Electronic Design

# Bob's Mailbox

HI BOB,
Quite some time ago, I sent you a circuit similar to this file (see the figure). You were very kind and answered all of my questions. However, one thing you said was that the noise gain of this circuit is 1. How did you arrive at that value?

(If the VOS of the op amp changes by 1 mV, the voltage across the load R will be changed by 1 mV. The delta VOUT of the op amp will be the same 1 mV, so the dc noise gain will be 1.0 by inspection. Now at 10 kHz, the noise gain will surely rise above 1. It should never be a surprise if the noise gain rises. This could easily be because the ZIN of the op amp is capacitive, and the ZIN of the MOSFET will be capacitive. So, the noise gain can rise at 9 to 12 dB per octave. This will lead to oscillation and instability. One of the first Band Aids is to put a small CF or series R-C network from the op amp’s output back to its negative input. Maybe 100 O and 100 pF? Don’t try just one value. Try several R-C values and see what works better in terms of not so much ringing. Have a ball! /rap)

The formula I have for noise gain is (RF/RIN) + 1. (Uh-uh. It is ZF × (sum of all input admittances) + 1. Your formula only applies to a case of an inverting op amp. The definition is delta VOUT/delta VOS, and it will be frequency-dependent. /rap) So in my circuit, I’m assuming RIN = R2 = 1k. What is RF in my circuit? (Not applicable. /rap)

Or is there a different formula to calculate noise gain for my circuit? Why I’m asking about noise gain is that, as you have pointed out in your article on noise gain, noise gain affects stability (see “\\[\\[What-s-All-This-Noise-Gain-Stuff-Anyhow-7164|What’s All This Noise Gain Stuff, Anyhow?\\]\\]”). So, I’d like to understand how the stability of my circuit is affected by noise gain.

RICHARD DZIOBA

HELLO, RICHARD,
If the noise gain stays low or constant, you are usually in good shape for stability. If the noise gain starts rising, at midfrequencies, it’s important to do something about it.

RAP

DEAR BOB PEASE,
I wonder why Early voltage never seems to be shown on a transistor sheet. (Or am I wrong about that?) (When transistors were new 40 years ago, they would show a typical family of IOUT versus VCE for various values of IB. You could use that to estimate the working range of VA. These days, paper is too expensive. /rap)

It seems to me that would be useful information, even if it’s only a range. (I’m thinking it would help a person to estimate the output impedance of a current source particularly.)

(High-beta transistors tend to have low VA, and low-beta transistors have high VA. What you want to know is the product of mu × beta, which tends to be a constant for any specific type. Different types may have higher or lower for the product. Also, is that product 1 million or more? Or lower? Then if you know the range of beta, you know the range of VA. /rap)

Until recently, I had thought the range was narrow, so it might be settled by measuring the effect for a few transistors. But I learned better. A colleague’s data taught me that the range of VA from a super-beta’s to a high-voltage part is very wide: from about 130 to 6400 (430 for a 2N4904). (And it can also go as low as 50 or as high as 25,000. /rap) Is there a complication that I don’t know about that precludes giving this specification?

TOM HAYES

HELLO, TOM,
Hardly anybody asks for it. And those that need this info know how to get it. Now you do. Have fun!

RAP

HELLO, BOB,
Here is a little something for you to think about. How do I drive a high-impedance microelectromechanical-systems (MEMS) filter from a 50-O source and then transition back out to a 50-O load and keep 50 O friendly?

DAVID SMITH

HI, DAVID,
You must ask the guy who makes and sells the MEMS filter. I mean, how am I supposed to guess what the MEMS likes to be loaded with? I have no idea.

If it likes to see a high-z load, then use an op amp as a unity-gain follower. Or an emitter follower. Or a source follower. Or a current-feedback amplifier. Let it drive the 50 O.

The fact that you say you want to work at 50 O indicates that you have some preconceived notion of the frequency range where this is supposed to run. And I can’t guess that either. And if you ask Dear Abby, I bet she can’t guess either!

RAP

Comments invited! [email protected] —or: R.A. Pease, 682 Miramar Avenue San Francisco, CA 94112-1232