High input impedance opamp gain choices

[philwinder]

op amp high-input

Hi,
When designing opamp based low noise amplifiers, is it better to have only one amplifier providing a high input impedance and the gain, or is it better to add another opamp to provide the gain?

My situation is that I am designing a high input impedance opamp based amplifier for a hydrophone, and I dont know whether to put any gain on the first stage, or just leave it as a follower.

I understand that the gain resistors are one source of noise, so leaving them out would reduce the noise in the first stage, and this should improve things quite a lot considering the high input impedance of the hydrophone transducer. However, this would mean that I would have to add another gain stage and therefore adding more opamp + component noise.

Which is best?

Thanks,
Phil Winder

 

[saro_k_82]

high input impedance op amp

The first stage should have gain to reduce the overall noise. Source-follower first stage is a bad idea as it introduces lot of noise without improving the signal level. If the first stage is a low noise gain stage, you need not worry about the noise from the subsequent stages.

 

[philwinder]

single transistor input

Yeah, makes sense. Thanks!

 

[FvM]

op amp high input impedance

I understand that the gain resistors are one source of noise, so leaving them out would reduce the noise in the first stage.

I don't agree to the conclusion. You rather should use a low ohmic feedback network ("gain resistors") to minimize the noise. This is generally easy for voltage amplifier.

As another remark, the source impedance and the frequency band of interest should be considered for a more detailed noise analysis. In principle, a single transistor input stages achieves lower noise than an OP, except for the low frequency range, where thermal fluctuations can be a dominant noise source and a differential input stage is needed to cancel the temperature sensitivity.

 

[philwinder]

low noise op amp high input impedance

FvM,

I don't agree to the conclusion. You rather should use a low ohmic feedback network ("gain resistors") to minimize the noise. This is generally easy for voltage amplifier.

Really? Using the resistor noise equation 4kTR, with a R of 0 (i.e. no resistor) the noise will be at its lowest.

In principle, a single transistor input stages achieves lower noise than an OP, except for the low frequency range

A PN4117 (a fet that I like for its high i/p impedance), has a noise of 10-50nv/sqrt(Hz) over the 1-100kHz range. The AD8066 (the amp I am using at the moment) has a voltage noise of 6-7nv/sqrt(hz) and 0.6fA current noise. (Nice!)

The datasheet's would argue that the opamp choice would be better.

the source impedance and the frequency band of interest should be considered for a more detailed noise analysis

Well the source impedance is high, so I know this will introduce a lot of noise itself, and I have used high value resistors on the input to match it, so what more could I do?

Thanks for you thoughts,
Phil

 

[FvM]

ad8066 current noise

Really? Using the resistor noise equation 4kTR, with a R of 0 (i.e. no resistor) the noise will be at its lowest.

Yes quite right. Unfortunately other criteria are setting a lower limit to the feasible resistance range. That's why a suggested low ohmic rather than zero ohm feedback...

A PN4117 (a fet that I like for its high i/p impedance), has a noise of 10-50nv/sqrt(Hz)

No problem to find a transistor, that has worse noise performance... Obviously, my remark is related to the comparision of single ended to differential amplifiers using the same transistor type.

Cause the improvement is only 3 dB, a low noise OP may be still the most convenient choice, and possibly the best among available parts.

I have used high value resistors on the input to match it

I guess, you are talking about bias resistors. They should have a resistance level that causes only a neglectable current noise amount, if possible.

 

[philwinder]

why the input impedance of opamp is high

I guess, you are talking about bias resistors. They should have a resistance level that causes only a neglectable current noise amount, if possible.

Yes I was, sorry. However I need the large resistances to get an efficient energy transfer, and the current noise will be proportional to the paralell resistance of the source and the bias resistors. So I do not have much leeway on the values.

However, I have been looking up bootsrapping to increase the input impedance rather than using large resistors, but I will ask that question in another post.

Many thanks,
Phil