Need suggestion on high-precision amplifier

Hi all, I am new in this forum and find it quite helpful.

Here is my design of the amplifier:
Signal will come out from a photo diode with 300mV DC component and 60mVpp AC component.
I will use an opamp to amplify the signal from photo diode firstly.
Secondly, a capacitor will be used to block the DC part of the signal.
Thirdly, the AC part of the signal will go through a bandpass filter(1kHz-20kHz).
Fourthly, the signal will be amplified with another opamp.

Very high precision is needed(1uV/60mV), will the amplifier work?
Are matching networks needed? Opamps have good input/output impedance.

Really appreciated.


Richie

 

Define "precision": do you mean that the noise level must be 1μV in 60mV?
This seems a very strict requirement, and would imply that the incoming optical signal was itself very clean, perhaps impracticably so.

 

Depends on how much money you want to spend in components.

Yes, it's possible, but the level you are looking for will require a very strictly regulated supply, and high end operational amplifiers. I'm assuming this is not for audio use. All of which adds zeros at the end of the pricetag before the decimal point.

What is the actual application that needs to be measured/amplified? There may be a better solution to the problem than a solution for your solution.

 

My guess is that your 300mv DC (where ever its coming from) is not going to be noise free. It is probably going to contain noise in your band of interest. I bet that the op-amp you finally choose will *not* be the limiting factor in overall performance.

But, then again, I have no idea what you are trying to do.

 

Hi Adjuster

I made a mistake in the calculation. The noise should below 25uV instead of 60uV

 

Hi thatone, I plan to spend around 100USD on the amplifier part.

I made a mistake that the noise should below 25uV instead of 1uV out of 60mV.

Actually, I am building a bias controller. The controller will control the biasing voltage , which is a DC voltage, of an optical modulator.
If the bias voltage is tunned correctly, pure sine wave will be the output.
If the bias voltage is not tunned correctly, sine wave will be distorted and the Vpp value will decrease.

The precision needed for bias voltage is 1mV, which means 26.79uVpp distortion of the sinewave.

 

Thanks joey

I am building a controller and explained above.
I am trying to look at the signal in time domain since the algorithm is quite simple in time domain. But I am not sure if it is a good way.


Richie

 

Are there any considerations in designing schematic and PCB layout?

 

It may be helpful to "guard" the input of the op amp, the technique is discussed in Section 5 of the "Sensor Signal Conditioning" seminar by Analog Devices, starting on page 7.

http://www.analog.com/static/imported-files/seminars_webcasts/371366216sscsect5.PDF

It also contains some advice on how to select a suitable op amp for amplifiying the signal from a photodiode. But when you look at the examples they give, keep in mind that they are only considering op amps from Analog Devices. I'm not saying these are bad parts, but you really should look at op amps from other companies too before spending your 100 USD....

 

I am curious: Do you know the precise frequency of the sine wave? If so, there is a better way!

 

Yes!!!! Could I hear your idea?

The frequency is 1kHz.

 

Well, why didn't you say so?

What you want to build (or buy) is called a 'lock-in amplifier' (LIA). Also known as a phase-sensitive detector or a synchronous detector.

Essentially, you simply multiply your signal by a stable 1 khz sine wave (reference signal) that is in-phase with the signal you are trying to detect. The product contains a DC component which is proportional to both the amplitude of the of the input signal and its phase relationship with the reference signal. The last stage of the LIA is a low-pass filter. The longer the time constant, the slower the response, the higher the sensitivity, and the greater rejection of other interfering noise frequencies.

Generally, you use the same clock for both signals so that they are always frequency matched and you don't have to worry about frequency drift. Then, you have a phase adjustment where a knob is turned for peak amplitude output. Then, any changes in the amplitude of the incoming signal are reflected as a corresponding change in the resulting DC signal.

Sometimes, it is difficult to maintain the proper phase relationship, and you will get phase noise in your output (i.e. a change in DC due to phase differences in input vs. reference, not changes in AC amplitude). I prefer quadrature detection in these cases. This works by multiplying by two reference signals of the same frequency, but 90 degrees out-of-phase. Then you have a 'complex' quantity that contains both amplitude and phase info. You can extract that amplitude information while rejecting the phase info.

Even better, you can eliminate the requirement for a sine/cosine oscillator for either method above by multiplying the incoming signal by a 1khz square wave of +/- 1 amplitude. This is simply a matter of inverting/non-inverting your signal at 1khz and filtering the result. Very easy to do! I even do it with a CPU instead of analog components (haven't done it at 1kHz, though!).

Hope this helps!

 

Hey Blofeld

This is extremely helpful...
Really appreciated.

Richie

 

Hey Joey

Surely it helps!

You are right, I should have been clearer.
How is the accuracy of multiplexer? And how to generate accurate sine wave?
I have looked some material on the internet and it seems the best way is to use an DAC with a low-pass filter.

Richie

 

Again, you don't really need a sine wave...just invert/non-invert the incoming signal at 1khz. Then filter. It's really pretty simple. If I had time, I'd draw up a rough schematic for you. Unfortunately, I'm pretty busy these days...

 

I have a PDF on lock in amplifiers, maybe two.

See attached.