Low noise opamp design

low noise op amp design

Hi guys,

I'm working on a dc cancellation circuit in a optical receiver circuit. Currently, i'm facing a problem in designing a low noise opamp for my feedback circuit. The function of this feedback circuit is to compare the dc voltage at the positive and negative input terminal of the opamp to be 1.2V and substract the input dc current of transimpedance amplifier to eliminate dc offset. When i do the transimpedance input noise analysis in cadence, i noticed that the feedback opamp contribute the large noise at the input signal. The opamp gain is 40dB. I need some reference or guide on how to design a very low noise opamp, thus my total noise will be reduced. The opamp is a differential input and single ended output.

Thanks,
Suria

low noise opamp

I'm not exactly clear on what you're trying to do with the amp, what is the 1.2V. There are far less noisy ways to cancel a DC offset than with an op-amp.
I'm no expert in low-noise op-amp design, but I do design CMOS camera's which contain op-amps amongst othere things. We cancel DC quite often in our chips, but never use op-amps to do it.
If you're looking to increase your ac acainst the dc background, try doing a search for an HDA (designed by mead - vision chips), a hysteretic differentiating amplifier, unity gain at DC and an ac gain of (depends on the ota) around 100.

You can remove DC with high-pass filter, but I'm not sure of the constraints of your design.

I hope you resolve your problem.

transimpedance amplifier dc cancellation

Old Nick said:

I'm not exactly clear on what you're trying to do with the amp, what is the 1.2V. There are far less noisy ways to cancel a DC offset than with an op-amp.
I'm no expert in low-noise op-amp design, but I do design CMOS camera's which contain op-amps amongst othere things. We cancel DC quite often in our chips, but never use op-amps to do it.
If you're looking to increase your ac acainst the dc background, try doing a search for an HDA (designed by mead - vision chips), a hysteretic differentiating amplifier, unity gain at DC and an ac gain of (depends on the ota) around 100.

You can remove DC with high-pass filter, but I'm not sure of the constraints of your design.

I hope you resolve your problem.

Click to expand...

Hi Nick,

Thanks for the reply. Actually what i said 1.2V is the dc voltage at the positive and negative input of the opamp. In our system, the common mode voltage is set 1.2V, that is why we need to make the output dc voltage of the transimpedance amplifier to be 1.2V. If you at the attachment which show the receiver system, the opamp compare the different in dc voltage and start to sink the input dc current in order to eliminate dc offset. My problem is, i need the opamp to be higher gain about 60dB for less dc offset but at the same time the noise also is amplified 60dB from the feedback to the input of transimpedance amplifier which degrade the performance. So, i'm looking for a solution on how to decrease the total noise at the output of the opamp and at the same time need to maintain the opamp gain to be high. U said you have used some dc cancellation techniques..please share with me, i still don't understand them.

Thanks,
suria

noise opamp

Hi,

Is your design on-chip or are you building this from IC's?
So am I right in thinking that you're converting the singal ended signal AC+DC to a differential signal with a 1.2V offset, but the DC from the pixel is causing the differential signal to be unbalanced?
The best solution would be to filter the Pixel output if this is the case.
Consider using a HDA at the backof the pixel (which is basically an amp with a filter in the feedback loop) which feeds DC back to the inverting input, thus you amplify your AC signal and the DC component is left unchanged. This only works well with a small ac input signal. Then Filter the signal to strip the DC off, then convert to a differetial signal.
You should at least introduce your gain stage prior to your main noise source, to improve your signal to noise.

I hope some of this helps,

let me know how things are going.

Cheers,

Nick

op amp eliminate dc

Old Nick said:

Hi,

Is your design on-chip or are you building this from IC's?
So am I right in thinking that you're converting the singal ended signal AC+DC to a differential signal with a 1.2V offset, but the DC from the pixel is causing the differential signal to be unbalanced?
The best solution would be to filter the Pixel output if this is the case.
Consider using a HDA at the backof the pixel (which is basically an amp with a filter in the feedback loop) which feeds DC back to the inverting input, thus you amplify your AC signal and the DC component is left unchanged. This only works well with a small ac input signal. Then Filter the signal to strip the DC off, then convert to a differetial signal.
You should at least introduce your gain stage prior to your main noise source, to improve your signal to noise.

I hope some of this helps,

let me know how things are going.

Cheers,

Nick

Click to expand...

Hi Nick,

It is on chip design. Actually we are converting the differential signal to a single ended signal through a opamp. We are more interested in DC voltage as the AC is already filtered through a low pass filter. So, when the system is fully balanced, the dc voltage at both input of the opamp will be 1.2V but the output of the opamp will vary from 300mV to 1V in order to regulate the input dc current from creating a dc offset.

So, currently the opamp gain is 40dB but the noise appearing at the input of the transimpedance ampifier input is 100 times of the total feedback noise. Thus, i need to reduce the gain of the opamp (something in feedback of the opamp itself), thus the noise will be a few times only.

I don't understand what is Pixel here. If you don't mind can you show me your idea in block diagram approach, may be i can understand better in that way...

Thanks,
Suria

remove noise of a op-amp

Sorry about the late reply, I've been moving house and forgot about this.

Right, when I am talking about a pixel that is the photodiade and all processing circuitry unique to that photo-diode, i.e. in a 64x64 array it is everything attached to each pixel, not any coulmn based processing.

You say you're trying to convert a differential signal to a singe ended? It looks like you're doing it the other way round.
I assume the Low-pass filter you are talking about is in you're DC cancelation circuit (which we want to get rid of since it doesn't work properly). And since you're trying to cancel DC I am also assuming that it is in fact the ac component you're interested in.

If you tell me the black-box details of your pixel/circuit (not the DC cancellation thing you're doing) I mean the light in to signal out. I will happily draw a block diagram solution for you.

low noise op amp filter circuit

Are u not using CMFB in the fully differential amplifier(112) ?

If ur using why do you need the feedback of DC cancellation ckt?