analog output 0 ... 15mv with noise

[ixps]

Hello,

I'm building an circuit to generate an analog output with range 0 .. 15mV however i have getting a noise on my output when i measure (with instrument to read load cells, with range 0..50mV and 15bit). I can calibrate the output, and works fine, but sometimes the noise  is too much.

On oscilloscope, I cant measure "ripples or noises" because its very small.

Follow the schematic:



My 3.3V is used as reference in my DAC (and i use an voltage reference to generate it). I would like to supply the OP07 with +-5V, however my charge pump only generate -1.4V on negative supply, but, i think this doenst it the problem.

I attempted to put an low pass filter on AO+ and AO- with 159Hz and 0,15Hz, but the problem persist.

Ps: my resistors does not  precision and stable (are standards, with tol of 5%)

Someone has a suggest?   


thnx

[David Hess]

I have designed differential amplifiers which directly drove a simulated load cell output before and never had a problem with noise, but the output was differential and I used a precision operational amplifier in the class of an LT1001.  I think it was the LT1112 but there are better choices unless you need such a low input bias current.

In another design I started with a higher voltage, and then used a resistive attenuator to reach the levels that a load cell would produce.

However you definitely should be using 1% metal film resistors or better.

[ixps]

Hello David,

Thank you for your suggest. I never worked with output using differential op amps. Is it like this?

https://ars.els-cdn.com/content/image/3-s2.0-B9781856175050000119-gr5.jpg

For example, I need a reference voltage (precision) in one of inputs, and my DAC output in another and the output will be diference between they?

Thanks again.

[David Hess]

Thank you for your suggest. I never worked with output using differential op amps. Is it like this?

No, like the left side of the three operational amplifier instrumentation amplifier:

https://en.wikipedia.org/wiki/Instrumentation_amplifier

For example, I need a reference voltage (precision) in one of inputs, and my DAC output in another and the output will be diference between they?

I do not understand what you are trying to accomplish.  Load cell outputs are differential even if they only use a half bridge or one active element.  However the outputs do *not* need to be symmetrical as long as their output resistance is identical.  So for instance you could use a precision reference, probably the same as the DAC reference, divide it in half for one of the outputs, and then use a voltage divider between that output and the DAC output to make the other output.  I remember seeing a load cell simulator which did exactly this although the reference used was the excitation voltage.

[jmelson]

Large capacitors to ground on the output of opamps can make them unstable.  You could put a small capacitor across R5 and let the op-amp do the work of smoothing the output.  Possibly as little as 100 pF could solve it.

If you MUST put a large cap across the output, place a 100 Ohm resistor at the output pin of the op-amp.  if you connect the feedback resistor to the far side of the 100 Ohm resistor, then the feedback loop will compensate for any DC error introduced by the series resistor.

Jon

[mikerj]

however my charge pump only generate -1.4V on negative supply, but, i think this doenst it the problem.

Charge pump style supplies can be extremely noisy.  Is you circuit on a PCB with good grounding, or on a breadboard?

[ixps]

Hello David,

Thank you for your suggest. I never worked with output using differential op amps. Is it like this?

https://ars.els-cdn.com/content/image/3-s2.0-B9781856175050000119-gr5.jpg

For example, I need a reference voltage (precision) in one of inputs, and my DAC output in another and the output will be diference between they?

Thanks again.

I would to connect my circuit in analog input of a indicator (in your input of 0 .. 50mV). 

Thanks, I will study about that you say!

Large capacitors to ground on the output of opamps can make them unstable.  You could put a small capacitor across R5 and let the op-amp do the work of smoothing the output.  Possibly as little as 100 pF could solve it.

If you MUST put a large cap across the output, place a 100 Ohm resistor at the output pin of the op-amp.  if you connect the feedback resistor to the far side of the 100 Ohm resistor, then the feedback loop will compensate for any DC error introduced by the series resistor.

Jon

Got it! But, i must to keep only capacitor in output of Op amp, without a load resistor? And others capacitors I should take out? I will try now! Thanks.

however my charge pump only generate -1.4V on negative supply, but, i think this doenst it the problem.

Charge pump style supplies can be extremely noisy.  Is you circuit on a PCB with good grounding, or on a breadboard?

Im using a board FR4 that I made and manufactured in China, but, i'm learning about good practices in PCBs. So, i consider that this ground is horrible, if I made today, I'd make totally diferent.

I'm developing a new board, and i changed the charge pump that I made with microcontroller to IC (that is charge pump also), but, believe that better than me.

Thank for all.

[David Hess]

Large capacitors to ground on the output of opamps can make them unstable.  You could put a small capacitor across R5 and let the op-amp do the work of smoothing the output.  Possibly as little as 100 pF could solve it.

That is not always true; it depends on the details.  If the capacitance is large enough, then it will result in dominant pole compensation when combined with the operational amplifier's output resistance.  It works especially well with an aluminum electrolytic or solid tantalum capacitor because the ESR of the capacitor adds some phase lead to the response.  10 to 100 microfarads commonly works but tests should be done to verify stability.

If you MUST put a large cap across the output, place a 100 Ohm resistor at the output pin of the op-amp.  if you connect the feedback resistor to the far side of the 100 Ohm resistor, then the feedback loop will compensate for any DC error introduced by the series resistor.

That works for any operational amplifier but if a low impedance is required at middle frequencies, then a capacitor as I described above directly across the output works better.