[SOLVED] AC voltage measurement without transformer issue

Hi All,

I am planning to measure the AC RMS using microcontroller. Because of size constrain I don't want to use step down transformer. By using bridge rectifier the input AC voltage is converted into DC, then using voltage divider network the rectifier volt is reduced and out put of the voltage divider is filtered and given to microcontroller ADC.

Issue:

After the bridge rectifier I am not getting the proper rectified output.
The actual output is attached below.




Rectifier part number: MB10s
Voltage divider values : High side 110K 4 numbers. low side 3.3K
Capacitor : 1UF/63 volts

Block diagram:

Hi,

show the complete schematic. With all connections and values.

(my first assumption: the scope connection generates the proboems.)

(BTW: It seems you are NOT measuring RMS voltage. You more measure peak_volatge or average_voltage - depending on filter circuit. Only for a clean sine waveform you may use a fixed factor to come to the same value as a true_RMS measurement, but with any other waveform you don´t get the exact value)

Klaus

Thanks for your reply,
Here is my schematic .

Actually I am measuring the peak voltage from the peak planned to get the RMS.
Example :

Input 230 Volt RMS
Expected Vmax = Vrms*1.414
=230*1.414
=325
From this Vmax I planned to get the RMS
Vrms=Vmax*0.7071
=325*0.7071
=229

Comments:

The time constant of 3.3K across 0.22uF is too short, you will get quite a lot of ripple. I would suggest you use a higher voltage capacitor before the voltage divider (after the bridge rectifier) maybe with a low value resistor (~1K) between the bridge and cap to limit surge current. That cures the time constant problem without making the circuit complicated.

There is a safety issue, you should have a fuse or fusible resistor in series with the incoming AC in case the rectifer or capacitor fails and you absolutely must ensure the "PGND" point is insulated and isolated from any real ground connection. PGND carries dangerous voltage in that circuit at all times.

The oscilloscope ground must not be connected. To look at the voltages/waveforms you must use suitably rated probes and take a differential mesaurment. Set the scope to invert one channel (not both!) and add them together. That will give you the same measurement as a "ground and tip" probe but without any real ground connection.

Brian.

This could be slightly dangerous if you lose your power ground, but you already know that.

My approach would be to connect a suitable parallel RC time constant network directly across the bridge, as already suggested by Brian.

Then use a high voltage differential amplifier to scale the output down, and ground reference the output. Some thought needs to go into providing sufficient common mode range in both directions, but its doable.

Hi,

I connected 2UF/ 450 Volt across the bridge rectifier and I measured the Vmax, but I am not getting the expected result the readings noted below.



Please give some suggestion

It would be nice if you bypass the two resistors (voltage dividers) with two small (say 10 nF) capacitors.
How about using two capacitors in series (both on the line and neutral side) of the input?

Hi,

I connected 2UF/ 450 Volt across the bridge rectifier and I measured the Vmax, but I am not getting the expected result the readings noted below.

Click to expand...

If you changed your circuit, then please show us the complete circuit.

How did you measure V_Max?

Klaus

Here is my updated circuit, I am measuring the 2.2uF Capacitor voltage( Vmax ) using multi-meter. Please correct me If Iam measure wrong.

I am considering the capacitor volt is Vmax, from this Vmax calculating Vrms. by using the following formula Vrms=Vmax*0.707

satiz said:

I connected 2UF/ 450 Volt across the bridge rectifier and I measured the Vmax, but I am not getting the expected result the readings noted below.

Click to expand...

The peak voltage is reduced because of the load. The expected voltage will be seen only when the capacitance is very high (it does not discharge appreciably in one period). You cannot expect theoretical value with this values of R and C.

hello,

it is very dangerous to connect one phase (even the neutral ,wich could be sometimes not so neutre!) to gnd
and if you plug the AC input in reverse way..=> phase to Gnd !

you find less than theorical value because
- loosing drop voltage acrross diodes
- impedance accross the ADC input acrros R17
_ form factor sinus disttortion..

you can use your voltage divider on the 2 diode bridge branch .. and maybe connect the Earth to ground
The very low current leakage to earth will not affect differential breacker 30mA! on main AC power supply.

paulfjujo said:

hello,

it is very dangerous to connect one phase (even the neutral ,wich could be sometimes not so neutre!) to gnd
and if you plug the AC input in reverse way..=> phase to Gnd !

you find less than theorical value because
- loosing drop voltage acrross diodes
- impedance accross the ADC input acrros R17
_ form factor sinus disttortion..

you can use your voltage divider on the 2 diode bridge branch .. and maybe connect the Earth to ground
The very low current leakage to earth will not affect differential breacker 30mA! on main AC power supply.

