Problem with an opamp circuit picking up interference

[Ubergeek63]

I've attached a schematic of my circuit.

as pdf or graphic? i, for one, despise all things microsoft

 

[ljbdragon]

Ok, my plan is to redesign my circuit, but this time to use INA126 instrumentation op amps. My ultimate goal is to create a circuit that will take the input from a current transformer (5 to 500mV) and convert it into a 0 to 10 volt dc output without being influence by external electrical noise, having said that I do expect to see dips and rises as other loads are being switched on and off, what I'm trying to avoid are the spikes created when inducted loads are switched elsewhere on the mains supply, and as I've discovered more recently the noise that a Ethernet over mains adaptor can cause.

I'm sure this will lead to new and interesting problems.

 

[Roff]

Post a schematic before you breadboard. We can probably suggest improvements (or corrections).
You only need one inamp per sensor. Do the rectifier with an op amp, as before.

 

[ljbdragon]

I've attached a drawing showing how I intend to use the INA126 instrumentation amp, I only have the datasheet to go on, so I would appreciate some advice on how to improve my circuit.

 

[Roff]

Is 820 ohms the recommended burden resistor for your CT?
Your IA needs a path to ground for input bias current (and common mode voltage reference). Either split the burden resistor into two resistors of equal value (i.e. 410Ω each if you are going to stick with 820Ω), then ground the center tap. Alternately, add 1Megohm to ground from either pin 2 or pin 3.
You have the gain set at 10. You can add your original precision half wave rectifier at the output.

 

[ljbdragon]

Yes, the 820Ω resister is the burden resister for the CT, although the CT may change so I'm going with your second suggestion of a 1MΩ to ground from either pin 2 or 3. I'm planning to put it on to a breadboard at the weekend and see how it performs.
Thanks for the assistance so far Roff, much appreciated.

 

[rumpfy]

Hi ljb,
Looks like yr getting the good oil so stick with Roff. I note you say you have the data sheet for the INA126. I find with this stuff that's all one needs.
I downloaded the ti data sheet for the INA126, and you will get lots of stuff by reading that. On p8, fig.3, there are 3 'typical arrangements'. Yours is the 'microphone, hydrophone etc'. and note the resistors to ground. To achieve a balanced input, we are talking about 'the same impedance from each terminal to ground'. To go the way Roff suggests is ok, but you need to consider that the impedance to ground is not well defined. It consists of stray capacitance and the high impedance of the input circuitry. In your case, I would go the two resistors of 410 ohm to ground method. P8 also describes the bias requirement for the device. Since the CT is operating at mains frequency, you can use some capacitance to ground from each input to attenuate the higher frequencies which will be mostly interference(probably). Your sketch doesn't show any filtering. On p4 of the INA datasheet, bottom left, is a diagram of common mode input voltage against output swing and the operating region shown is the region in which the common mode performance will be satisfactory. However, the degree of balance of the input wiring and circuit will probably dominate the common mode performance.

Your arrangement only needs to be satisfactory, but the fine tuning can make a huge improvement in the end result.

 

[WTP Pepper]

Is it possible to create a reliable instrumentation amp from ordinary amps like 741 or LM348, or is there a particular chip you would recommend.

Yes if you design your circuit properly. Instumentation amplifers generally have very low bandwidth.

 

[Roff]

Hi ljb,
Looks like yr getting the good oil so stick with Roff. I note you say you have the data sheet for the INA126. I find with this stuff that's all one needs.
I downloaded the ti data sheet for the INA126, and you will get lots of stuff by reading that. On p8, fig.3, there are 3 'typical arrangements'. Yours is the 'microphone, hydrophone etc'. and note the resistors to ground. To achieve a balanced input, we are talking about 'the same impedance from each terminal to ground'. To go the way Roff suggests is ok, but you need to consider that the impedance to ground is not well defined. It consists of stray capacitance and the high impedance of the input circuitry. In your case, I would go the two resistors of 410 ohm to ground method. P8 also describes the bias requirement for the device. Since the CT is operating at mains frequency, you can use some capacitance to ground from each input to attenuate the higher frequencies which will be mostly interference(probably). Your sketch doesn't show any filtering. On p4 of the INA datasheet, bottom left, is a diagram of common mode input voltage against output swing and the operating region shown is the region in which the common mode performance will be satisfactory. However, the degree of balance of the input wiring and circuit will probably dominate the common mode performance.

