-1

In order to make analog calculation (for analog regulator simulation), what would be the means to make an squaring amplifier ?

  • There is an obvious solution with a multiplier circuit as pointed here;
  • I also imagined using a mosfet like in the following drawing and in this case I would need another similar circuit for the negative part of the signal;

but before trying to effectively design it, I wondered if there would be other well known simple solution.


Edit : Some people seem to dislike the application of analog computing. Just think of it as an educational application.

Edit2 : To make things more precise, let's say that to avoid saturation problems, the identity element would be 10v, So for a 1v input we get a 0.1v output and for a 0.1v input, we get 1mV output, and it would be even better if we had a circuit which would give the same positive output for the same negative inputs.

enter image description here

Camion
  • 323
  • 2
  • 9
  • depending on your definition of simple, one method that is very effective is to run the signal through a noise-shaped delta sigma converter, which turns it into a bit stream. Use the bit stream to switch the signal polarity, and filter the output. Linear (now Analog) use this technique in their true RMS converters. – Neil_UK Aug 06 '19 at 13:44
  • 2
    [This question and answer](https://electronics.stackexchange.com/questions/373492/analog-analog-multiplication-part-of-a-hybrid-cpu-for-fun?noredirect=1&lq=1) may be of some interest as it also uses a technique similar to that mentioned by Neil above. – Andy aka Aug 06 '19 at 13:47
  • I you convert it into a bit stream, you loose the properties of being an analog circuit. The mathematics behind this are not the same anymore (z-transform instead of laplace transform), and it can cause instability in a regulation system, if you try to model it as if it was analog – Camion Aug 06 '19 at 14:04
  • 1
    This is too broad as you have failed to state any specifics of the application requirement as it impacts the performance required of the solution and failed to give any actual justification for why an analog solution is required. There are many ways to solve problems but you give no basis for picking the appropriate one, thus failing the SE requirement for specific answerability. All of this was pointed out when you tried to reopen a similarly flawed question. It was explained then what you would need to have a valid question, essentially none of which you have provided. – Chris Stratton Aug 06 '19 at 14:23
  • @Chris Stratton : did you ever heard about analog computers ? This is a very specific scope where one combine "all purpose" modules in order to build regulators and differential equations simulators. I wonder how I can be more specific than that. https://en.wikipedia.org/wiki/Analog_computer#Electronic_analog_computers_2 – Camion Aug 06 '19 at 14:41
  • 1
    Analog computers existed, but are largely extinct as they have many implementation problems. You've failed to state specifics of your requirement ruling out a digital solution or even allowing a selection to be made between analog techniques. **THIS IS A BAD QUESTION** showing that you learned nothing from your meta question about why the other bad question was closed. – Chris Stratton Aug 06 '19 at 14:43
  • The fact that you may not be interested in this in not a reason for other people not to be interested in it. – Camion Aug 06 '19 at 14:48
  • At least, if you are not interested in this topic, just admit that it can be of use for learning purposes. – Camion Aug 06 '19 at 14:51
  • 1
    This is not a discussion forum, it is a site reserved only for *specific questions* which can have specific answers. **You do not have an answerably specific question**, if you want to have an open ended discussion about various techniques, you need to take it to a discussion forum, which is not what an SE site is. – Chris Stratton Aug 06 '19 at 15:22
  • Designing an square amplifier for use in analog computing (which can be used for educational purposes) is a perfectly specific question. – Camion Aug 06 '19 at 15:34
  • 1
    It's an overly *broad* question without a specific answer. Nevermind that you haven't actually asked a question anywhere in your 'question'. – brhans Aug 06 '19 at 16:51
  • @brhans : What kind of precision would you want me to bring ? In what way analog computing is not specific enough ? – Camion Aug 06 '19 at 17:03
  • You could start by asking a question. Questions often start with words like 'how', where', why' - and end with a `?`. It would probably also help you to spend a little time reading some of the [asking help pages](https://electronics.stackexchange.com/help/asking). – brhans Aug 06 '19 at 17:08
  • Ok, I put the question more clearly in the title and reworded the text somehow. – Camion Aug 06 '19 at 17:26
  • This app note from Analog Devices may give you some ideas: [Analog Multipliers](https://www.analog.com/media/en/training-seminars/tutorials/MT-079.pdf) – Justin Aug 06 '19 at 18:02
  • Oh yes... I didn't think about the possibility to use log and antilog amplifiers – Camion Aug 06 '19 at 19:31

1 Answers1

3

This circuit will do squaring. I leave calibration, and temperature compensation, to you.

What is the bandwidth? at 1mA input, the 2N3904 will have about 40pF Cbe (computed from Ftau). With 100 ohm resistors, thus 4nanosecond Tau on nodes, BW is about 40MHz.

schematic

simulate this circuit – Schematic created using CircuitLab

analogsystemsrf
  • 33,703
  • 2
  • 18
  • 46