Square rooter

[epilot]

Hello,

Does anyone have any info about "square root" or "square rooter" circuits?

I have heard that they are used in sound devices, is this correct? If so what is their Benefit there.

I made a search about those circuits but the result was not so good.


Thanks.

 

[Nigel Goodwin]

No idea what you are on about!.

Only thing that comes to mind is doing true RMS readings on AC voltmeters?.

 

[stevez]

One very common thing to measure is the volume of air or gas and with a known cross section one only needs to know the velocity at any given moment to know the volume.

Pressure is relatively easy to measure. The square root of the pressure times a constant will yeild velocity.

I know on old pneumatic control systems there was a device we used in boiler controls called a square root extractor. It was a device of non-uniform cross section that essentially delivered the square root that was needed. Electronics systems and pneumatics systems were done in building blocks that were patterned after each other. I'd expect there to have been electronic square root extractors - most likely some kind of amplifier or attenuator that exihibited a non-linear, square rootish behavior.

Kind of a long winded way of suggesting googling on square root amplifier, extractor, etc - consider looking under industrial controls - possibly ISA publications.

 

[stevez]

I need to correct myself - pressure is easy to measure. Velocity pressure can be measured with a probe facing directly into the flow - and after static pressure is deducted, the result is velocity pressure. The square root of velocity pressure times a constant yeilds the velocity.

 

[eng1]

You can extract the root of a voltage with log amplifiers: take the log, divide by 2 and then take the antilog.

 

[zevon8]

See if you can find a Natiional Semiconductor Linear Devices handbook. There may be one available online. In it you will find all sorts of circuits for op-amps that will do log, square root, etc. Very common circuits used to do caluations in industrial control circuits.

Here are some examples, the linear handbook has far more:

**broken link removed**

 

[Hero999]

Wow, these are kind of like analogue computers.

 

[zevon8]

Yes, that is exactly what they are. Often in industrial controls you would need to do some cacluation, and this is how you do it. On machinery you could have sensors measuring a speed, weight etc and want to know some other variable that is very difficult to measure, but easy to calculate. You do it with op amp calculations, and show it on a display on the control panel.

The example of volume and pressure given above is a perfect example. One place I worked we developed controls for the sheet metal industry. You would need to do calculations of area and line speed of material that was coming off a very large coil that was rapidly getting smaller as the material came off the roll. You could not have a roller measuring speed, and often the material entered an accumulator where it stopped moving all together. The only data you had was the thickness of the material and the spped the coil was turning, the rest you calculated.

 

[Roff]

Check out an analog multiplier, such as Analog Devices' AD633 (page 5).

 

[zevon8]

Very useful, thanks for the link. People tend to forget how things were done "back in the day", LOL. But seriously, even with computational power so cheap these days you can't beat one chip square root or multiplication when that's all you need. Even using a PIC, you still have the developement time for the code, debugging, etc. One engineer I worked with did everything with either a LM324 or a LM339, from complex machinery controls to intrinsically safe battery chargers used in mining.

 

[Roff]

For fun, I once made a "first and third" generator, whose input was a sine wave and whose output was a waveform which was the fundamental plus 1/3 of the third harmonic. The idea was to generate the first two terms in the Fourier series expansion of a square wave. I did it with two analog multipliers (LM1495's, now extinct), and maybe an op amp (I can't remember - this was circa 1970).
The key identity is

sin 3(x) = 3*sin(x) -4*sin^3(x).

I then had to add 1/3 of that to the fundamental. The first multiplier generated sin^2(x), and the second one took sin(x) and sin^2(x) and generated sine^3(x).
You can do this with a filter, but it will only work at one frequency unless the filter is tunable, such as a switched capacitor filter. I don't think they were available in integrated form in 1970, and besides I probably wouldn't have understood them then.
Below is what the output waveform looked like.

This is obviously the rambling of an old fool.

 

[epilot]

Check out an analog multiplier, such as Analog Devices' AD633 (page 5).

Hi Ron,

Thanks for the link,
yes I saw that circuit before, I want to use it but my circuit is working with 12 to 15V there is no NEGATIVE voltage for this square rooter,
I bought few 633 IC's few months ago for 5 or 6$ but did not use them yet,

Anyway the power supply is not able to generate any negative voltage, is there any way to convert or modify this circuit for a common power supply?

I need it for a kind of audio amplifier and want to get a square root from its ouput signal, to remove or limit kind of noise.

 

[audioguru]

Ron,
It looks like the output of a vacuum tube guitar's amp.

 

[Roff]

Hi Ron,

Thanks for the link,
yes I saw that circuit before, I want to use it but my circuit is working with 12 to 15V there is no NEGATIVE voltage for this square rooter,
I bought few 633 IC's few months ago for 5 or 6$ but did not use them yet,

Anyway the power supply is not able to generate any negative voltage, is there any way to convert or modify this circuit for a common power supply?

I need it for a kind of audio amplifier and want to get a square root from its ouput signal, to remove or limit kind of noise.

I don't think you can take the square root of audio, because you can't take the square root of a negative number.

 

[epilot]

I don't think you can take the square root of audio, because you can't take the square root of a negative number.

Yes that is correct, but you can do the job with offsetting the adudio signal.at the moment the biggest groblem is finding a sq rooter circuit with a common supply...

 

[Roff]

Yes that is correct, but you can do the job with offsetting the adudio signal.at the moment the biggest groblem is finding a sq rooter circuit with a common supply...

OK, if you say so...
This circuit just creates an "artificial GND" at VCC/2, similar to many single-supply op amp circuits.

 

[epilot]

OK, if you say so...
This circuit just creates an "artificial GND" at VCC/2, similar to many single-supply op amp circuits.

Ron,
Thank you very much for the circuit.

most of opamp circuits I have seen use TL072, I have no access to it, can I use LF411 or LF351 instead?
I think the answer is yes because it seems there is no difference rather than freq response and..

I"ll try to make your circuit tomorrow, hope to get a good result with it on audio freq.

 

[Roff]

Ron,
Thank you very much for the circuit.

most of opamp circuits I have seen use TL072, I have no access to it, can I use LF411 or LF351 instead?
I think the answer is yes because it seems there is no difference rather than freq response and..

I"ll try to make your circuit tomorrow, hope to get a good result with it on audio freq.

Those op amps should be OK.

 

[epilot]

Those op amps should be OK.

Ron, I made it, there is a problem, the output voice is not so clear(cut cut or something like that is in the voice) like when the dc voltage of an amplifier is not perfect?!
Do you have any idea?
perhaps we need to make more modification?

 

[Roff]

Ron, I made it, there is a problem, the output voice is not so clear(cut cut or something like that is in the voice) like when the dc voltage of an amplifier is not perfect?!
Do you have any idea?
perhaps we need to make more modification?

1. You won't be able to understand the square root of audio.
2. If you don't have the offset pot set high enough, I think some sort of clipping will occur. The op amp offset voltage should not be important, since this is an AC-coupled application.