View attachment 131321

Click to expand...

Thanks for your reply..
Could you tell me how to use two diode bridge here ?

Please provide one simple solution to convert RMS into DC with 1 to 2% output tolerance.....

paulfjujo said:

- impedance accross the ADC input acrros R17
_ form factor sinus disttortion..

Click to expand...

If the RC time constant is much larger than the AC period, the capacitor charges to the peak voltage in a form-factor independent way. If the time constant is comparable to the AC period (or less) the capacitor discharges in a manner that depends on the shape of the waveform. It can become computationally intractable.

Hi,

measuring peak voltage is the worst i can think of.

It just uses one single point (of time) to estimate the voltage.
* It´s easy for a spike (caused by a motor, a triac, an SCR...) to add a little peak to the top of the sine waveform. --> resulting in a higher output voltage than expected..
* but usually mains voltage sine has a flat top caused by switch mode supplies (where most of the input current is drawn by the very peak of the sine voltage...not all SMPS are that bad..)
therefore you see high current around the peak of the sine, but low current at the rest of the sine. This current cuases a voltage drop in the wires.. and the voltage drop causes a flattened top of the sinewave. --> thiss effect results in a lower output value than expected.

Therefore - if you are interested in some reliable voltage measurement - you should use other measurement methods.

Either a true-RMS, where you take a lot of samples of a sine fullwave do the squaring, averaging and take the sqare-root from it. Best results, but with some software effort.

Or do a true rectified_average measurement, where the complete waveform is taken into account. --> DC output, easy to sample with an ADC.

Klaus

Hi

Thanks for your reply.

Please suggest some example circuit for true rectified_average measurement.

Hi,

@paulfjujo:

I agree that your solution brings some safety, but it generates measurement problems.

In the LTspice simulation - red line - you see what happens.
The problem is that often "neutral" mains is connected to Earth-Ground. If you now additionally connect your measurement-Ground to earth ground (like shwon in your schematic), then the result is a half wave recified and less attenuated output.



Klaus

- - - Updated - - -

Hi,

here a circuit:

It has
* a full bridge rectifier (loosing about 1V RMS)
* an attenuation circuit 100k:1k
* and a two stage ripple filter.

example: 200V RMS input
after rectifier: 200V RMS - 1V RMS = 199V RMS
after attenuation: 199V RMS / 101 = 1.9703 V RMS
After filtering (averaging) 1.9703V RMS * 2 * sqrt(2) / Pi = 1.774V (rectified averaged) expected value.

The simulation shows 1.771V DC (=0.2% too low), with a ripple of about 0.5mVpp. Tau is about 270ms.

--> take just one ADC value and calculate: U_RMS = ADC_value_in_V * 101 * Pi / (2 * sqrt(2)) +1 = ADC_value_in_V * 112.2 + 1

**
For sure the safety probelm is not solved.
And the result may be affected by stray impedance of maes_gnd to earth_gnd.
--> You need proper isolation with low leakage currents

Klaus

- - - Updated - - -

Hi again.

Update: improved version: Less part count, higher input impedance, lower output impedance, same performance.

Btw: in all my circuits R4 is just for simulation. You don´t need it in reality.

Klaus

Hi ,

I got the result with 2 Volt difference. I attached the schematic below for your reference.(In real time test the output is 2 Volt difference)



Formula used to calculate the Rms is Vrms=((R12+R8+R9+R10+R11)/R12*AC sense volt )+1.4

Please correct me If I am doing wrong calculation

Hi,

this is very close the the true_rectified average.

I expect the ripple to be worse than mine: about 11mV at the ADC
and a tau of about 0.3 s

*****************

Formula used to calculate the Rms is Vrms=((R12+R8+R9+R10+R11)/R12*AC sense volt )+1.4

Click to expand...

with a clean sine:
V_RMS = U_ADC_in_V * Pi * (R8 + R9 + R10 + R11 + R12) / (R12 * 2 * sqrt(2))
Add 1 VRMS as voltage drop at the rectifier.

Klaus