nV/sq root hz question...

[GT Blocker]

Told ya I'd have more stupid questions!

After finishing the build 'stock', I've started playing with it a bit.

So here goes...(bear with me) I am using the LME49710 in place of IC6 (the NE5534). The 49710 has 5x the GBW, 25% higher slew rate, and 38.5% lower noise/sq.rt. hZ.
So far it works great, a drop in replacement.

Part 2 - I have also replaced IC8 (the TL072/082) with the LMP7702 precision RRIO (phono/cassette head) dual opamp that has a MUCH lower slew rate, a bit over half of the GBW, but 43.5% less noise/sq.rt.hZ.
I haven't noticed a problem with the low slew rate or GBW, but I'm guessing that by this point in the circuit we're back to the 'audio' range (which is what the 7702 is designed for).

So here's the question...NOT counting the noise from passives, etc, Would the noise from these two chips average out ({38.5%+43.5%}/2) to 41% less noise (nv/sq.rt.hZ), i.e. that added by these two chips in the circuit? Told ya it was a dumb question, but I wondered how all of you designers of these things do the figuring out on stuff like this. Thanks...GTB


PS- the LMP7702 is only available in surface mount soic, I had to make a 'carrier' from an IDC header in order to place it in a dip socket. Jameco has soic to dip adapters pre-made, but at a crazy cost...

[homefire]

Noise is accumulative.... It adds! No Averaging to it.


Do you Really Want to go there?


This is a example of thermal noise in a resister. OP amps main noise figure comes from THD Total Harmoni Distortion and Thermal noise. Noise is Noise and adds.

http://www.ecircuitcenter.com/Circui.../res_noise.htm

[GT Blocker]

What?

Gee, Homefire, that really cleared it up. Like duh.

Sorry if I seem snarky, but that 'answer' had nothing to do with the question.
SINCE the noise is additive (a given), AND if all passives remain the same (another given), WOULD the noise contributed by THESE TWO chips average out to 41% less than the 'stock' ones contributed to the circuit?

Simple question, I thought. Though the answer is probably a bit more complex than that, at least I hope it is more complex than '"gee, the noise adds up, dude".

[homefire]

Sorry Your Right!

I just got off work and snuffed it.

Circuit noise should be reduced by the reduction of noise of the two chips you replaced.

Still not averaged.

Better?

[Wreckdetector]

Re: nV/sq root hz question...

Without knowing anything about the circuit that you are using the chips in, it's difficult to give the definative answer. I'll give it a try though.

The simple answer is no.

It depends on the amplifier gains. If for example you connect two amplifiers one after the other to get a higher total gain. You would normally use the lower noise of the two as the first stage as it will contribute the bulk of the total noise.

This is a bit simplistic but it goes something like this:

Say both amplifiers have a gain of 10, one has an equivalent input noise of 100nV and the other 50nV.

If you use the low noise amp as the first stage you get an output noise of:

((10*50nV)+100nV)*10 = 6000nV

If you use the high noise amplifier as the first stage you get:

((10*100nV)+50nV)*10 = 10500nV

So you are using the same amplifiers but getting a lot more noise in the second case.

Hope that helps.