Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

Op amp with 120dB PSRR exhibits weird "noise" at 60dB gain when powered by traf

I have built the circuit in the schematic bellow which seems to work acceptable when is powered by a 9V battery.
The circuit measures the current in the mains by the means of an very small resistive shunt (R20 10mOhms 20W) connected in series with the load. This will produce a small voltage signal proportional with the current in the load, that needs to be amplified. To do so I use a very performant op amp AD8551, used as a non-inverting amplifier that is controlled by a microcontroller to have 4 different amplification levels (2x, 10x, 100x and 1000x).

When the circuit is power by BAT1 (a 9V battery) and there is no current in the sensor, the output is relatively flat, which is ok. There is a little bit of 50Hz noise picked up from the mains that is showing up in the 1000x amplification level, but that is manageable.



But when the circuit is powered by the transformer as in the schematic bellow, the 1000x amplification level is rendered useless due to a very weird constant "noise".



The noise is not induced by the transformer, because if I power the circuit with the battery and keep operating the transformer near the circuit, the output is the same as in the first picture, without the transformer.
I have also tried powering the Tl431 from the battery while the rest is powered by the transformer, to exclude the possibility of noise induced by the 2.45V shunt regulator to the non-inverting input. But nothing changed.

What could cause that noise to show up only when powered by the transformer?
I tried everything and can't link it to something.

Here is the schematic:

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

Is the short between transformer primary and secondary real? There may be ground loops involving the oscillsocope. Generally one would prefer a 4-wire shint measurement.

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

Yes, the circuit ground is connected to neutral. This is done so that the voltage on R20 can be measured.
There is no oscilloscope. The plot is obtained from the data saved on a SD memory card by the microcontroller from ADC.
The uC saves 128 samples at 6400 S/sec (128*50), switches amplification level, waits a few seconds and samples again.

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

PSSR is not perfect over the entire input range.

THe datasheet for 5V single supply says ....

Power Supply Rejection Ratio PSRR VS = 2.7 V to 5.5 V 120dB min 130 dB typ.

That means PSSR is poor below 2.7V input, so you may have to condition your input range or pick a different OA.

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

My Vs is 5V. Not sure what are you suggesting. Limit non-inverting input from 2.7V to 5V ?

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

Yes, the circuit ground is connected to neutral. This is done so that the voltage on R20 can be measured.
There is no oscilloscope. The plot is obtained from the data saved on a SD memory card by the microcontroller from ADC.
The uC saves 128 samples at 6400 S/sec (128*50), switches amplification level, waits a few seconds and samples again.

Click to expand...

Slowly approaching a complete problem description...

"No oscilloscope" also means there's no second ground connection besides neutral? Was the neutral line also connected in battery supplied measurement?

What's the peak-to-peak noise at amplifier output, any low-pass filtering between amplifier and ADC involved?

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

Yes, there is no second ground connection besides neutral, so no ground loops.
Yes, during the battery test, the neutral was still connected.

Actually, the op amp output is fed to an 5KHz PWM, then transferred through an optocoupler, then on the other side filtered through a 200Hz low pass, and then fed to ADC.

The peak-to-peak of the op amp output noise at 1000x gain is about 1.25V. The full range on the plot in the picture is from 0.5V to 4.5V.

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

expert_vision said:

My Vs is 5V. Not sure what are you suggesting. Limit non-inverting input from 2.7V to 5V ?

Click to expand...

Yes that is what I meant.

That means Vref changes from 2.5V to 4V with +/- 1.3V swing input, attenuate input and increase gain on output. ( or possible 4.1V ref for 1.4V swing as Vmax in is 5.5 for PSSR for V+=5.0)

OR change to slightly better OA rated for 1.8V instead of 2.5V

-PSSR is only 80dB with 50mV input at Vref=2.5 on V+=5 but you have 60dB gain.


OR improve your AC ripple on the V regulator.


maybe AD8515: 1.8 V LOW POWER CMOS RRIO OPERATIONAL AMPLIFIER is worse.
-worse PSSR in nominal range but wider range.


Test PSSR vs Vcm in vs T('C) to validate assumptions on PSSR.

Rail-Rail in but not perfect PSSR.

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

Oh. Does "Vs" in datasheet refers to "Voltage signal" or input swing? I thought it refers to "Voltage supply".

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

According to datasheet values, insufficient PSRR can't explain the output ripple, even if the 7805 input voltage drops below minimal voltage. There must be > 1 mVpp OP input voltage at 100 Hz or 10V supply ripple.

A possible explanation would be a bad circuit layout where capacitor ripple current couples into the amplifier input. The amplifier ripple is however large enough to see it with a sensitive oscilloscope input and 1:1 probe and trace it's origin.

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

I just replaced the 220uF cap with a 470uF low ESR cap and the output waveform didn't change at all. I guess it's not caused by supply voltage ripple.
I wish I had access to an oscilloscope ...

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

Can you clarify the test results? Input being tested, Measured Results. Vmin,max

Re-check assumptions;
CMMR issues? Ground shift issues? Twisted pair or not?

Can you make the noise worse by moving sense cable?
Using HE/CT? is it saturating?

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

In the schematics, I replaced C2 220uF with a 470uF low ESR cap which should reduce the supply voltage ripple and should reduce the noise observed at 1000x gain when powered by transformer. But after I replaced the cap I could still observe the noise at 1000x gain with exactly the same amplitude, which leads me to believe there is no problem with the voltage regulator.

CMRR issues: Shouldn't be the case. The OA is used as a non-inverting amplifier not as a differential amplifier. Given the configuration I use I don't see how I could have a common signal at both inputs.
Ground shifting: Very possible, but I have no idea how to test it or the instrumentation to measure it.
Screening: Not really necessary. The sensor has such a small impedance (10 mOhms) the EMI would have to be insane to create a differential potential on such a small impedance. In fact I placed the wire from the sensor (1 mm copper solid wire) right on top of the transformer, and the output noise barely changed (maybe 10 mV measured from the plot).

I cannot make the noise worst, but I can get rid of it either.
I didn't use an HE or a CT sensor because I wanted a wide range sensing, didn't really had any available and wanted to keep thinks simple. Now it's kinda of a bugger to switch since I've already built the thing.

The best candidate in my guess would some crazy ground currents. On a second thought, I can actually measure that.

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

I think the noise is produced by 78L05 voltage regulator.

Re: Op amp with 120dB PSSR exhibits weird "noise" at 60dB gain when powered by traf

Ugh. I give up. I tried the biggest caps I had laying around to filter that Vcc, but no dice.
But, I replaced the 9V transformer with a tiny old 9V phone adapter and noise is .. puff .. gone. :grin:
So I'll just use the tiny adapter and save my self from the trouble. I'll still use the 5V regulator but since the 9V source is so much cleaner, I guess the output is cleaner too. Hard to tell with a lousy DVM.

I probably should have included a voltage stabilizer or something in the design for such a sensitive circuit. I certainly learned my lesson.