Newbie: analog delay circuit for delays of 1us to 50us

Moderator edit:
Title changed after move.

Hello,

I am trying to learn electronics by doing a simple project. I have broken this down into several "baby steps" so as to be simple enough for me to build up my knowledge bit by bit and spur me on.

Context: I would like to create a simple delay circuit that takes as its input an analogue signal and generate 4 new signals, each a replica of the original but with a time delay. I would like to experiment first with signals in the audio range (up to 20KHz) and then the ultrasound range (up to 40KHz) range. I am looking for delays of 1us to 50us.

After scouring the Internet and reading up on some stuff I figured an all-pass filter would serve my purposes (555s, simple RC filters, bucket brigade ICs and Linear's Timerblox don't seem to fit). I set-up an LTspice simulation using a bank of 4 all-pass filters just to see what is possible. After some playing around with this something did not seem ok (delay seemed to depend on the frequency of the (square) wave I was using).

Issue: After some more searching I found a thread in this forum, which explained (via LTspice AC analysis simulation, which I also reproduced), that only a phase shift is done at a given frequency. More specifically the "all" in the all pas filter refers to the gain that remains close to 0dB over the full frequency range.

Question: so my question is, according to the specs above, what is the best way to delay the signal with that rage and precision for the full frequency range? Is it doable with a minimum of distortion? Appreciate any suggestions and/or pointers.

TIA.

delay, analog, audio, ultrasound, microseconds

Have you considered delay-line technology?
http://en.wikipedia.org/wiki/Delay_line_memory

Alec_t said:

Have you considered delay-line technology?
http://en.wikipedia.org/wiki/Delay_line_memory

Click to expand...

Yes I did. However a search Digikey did not help me.
I found that:
1. Most of the ICs I found were in the nano second range.
2. Those that were in the microsecond range had few taps
so I would require several of these chips (complexity)
3. Packaging does not seem to be easy to work with (surface mount, I
was hopping to breadboard this)
4. The data-sheet of Linear's product seems to be designed for digital signals (I assume the same for the others)

Do you have any suggestion of any of these for use in analogue?

Thanks.

hugo23 said:

I am trying to learn electronics by doing a simple project.

Click to expand...

This is an unusual choice for a "simple" project for a newbie. Flashing LEDs or crystal radios are more typical.

What are you really trying to do? You might be amazed at the creative alternatives you'll hear if you just ask.

wayneh said:

This is an unusual choice for a "simple" project for a newbie. Flashing LEDs or crystal radios are more typical.

What are you really trying to do? You might be amazed at the creative alternatives you'll hear if you just ask.

Click to expand...

Granted it is unusual. I am simply trying to set-up an experiment in order to complete a more complex and challenging project which I can only built step by step:
1. emulate the delay that occurs when sound travels to several (2) sensors
2. use this input to calculate the TDOA (time difference of arrival).

Naturally its point 2 that is my goal, which will also be broken down in very simple exercises. Point 1 just serves as a basis for experimentation and is a way of avoiding starting with the complications of acoustics.

BTW, wasn't aware of crystal radios. Interesting.

Thanks.

A very easy solution is a PT2399 echo chip. These chips digitize your analog input and store in registers. Once the register bank is full, they start feeding the output with the oldest stored input value. The rate of sampling determines the echo delay - faster sampling rate means less delay, slower sampling is longer delay. Just a handful of resistors and capacitor.

You might want to add a jfet input stage (or op amp follower) if you are using this to drive a guitar or other high impedance audio source.

datasheet pt2399

hugo23 said:

...
1. emulate the delay that occurs when sound travels to several (2) sensors
2. use this input to calculate the TDOA (time difference of arrival).

Naturally its point 2 that is my goal, ...
...

Click to expand...

You might want to check out this project;
http://www.romanblack.com/SonicRanging/Sonic_Ranging.htm

All you need to do is hook up two or more microphones to opamps, and measure the different TOAs with any microcontroller (or on a 'scope).

To determine the sound sorce location you will need to process a lot of math anyway, which will be a microcontroller solution as you are not going to be able to do that math in analogue hardware.

My suggestion is to get up to speed with microcontrollers at the start, and the rest of the project will be easy. Or hook up with someone who can do the micro side.

GopherT said:

A very easy solution is a PT2399 echo chip.
....
You might want to add a jfet input stage (or op amp follower) if you are using this to drive a guitar or other high impedance audio source.

datasheet pt2399

Click to expand...

Initially I had looked at DIY audio stuff but came to the conclusion that the time delays were to large. I have looked at the data sheet and the shortest delays are in the tens of milliseconds range. I would need 3 orders of magnitude less. The objective is to measure TDOA between mics.

Thanks for the suggestion.

THE_RB said:

You might want to check out this project;
http://www.romanblack.com/SonicRanging/Sonic_Ranging.htm

Click to expand...

This is what I am thinking about. I will have to read it more carefully though.

THE_RB said:

All you need to do is hook up two or more microphones to opamps, and measure the different TOAs with any microcontroller (or on a 'scope).

Click to expand...

I will start with a scope. Have a 20 year old 20MHz Kikusui with microsecond range, so that should do it.

THE_RB said:

To determine the sound sorce location you will need to process a lot of math anyway, which will be a microcontroller solution as you are not going to be able to do that math in analogue hardware.

My suggestion is to get up to speed with microcontrollers at the start, and the rest of the project will be easy. Or hook up with someone who can do the micro side.

Click to expand...

Micro-controllers should not be a problem for me (after getting up to speed of-course). Used some a very long time ago (25+ years). I was thinking Arduino. Maybe some PIC controller. Anyway, that will come later.

Thanks.

Additional ifno you can get by googling "Laurent Kneip ears robot mic amplifier distance sound".

Somewhere there you could find the concept of "window" and "autocorrelation" straight to the point. Do not waste your time simulating. Measure the real thing as RB says.

atferrari said:

Additional ifno you can get by googling "Laurent Kneip ears robot mic amplifier distance sound".

Somewhere there you could find the concept of "window" and "autocorrelation" straight to the point. Do not waste your time simulating. Measure the real thing as RB says.

Click to expand...

Found the link. Looking at it now.

Thanks.

hugo23 said:

Found the link. Looking at it now.

Thanks.

Click to expand...

If by any chance you run across an Elektor (Spanish - 2007) issue with an article about robot ears, be warned that the translation of certain concepts are wrong. Checked that with Laurent myself.

Would you post the outcome of your tests? I have that in my list since last year.

Tested the SAD concept in Excel and it is essentially simple to see at work.

atferrari said:

If by any chance you run across an Elektor (Spanish - 2007) issue with an article about robot ears, be warned that the translation of certain concepts are wrong. Checked that with Laurent myself.

Click to expand...

Did not. But thanks for the heads-up.

atferrari said:

Would you post the outcome of your tests? I have that in my list since last year.

Click to expand...

No promises. Let's see how far I can go.

atferrari said:

Tested the SAD concept in Excel and it is essentially simple to see at work.

Click to expand...

Isn't that "simulating"?

hugo23 said:

Isn't that "simulating"?

Click to expand...

No; it was implementing the SAD calculation to confirm it worked as I understood.

When I say "simulating" is related to avoid creating two signals delayed to each other as in your OP but having instead a real source of sound (simple speaker emitting audio bursts) and two microphones at different distances (thus different TOAs) to implement the basic measurement and further calculation.

atferrari said:

When I say "simulating" is related to avoid creating two signals delayed to each other as in your OP .....

Click to expand...

I see. Thought you were referring to the SPICE simulation.