Your arrangement only needs to be satisfactory, but the fine tuning can make a huge improvement in the end result.

I agree that splitting the burden resistor should result in less common mode noise reaching the output.

 

[lebevti]

take a grounding and shielding course

 

[ljbdragon]

Tried out the INA126 amp today and the results were very good, I was also able to accurately workout the current flow through the CT based on the output of the amp. Most of the interference has gone with the exception of the occasional spike on the output when inductive loads elsewhere on the mains supply were switch on or off, at first I thought this was via the CT however they continued with the CT disconnected, as far as I can tell this interference is getting in via the 0Volt rail of the PSU.
I did go with splitting the burden resister into two separate resisters in the end.

The only problem I did have was with the half wave precision rectifier, I used the one from my original circuit and found that the ratio between output voltage and actual current kept changing, I'm thinking this may be due to the way the diodes behave at different voltage levels, the rectifier input voltage varies between 200mV and 3 volts depending on current sensed, the only diodes I had to hand were IN4001.

All in all a good result, thanks to Roff and Rumpfy for pointing me in the right direction.

 

[rumpfy]

Hi lb,
That's good.
You seem to have a good grasp of the subject. I didnt look at your circuit arrangement before writing this, But I note you remark about the rectifier. From memory, you have the diode as a series diode at the output of one of the op amps. I thought at the time the diode would have been better to be inside the feedback loop of the opamp. I have an old National semiconductor 'Linear Applications handbook' and this has a wealth of quick reference ideas. if you need some ideas on precision rectifiers please advise.
With power rail filtering, remember that electrolytic caps are a bit suspect for HF filtering. Try some small ceramic caps in parallel with them.

 

[Roff]

In your original circuit, U3, D1, D3, R7, R9, and C1 form a precision half wave rectifier.
I would use a JFET-input op amp here, such as TL072, and change the resistors to R7=100k and R9=270k. Make C1=1uF.
If you are digitizing the half wave signal, then you should be good to go. If you want peak rectification, you will need a cap (1uF?) to ground from the cathode of D1. The loop may oscillate unless you add a small (100Ω) resistor in series with the cap. The tradeoff here as you select the cap value is between ripple, and response time to changes in current.

 

[ljbdragon]

Thanks guys, I've ordered some extra caps for the PSU, and a TL072. With luck I should be able to try your suggestions in a couple of days. My ultimate goal is to plug the DC output into a data loggers analogue input, so I'm thinking response and ripple are equally important to get good quality data in return.

 

[atferrari]

Not clunky .doc files

as pdf or graphic? i, for one, despise all things microsoft

Hola dragon,

You were suggested to use graphic files. Better for everyone than .doc.

Be a good boy and learn how to post graphics.

 

[ljbdragon]

Hola dragon,

You were suggested to use graphic files. Better for everyone than .doc.

Be a good boy and learn how to post graphics.

Hello atferrari,

Didn't realise that word documents would cause a problem.

 

[Roff]

Thanks guys, I've ordered some extra caps for the PSU, and a TL072. With luck I should be able to try your suggestions in a couple of days. My ultimate goal is to plug the DC output into a data loggers analogue input, so I'm thinking response and ripple are equally important to get good quality data in return.

I can think of a way to get both at the same time. It would require adding a lot more parts, and I would have to do some circuit design.

 

[Roff]

Hello atferrari,

Didn't realise that word documents would cause a problem.

PNG files make cleaner schematics, and the files are smaller.