Ultrasound transducer for bat detector project

Working fine here. Sure you don't have a virus or something? Have you tried "Insert Image" instead of "Manage Attachments"?

I cleared my temporary internet files and cleared my cookies then I logged in.
This site is still extremely slow to open and load. I still cannot attach a schematic because the "upload" button times out after waiting for one minute.
I tried to "insert image" but it also timed out and dropped the connection to this site.
I tried "insert image" again but it does not insert it.

ALL other websites I visit work immediately and perfectly.
Something is still timing while I type this.

I think this website works properly today.
Here is my schematic:

Excellent. AG's right about that unreferenced + input on the amp, breakshift - on the other side of that blocking cap it can float to any voltage. This circuit also provides high-pass filtering with a rolloff that starts in the audio range.

Thanks for the advice. I did it that way because I was copying the circuit from Figure 3 (non-inverting) from:

https://www.electro-tech-online.com/custompdfs/2008/01/sloa058.pdf

What's wrong with that circuit?

Believe it or not, the TI circuit in the spec sheet is wrong, and AG's circuit is right. What's the voltage at the non-inverting terminal? You have the cap blocking DC on one side. You have the opamp presenting near-infinite resistance on the other. Without a DC reference at that non-inverting input, it can "float" to any voltage.

duffy said:

Believe it or not, the TI circuit in the spec sheet is wrong, and AG's circuit is right.

Click to expand...

TI hired a stupid kid straight out of school to draw that schematic. Then the schematic is missing some intelligence and experience.

Oh, I see. Well thanks for taking a look at my schematic

So you've connected the feedback loop to ground - why not to Vcc/2? I thought it should be connected to Vcc/2 since its grounded for dual supply op amps. I guess it doesn't matter so long as R1 is there to drop volts to keep both inputs at the same potential?

And the 10nF cap you put in the feedback - to stop DC current flowing through the feeback loop (saving power)?

And the 1nF input cap - isn't that a bit low? That's presenting a few thousand Ohms across my frequency range.

My circuit has a DC gain of 1 so its input offset voltage is not amplified.
If the feedback connects to the input reference voltage then it has plenty of DC gain that amplifies its input offset voltage.

First I designed an audio amplifier with a response down to 40Hz. Then I realized that you are amplifying ultra-sonic bat noises not sounds so I changed all the capacitors to cutoff frequencies below about 10kHz but I forgot to change my note about 40Hz.

The 10nF cap for the feedback to ground is a highpass filter with the 3.3k resistor in series have a cutoff frequency of 4.85kHz.
The 1nF input cap with the 100k input resistor have a cutoff frequency of 1.6kHz. You can barely see any attenuation at your ultrasonic bat frequencies. The output cap and 10k load have a cutoff frequency of 7.3kHz. All the filters combine to produce a cutoff frequency of about 10kHz.

I see. Having drawn up your circuit in LTspice I can see how you've shaped the frequency response. I think I have the pre-amp stage sorted now (at least a preliminary design), so thanks for that The next stage is the actual heterodyning.

I'm interested in frequencies from 15kHz to 125kHz. If mixing down to ~10kHz, I'll need an adjustable oscillator that can do 25kHz to 135kHz. One thing that strikes me here is how when tuned on a specific frequency, if the incoming ultrasound rises in frequency then the user will hear this as a shift down in frequency, because the difference is reduced. And vice versa.

I would much prefer it to be normal - how could I go about doing this?

I made a voice scrambler circuit for a wireless boardroom tele-conference system. I used a balanced modulator IC that heterodyned voice sounds with an oscillator then I filtered only the reverse difference sounds. High frequencies sounded low and low frequencies sounded high. It worked very well.

I don't know how you can heterodyne and keep the same low= low and high= high sounds. Maybe you must record the bat noises then play them back slower.

If I go for a double superhet, this will reverse the effect and hence remove it. This is what I've come up with. This is just a first stab kind of effort - I don't know how easy it will be to implement. It can easily be modified.

Incoming f: 15kHz to 125kHz

Local oscillator 1: variable from 265kHz to 375kHz
IF: 250kHz

So at f(in)=15kHz, and LO1 at 265kHz ---> 265 - 15 = 250kHz
And at f(in)=125kHz, and LO1 at 375kHz ---> 375 - 125 = 250kHz

Filtering at the pre-amp stage will stop image frequencies getting to the IF stage. A narrow passband filter to pass IF +/- 5kHz.

Now IF signal mixed again with a second oscillator, this time fixed at 260kHz.

So LO2 - IF = 260kHz - 250kHz = 10kHz

Followed by a low-pass filter and audio amplifier etc.

I went for a 10kHz audible range, from 5kHz to 15kHz. Perhaps I should increase this to something like 20kHz, so that the user is more likely to hear something...?

What do you make of this superhet idea? How easy will this be to implement and what's the best way of doing just that? I assume something like an IC for the mixer & variable oscillator, either IC or lumped component IF filter, and another IC for the second oscillator/mixer?

I'm aware of how old the superhet design is now - but it should be fairly easy to get components for this, right?

I don't know if the old 455kHz IF transformers used in AM radios are still available. Their bandwidth was too narrow anyway.