Opamps are noisier...is this article really true?

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have got an old article about noise in high gain devices. The article says that by nature opamps are noisier than low gain amps. Is this really true or was it true when the article was written?


  • noise1.gif
    50.8 KB · Views: 1,011
Hey Artnyos,

Have you seen the schematic for the AD devices? Most of their amps have emitter resistors in their first stage diff pair.

I was struggling with the problem of noise in building an amp of my own design using MOSFETs source connected to bipolars bases while the bipolars were connected as a diff pair. Noisy like crazy.

Until that is I added 330 ohm emitter resistors.

I actually got the idea in the book "The Art of Electronics" under the chapter on low noise design.

I guess too with most op-amps noise is a relative term.

My 2 cents,
peranders said:
My own theory in the subject is that open-loop gain doesn't matter at all when we talk noise. The only thing which determines the noise is the first stage if the gain+noise is much more than the noise in the second stage (which is usually the case)

mmhhh .. you are on the right track...

Somewhere i have chapter 10 from M.Leach ... NOISE IN FEEDBACK AMPLIFIERS .. I cannot find it right know :bawling:

I will search on my laptop later!

Resistors and capacitors are the poorest elements that one can make on a chip. Inductors are just about impossible ... Thus, if you rely on quality components of those types to get great results, you might want to go discrete :)

The other problem that chip designers have is that you initially choose to go with an n-substrate or a p-substrate. Those substrates are ideal for one polarity of transistor. If you want the other polarity (complementary design), you need to invert the substrate, not really an ideal situation. TI has made some noise recently in this very topic.

To achieve noise levels of the best op-amps of today with discrete elements you have to work very hard. Some op-amps are close to the level of a resistor of 50 Ohm (intrinsic noise from physics).

Noise is also at least two things: Current noise and voltage noise. Some types of circuits excel in one (typically due to choice of input transistors), few excel in both but op-amps are still very very good on paper!


I did a tiny noise experiment long ago at my college. It isn't all that prestigeous or anything, I got really poor score on that paper. ...

Let's talk about the three noises, Johnson/thermal, ficker, and shot noise. Johnson noise was most prominent from my readings. (I built a small thermometer with resistors, cool huh?) I got ficker noise, but it is smaller than the noise due to 50/60Hz AC. Shot noise was so insignificant, I skipped it.

Thermal noise, ... is rather inevitable like Gabevee said. I wonder how I would kick that down. The integration and the minituralization should not result in distinct disadvantages. Rather wires are shorter/optimized and thermal noise could be treated as common mode (?). Sounds like it is a better idea to me. (I am skipping the part about fabrication and stuff 'cause I am learning that now ... :) )

1/f noise was a little bothering when I was insane and build a preamp with RF chips. 1/f noise is somewhat higher with these chips. This is not mentioned in specs because it assumes you won't use it in an audioamp. (Note this noise is not caused by 50/60Hz AC. The peaks you see at 50~60 and 100~120Hz aren't the result of 1/f noise.) To be honest, I did not observe any advantage or disadvantage of minitualization/integration.

Quantum level noise shouldn't come into problem when we are talking about audio frequencies. (I believe author is refering to the problem rising for high density high frequency modern microprocessors or something.) I think shot noise can simply be ignored in most of our applications. I wonder why the author spends long time describing this. Besides, I think shot noise presence shouldn't decrease or increase due to minitualization and integration.

As for stray properties, I think that is considerably higher with descrete counterparts. You know, wires and patterns are much bigger. Stray resistance could be a problem, but we are talking really small length and stray cap being so small, the chip stray properties seems insiginificant.

Personally, I think people prefers descrete designs for other reasons. I could be wrong but noise level isn't the reason for the preference. The real reason is ... I have no idea. All I know is ... some opamps sound gooood.

Was I correct in my deliveration? I am no Mr Pass. I could be saying something wrong. I am copying the results on my paper. ...

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.