A neat little trick

[Sean_Goddard]

Has anyone tried reducing system noise by PARALLELING amplifiers or filters?

If you combine the outputs of two identical amplifiers, the overall noise level SHOULD drop by 3dB. If you use 4, that's 6dB, 8 gives 9dB and so on.

What happens is the random noise, correlates and cancels out.

Of course it's NOT actually 3dB you get, more like 2.5 but this can be improved by keeping resistor values low thus reducing Brownian noise. Not too low though, or you run the risk of increased thermal noise.

This technique is used in military sonar arrays, and whilst I can't give you performance figures, let's just say it is possible to detect targets a thousand miles or more away

[Carl-NC]

I have not personally used it, but it's been discussed elsewhere on the forums. I think Eric may have had a design with either 2 or 4 parallel preamps.

[Altra]

There is an example in the LT1028 datasheet. You can reduce the voltage noise at the expense of increasing current noise.

lt1028.pdf

[Sean_Goddard]

As each amplifier is effectively "fighting" the others, it stands to reason the noise current will increase, but, as the LT1028 data sheet says, the relationship is not linear. Extra current I can handle, but in our applications, we are, I believe, always fighting voltage noise.

[lucifer]

I've used 4 op amps in parallel for the first stage PI preamp and it worked well.

[Riss]

I've used 4 op amps in parallel for the first stage PI preamp and it worked well.

only 4 ?!? aren't you from Gabrovo? you are incredibly big misers there !!! the Aberdeen Scots are a very generous people compared to you ...with 8 pieces you will earn another 3 db, and thus you will register the cauldron full of gold. and you will stop writing in the forum

[lucifer]

only 4 ?!? aren't you from Gabrovo?

Nope, not even close.

with 8 pieces you will earn another 3 db,

Greater quantity doesn't mean higher quality.

and you will stop writing in the forum

What's your problem with my presence on the forum?

[Altra]

More on this subject.

https://www.analog.com/en/resources/...y%20%E2%88%9AN.

[ivconic]

Maybe it doesn't matter and it's not the same thing, I think I've seen "duplicated" opamps on some of the older Minelabs.
At the end, they are all summed up. I've always wondered what the point is?
It would be good if I could remember exactly which schematic I saw it on.

[lucifer]

At the end, they are all summed up. I've always wondered what the point is?

The point is that the signals are summed up but the noise being random in each op amp, when summed up actually gets reduced.

[Carl-NC]

You get an SNR improvement that is proportional to sqrt(N), where N is the number of parallel opamps. Each time you double the number of opamps you get a 3dB improvement in SNR. However, input noise current goes up by sqrt(2) and power & cost double.

[Riss]

Nope, not even close.


Greater quantity doesn't mean higher quality.



What's your problem with my presence on the forum?

I have no problem with your presence in the forum. if he dug up a cauldron full of gold, everyone would spend the gold on entertainment - cruises, travels, women - a merry life. and he wouldn't waste his time entering forums. Or am I wrong?

[lucifer]

I have no problem with your presence in the forum. if he dug up a cauldron full of gold, everyone would spend the gold on entertainment - cruises, travels, women - a merry life. and he wouldn't waste his time entering forums. Or am I wrong?

I'm rather interested in the technology and exploring new ideas for improving the circuits and not in chasing the wind / cauldrons full of gold.