Filter 60 Hz component without altering signal of interest

[Khelz]

Hi Guys,

Here is the situation:

I am measuring the current flowing in a high voltage circuit (attached). Basically I am measuring the current through the shunt resistor which outputs a voltage proportional to the current
then I amplify the voltage difference with an in-amp and I acquire that on a scope. The problem is that the HV supply carries a large 60 Hz component noise and this gives a poor signal to noise ratio. Now the thing is that I cannot low pass or high pass filter the whole thing because the signal of interest has 2 components occurring at frequencies lower AND higher than 60 Hz.

So I thought that a solution would be to filter the High dc voltage BEFORE it reaches the shunt resistor, which could get rid of the 60 Hz component without filtering the signal at the same time. Since the current pulses come from the reactions at the load (chemical reactions) I thought that may make sense, but I am not sure if it possible to filter the voltage before it reaches the shunt resistor without affecting the whole circuit response.

What do you guys think?


Thanks a lot. View attachment Electric Circuit for support.pdf
Khelz

 

[crutschow]

Are you sure it's the HV supply and not pickup from the mains radiation? Certainly you can add additional filtering for the HV supply output, it that helps. If the load current is small you could use a simple RC filter but the capacitor would need a voltage rating over 3.5kV.

What is the value of the shunt resistor R1? Are the leads twisted-shielded from R1 to the amp input.

Another approach often used is a notch filter at the 60Hz line frequency. Those can provide significant rejection without significantly affecting signals on either side of that frequency.

 

[Khelz]

Are you sure it's the HV supply and not pickup from the mains radiation? Certainly you can add additional filtering for the HV supply output, it that helps. If the load current is small you could use a simple RC filter but the capacitor would need a voltage rating over 3.5kV.

What is the value of the shunt resistor R1? Are the leads twisted-shielded from R1 to the amp input.

Another approach often used is a notch filter at the 60Hz line frequency. Those can provide significant rejection without significantly affecting signals on either side of that frequency.

Hi Crutschow,

I tried your approach for the notch filter. I reproduced the exact circuit attached that has been suggested from an analog engineer.




The only difference with the circuit is the capacitor that is 0.22 instead of 0.2 and resistors that are 13.7 and 26,7 kohms. The amplifier I use is AD711, and I power with 2 nine volt batteries in series and I hook the reference to the ground of the circuit. Now I tested it with a waveform generator (output a sine wave across a resistor and put the positive lead to the circuit input and negative lead to the reference of the battery) and it is not working at all and I do not know why:

When I hook the reference (at 9 V) to the ground of the scope and generator it actually creates a more noisy waveform with the same frequency. When I do not hook the reference at all (so I supply straight 0 and 18 V to the op amp) the waveform is perfectly reproduced. However, it does not reduce the magnitude at all at 60 Hz.



Can you advise?

Khelz

 

[crutschow]

What do you mean "reference"? Is that the junction between the two 9V batteries in series? That junction is your circuit ground.

Do you have 0.1uF decoupling capacitors directly from the op amp power pins and ground?

 

[Khelz]

No I don't have any decoupling capacitors from the op amp power pins. The 0.1uf I have are connected as the circuit attached.

And you're right about the reference, I mean the junction between the 9 V batteries. Isn't it weird that the scope signal makes more sense when the reference is not connected??

 

[Venkadesh_M]

what about the high voltage ground and your low voltage circuit ground? If there is a proper isolation You need not to worry about the field effects.. If you have any field problems try the bottom side of the line to tape...

 

[crutschow]

I simulated your circuit and it indeed does not work as drawn. Just shows you can't trust everything you find on the Internet. ;-)

Below is the circuit with an added op amp in the feedback to adjust the circuit Q, which works. Unfortunately the notch frequency is quite sensitive to component values and the values you used moves it significantly away from 60Hz. If you can put two 0.1µF caps in parallel to make a 0.2µF capacitance for C1, the notch frequency is much closer to 60.

I plotted the response for various positions of the pot wiper (P). P=0 is for the wiper at the top of its rotation (which is basically equivalent to the original circuit). This is the top green trace and you can see that there is virtually no notch depth, which is what was observed with the original circuit.

As you adjust P, the sharpness of the rolloff increases (higher Q) but the depth of the notch decreases, so that's a tradeoff.

The values of the resistors and capacitors can be changed if you have other values that would give you closer to 60Hz. You just need to maintain R1 = R2, R3 = 0.5*R2, C2 = C3, C1 = 2*C2 and the time constant of R1*C1 = 5.31ms.

If you need a better notch performance there are other, more complex circuits that should do that, for example this.

 

[Khelz]

Crutschow thank you very much for your help. I will try these and let you know how it went. Venkadesh I am not sure I understand your suggestion. There is a proper electrical isolation since I use an isolation amplifier, otherwise I connect the grounds of both the high and low voltage circuit together in order to avoid ground loops.

- - - Updated - - -

One question: I found a few documents on the net that depending on the supply of the op amps being single or split, you should connect half of the supply or the ground to R2. How would that work in the circuit you suggested Crutchschow? Also, how practically would you do that? Would you use 2 batteries in series and connect the junction of the 2 op amps on R2?

thank you

- - - Updated - - -

I mean the junction of the batteries supplying the 2 op amps*

 

[crutschow]

If you want to use a single supply, (single battery) then you connect 1/2 the supply voltage to the points shown with ground symbols. This 1/2 supply can be generated by two 1k ohm resistors in series between the supply and ground with a 10uF and 0.1uF capacitor from the junction to ground. In that case you need to capacity couple the input and output to block the DC offset of 1/2 the supply voltage. Here's a little more info on that.

If you use two batteries, then you just connect the junction of the two batteries to the ground point and don't need the capacitor coupling. The ground point is the ground for the rest of the circuitry.

 

[crutschow]

Here's a different circuit design that I played with a number of years ago for a subwoofer room equalization circuit. It has variable notch frequency and Q. This allows for component tolerances so you can adjust the notch for exactly 60Hz with pot U4 as well as varying the width and depth of the notch with pot U3. The graph shows how the notch frequency varies with pot U4 wiper position.

If you want to play with these circuits on LTspice (free download), I can post the circuit files for you also.

 

[schmitt trigger]

Please Crutschow, do post the LTspice files.

Khelz..... Crutschow's circuit is an excellent topology that will give you excellent results. All I have to add is that you must employ stable components for the resistors/capacitors. This means metal film for the resistors, and plastic film for capacitors (preferably polystyrene).

 

[crutschow]

Here are the LTspice files for the two circuits I posted. I added the potentiometer files since those doesn't come with LTspice. Because of download restrictions I had to send them all with a .txt extenstion.

Rename the two Notch Filter files from .txt to .asc files.

Rename potasy.txt to pot.asy and potsub.txt to pot.sub.

Paste pot.asy into the LTspice SYM folder.

Paste pot.sub into the LTspice SUB folder.

View attachment Notch Filter .txt
View attachment Notch Filter Adj 60Hz .txt
View attachment potasy.txt
View attachment potsub.txt

 

[schmitt trigger]

Did as you suggested, works excellent, thanks!

 

[crutschow]

I found a different circuit reference and it has an added pot to optimize the depth of the notch (U5 in the schematic below). There is some interaction between U4 and U5 so you need to tweak them both to get the best rejection.