1
\$\begingroup\$

I am new to your forum although I often read it and find great help.

I have a problem that I cannot resolve. I am using a 5A CT (1000:1) to measure 110VAC (50Hz) currents of 70 to 90mA which gives me a very small window (200mV in a 5V range) and I need to monitor this little window with a 10 bit ADC with some degree of accuracy and reliability. After many days, I ended up using a MAX368 audio amp and half wave rectifier to deliver the best DC result.

What I would like is to expand my small 200mV window to 2000mV biased at about 2.5V in order to be able to digitally track and react when the window is breached.

enter image description here

Please assist by suggesting aa additional DC amplifier that will expand my DC window./


Thank you for the responses.

Apologies, I meant 10 bit ADC! RMS, ave or peak, no matter. It's a DC output to an ADC.

Burden resistor has been optimized for best results.

40% is arbitrary! Gives me some lee-way for excess breeches of the window.

I want to expand my DC window to achieve the above "arbitrary" value.

\$\endgroup\$
  • 1
    \$\begingroup\$ A 1024 bit ADC? I'm not sure that's what you mean! \$\endgroup\$ – Finbarr Sep 29 '17 at 17:18
  • \$\begingroup\$ Essential to know what you want to measure avg, rms or pk with value for 2v and 4 V implies an offset with a gain of 10 using only 40% of a 10 bit ADC range. Is that optimal or arbitrary? Yes a gain , 2.5V Vref and window comparator is easy once you clarify above. \$\endgroup\$ – Sunnyskyguy EE75 Sep 29 '17 at 17:19
  • \$\begingroup\$ Audio amp directly connected to an audio opamp and outputting via a diode to a 1024bit ADC? I don't even know where to begin... \$\endgroup\$ – PlasmaHH Sep 29 '17 at 17:26
  • \$\begingroup\$ Also you must define voltage input waveforms with R burden, not current. That means crest factor & bandwidth if not pure sine. \$\endgroup\$ – Sunnyskyguy EE75 Sep 29 '17 at 17:27
  • \$\begingroup\$ A one input terminal op amp... Haven't seen that one before \$\endgroup\$ – Voltage Spike Sep 29 '17 at 19:29
1
\$\begingroup\$

No burden R shown, so CT voltage could be dangerously high if it's not included in your CT. If the current is non-sinusoidal, then you'll probably need to digitize at least 500-1000 times a second to get a reasonable result. I assume by a 1024 bit ADC you really mean an 8 bit ADC. Using a peak detector is ok for sinusoidal current flows but the losses in doing your rectification the way you have shown will kill your accuracy (Vf for the diode) and is only valid for the positive peak.

I'd suggest you read this.

Look at the full wave rectifier schematic and details starting at Figure 17

enter image description here enter image description here

As you can see you don't actually need a peak detector once you have a full wave rectifier. You can digitize the result with any reasonable MCU that includes an ADC with your required resolution.

\$\endgroup\$
0
\$\begingroup\$

If you can sample the ADC at accurate time intervals then use an AC-coupled amplifier with an output centered at midrange for your ADC. Then sample at a frequency that works and take the RMS of the samples.

There's a lot of sample frequencies that you can use. The one that's conceptually easiest is 200Hz. That's exactly four times your line frequency, so you can take the four samples, subtract the average, and compute the RMS.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy