I need to measure the peak value of the mains voltage, and I'm gonna use a peak detector. A simple diode, plus an RC in parallel with a suitable time constant. But there seems to be a problem: voltage surges, a high voltage pulse that can charge the cap to an incorrect peak and make my peak detector useless.
My question is: how much are these surges important, and is it possible to filter these surges effectively?

 

Use a zero-crossing detector, followed by a monostable, adjust it so you read at the peak part of the mains cycle, then average over a number of readings.

__________________
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk

 

That seems a reasonable approach.


I guess that depends on what you are trying to measure and why.

They could be very important, it depends on the equipment connected to the mains supply.

Yes.

Why are you measuring the mains voltage, just to monitor it to see how it varies with time and load?
Or are you looking for a problem?

There are companies (Dranetz amongst others) which produce items of test equipment to measure and record the spikes, dips and surges on mains supplies.

You could use a mains supply filter to remove the spikes before applying it to your peak detector, or you could increase the time constant of the RC filter after the detector.

JimB

__________________
Experience is directly proportional to the value of the equipment ruined.

 

I have to design a tap changer for a three-pahse transformer which has -10%,-5%,...,+15% taps in its primary (6 taps).
The tap changer must monitor mains voltage trends and choose a suitable tap to compensate mains amplitude increase/decrease in its primary. proper tap is chosen with back to back connected thyristors.
The approach I chose: use a peak detector, then a differential amp., and compare the output DC voltage with 6 references, then decode the result and choose the proper tap. Surge problem can make this circuit choose the improper tap.

 

That is why on a tapchanger hysteresis is applied to avoid unnecessary hunting up and down of the tapchanger.

In substations the TCOL (Tap Changer On Load) senses the voltage via a VT which is fed from the secondary of the transformer, and via the AVR it will get the instruction weather to step up a tap(s) or step down a tap or taps.
Usually a delay of 45 seconds is applied to avoid hunting of the tapchanger, it will cancel out brief dips and spikes and also will allowe for starting of heavy motors without unnecessary changing taps.

In your tapchanger you will have to allowe for a resistor adequately rated to be switched in circuit with one tap position on before parallelling the windings otherwise a shorted turn will quickly burn out your transformer.

Example:
Tap 4 active.
Raise tap (command from AVR)
Tap 5 and series Resistor in parrallel with tap 4
Break Tap 4.
Close Parrallel switch across Resistor.
Tap 5 active.

The resistor has to be able to handle the full load current for a couple of seconds during the tapchange.

__________________
There are more ways to get to Rome.

Electricity, Electric clocks, Meters and Trains are great.

 

Yes in real life transformer tap changer but not necessarily in his case. You have missed the fact that he uses thyristors to do the tap switching.

A clever design is to wait for current zero and then switches on the thyrsitors pair of the next tap.

__________________
L.Chung

 

You both are right. I have taken into account the shortening turns problem. An auxiliary circuit detects zero crossings and does not let tap changing to happen unless the current in the corresponding phase is zero.
Thank You

 

To get the system voltage, why would you need to use a peak voltage detector?

A three phase rectifier to a large filtering capacitor would be sufficient for your circuit. You will need to program a very long time constant(in ten of seconds) for the measured system voltage value to stay outside the present tap position value before you even commands the tap changer to initiate a change of tap so any surge on the line would be averaged out by the filter capacitor and not be a problem.

__________________
L.Chung