Measuring AC Power Line

[sougata]

Hi there,

My problem is I need to make a circuit that senses the AC RMS voltage and takes certain decisions. Now Interfacing to the power line my way are:

1. Use a 20:1 transformer , make a potential divider, on the dc side to account for the diode loss, and do the adc. Tried - results non-linear ,unsatisfactory for the input range.

2. Use an IL300 or similar circuit to make a linear optoisolator - Increases overall price.

3. I am wondering can I use a PIC12F675 to sense the input AC and PWM an I/O to an opto-coupler or shift- out the values

Can you shed some light on the third option.

By the way my circuit would be used on a three phase line-to-line sensing scheme.

Regards

Sougata

[Luciano]

Hi!

Voltage-to-voltage signal conditioner:
http://www.ayainstruments.com/products3.html


Other solution:

The link below is the schematic of an uninterrupted power supply.
See the signal IN-RECT on the schematic page 1 (Top left corner).
See the same signal IN-RECT on the schematic page 2. (Top left corner).

http://www.eserviceinfo.com/downloadsm/2270/APC_250i_400i_600i.html

Be careful with AC Mains voltage!

If you try it, let us know how it works.

Luciano

[sougata]

Hi Luciano,

Thanks for the post. I live in Kolkata India and would not be able to source the Aya products. Other design constraint are space and cost. Please let me know if you have any ideas. Whatever it is if I get through the problem rest assured that I will post the schematic and code.

Regards

Sougata

[sougata]

Hi everybody,

I am working on a routine with 12F675. Since the frequency of the ac line is quite predictable I am sensing the zero cross then starting the sample at almost the peak of the AC waveform. As you know that the average dc voltage is 0.707 times the ac RMS, so it is quite easy to sense it as long as it is sine wave. I am then shifting it out via an opto-coupler. By the way I am using a transformer less power supply (read non-isolated) for the 8 pin PIC.
Microchip has got an appnote on the same (PS).

Regards

Sougata

[jmgelba]

Are you trying to measure just one phase, or all phases, or the difference between one phase and another? Are you measuring power, volts, amps?
If you give me size and power constraints, I can probably come up with something since I used to design switch mode power supplies :-)

[sougata]

Hi jmgelba,

Thank you for your post. My application is basically a voltage monitor (via RS232) and trip circuit with presettable hysterisis and auto recovery. I would be measuring between R-Y and Y-B. It is the output of a DG and a three phase three wire system. Any idea or suggestions would be greatly appreciated. Thanks in advance.

Best Regards
Sougata

[jmgelba]

I guess the cheapest method would be to use an auto transformer. You could loop a piece of wire around the wire you wish to measure. You could calibrate the reading value as a percentage of the voltage in the wire you are measuring. Obviously each wire loop would be slightly different. You can account for this in software.
Take the two readings from each loop put them through the adc and measure the difference, or do whatever you need to do with the reading. Its cheap but its not elegant.

[sougata]

Dear Jmgelba,

As for the auto-transformer based scheme could you explain the construction process or help with schematics

Regards

Sougata

[psdayama]

Hi Saugata,
sorry I hadn't seen this post.
I must tell U that U need to brush up Ur measurement technics although
U are good on programming PIC.
a) Diode rectified output voltage of attenuated mains will be highly
non-linear as U are trying to convert about 1-2 volts signal and diode drop
is typical 0.6 Volts. Simple remedy is :- use diode directly on mains and then
attenuate the DC voltage. This will give somewhat better accuracy.
b) better use a small OPAMP based precision detector. This will give U best
accuracy at very little extra cost.
c) Avoid transformerless power supply. This will help in measuring multiple
inputs. Also instead of measuring RY-YB-BR fashion measure the voltages
at star point (neutral) and corrolate to line voltage. This way U can arrange
voltage divider network in star fashion and measure in cyclic way.
d) U can use Exxar or NS made voltage to frequency converters and transmit the results by optocoupler to RS232. Those have about 1-2% accuracy which is good enough. If U can't get them then send me private
email and I will show U how to get result from even 555 !

regards.
dayama

[sougata]

Dear Dayama,

This is an old thread and I have already designed and delivered the product. It has been tested by ERTL Kolkata.It is due for testing at NABL Hyderabad and has been approved by GE Bangalore.

1. I am aware of the diode loss and when use a transformer for isolation, I make sure of the following.
a. The transformer never saturates.
b. The sensing winding is sufficiently higher (at least 10x) than the measuring span.
c. Use Schottky diode (0.135 volts Fwd) or active rectifier (OP-Amp based)
d. Use a slope routine to nullify any non-linearity.
e. Do averaging of ADC routines.

Cost constraints do not allow me to use a rms converter chip from LT or AD.

For my typical measuring requirement it is a naval ship and you don't have a neutral there!!!!!!!

When cost permits I use the IL300 linear opto from Vishay or Siemens. I have my owm PIC (12Fxxx) that works as a voltage to serial converter (opto-interface) powered by non-dissipative capacitive power supply.

Low cost is important but I need to optimize my design around the thin line.

Regards

Sougata