blueroomelectronics]

The 18F1230 (18pin DIP) can run a single channel A/D at 100ksps but that would only allow 2x oversampling of a 40KHz signal. 4x or more would be better, many dsPIC30Fxxxx chips can do 500ksps to 1msps. The 30F2011 is 200ksps.

 

[DSG]

Roff,
I can live with the DS PICs suggested by Bill (thanks Bill). They are only a couple of bucks more than a regular PIC. Let's proceed forward with the circuit.
DSG

 

[Roff]

The 18F1230 (18pin DIP) can run a single channel A/D at 100ksps but that would only allow 2x oversampling of a 40KHz signal. 4x or more would be better, many dsPIC30Fxxxx chips can do 500ksps to 1msps. The 30F2011 is 200ksps.

But he's not trying to store a sine wave. He will have sample-and-held peaks that will be stable for most of the 25us period of the sine wave. See the waveform I posted on one of my previous posts.

 

[3iMaJ]

You might actually use a matched filter since it would be optimal for your application. It may get complicated if you're going to measure doppler though, but I don't think you're operating at a high enough frequency to monitor the changes you're interested in. Unless you pulse the sonar at a repitition frequency and then you can measure doppler differently.

Also, if you use a matched filter then you won't have to sample at a high rate, just a few times around where you believe the output of the filter to maximize.

 

[Roff]

Roff,
I can live with the DS PICs suggested by Bill (thanks Bill). They are only a couple of bucks more than a regular PIC. Let's proceed forward with the circuit.
DSG

This is a can of worms.
Do you prefer surface mount (SMT) or thru-hole components?

 

[blueroomelectronics]

I didn't see a circuit for a peak detector?
Even if it's just sampling the violet waveform you should trigger the conversion with the original waveform if you want to avoid oversampling.

The 18F1230 has a DIP version.

 

[Roff]

I didn't see a circuit for a peak detector?

I didn't post the circuit. I simulated one, and posted the waveform. I'm working with DSG here to come up with something "easy" to build, as regards supply voltage(s), packaging preferences, etc. I'll post the waveform again here.

Even if it's just sampling the violet waveform you should trigger the conversion with the original waveform if you want to avoid oversampling.

The 18F1230 has a DIP version.

I realize that the samples need to be synchronized with the input signal. I'll need to provide a start conversion pulse. Do you know how it should be timed relative to the waveform, and how wide it needs to be?

 

[blueroomelectronics]

The pulse to start the A/D can be edge triggered as soon as the peak has been detected. The PIC can return a reset pulse as soon as it's finished conversion ~10us if needed.

 

[Roff]

The pulse to start the A/D can be edge triggered as soon as the peak has been detected. The PIC can return a reset pulse as soon as it's finished conversion ~10us if needed.

Excellent. I think I already have that edge available.