Standard non-inverting opamp with a cap across + and - terminals...... Why?

A lot better than you could stand 90 degrees at 95% humidity.

Of course I do. That's why my stuff works. If I understand the power series, I know how many terms I need to use.

AKA "engineering."

Actually, most of them are collector's items, Tek 547s and HP 185s and such. Even a Lumatron sampler. We only really use 20 or so.

Why not do something on your own?

John

I know... I grew up in WV (90°F/90% was standard issue), that's why I live in AZ ;-)

...Jim Thompson

Maybe if you had anything of value to present, there'd be more interest.

Good Luck! Rich

You will have to get your preferred brand of intellectual junk food from some other supplier.

-- Bill Sloman, Nijmegen

input

ce) of

ter of

owed

s of

r op-

nt

N Q1

ate

So it isn't extra current running through protection diodes. In fact what you are seeing is the non-linearity in the response of the long- tailed pair as the voltage excursion across the long-tailed pair starts to approach the level where non-linearity starts to set in.

John Larkin likes to expnad the Ebers--Moll relationship as a power series. Since the high order terms are essentially the applied voltage over 26mV (kT/q at room temperature) you can see that higher order terms start getting significant as the signal voltage approaches 26mV (actually 52mV over the two transistors involved in a long-tailed pair, but 26mV is easier for people to recognise).

The 26mV comes from the Ebers-Moll relationship. If that looks like magic to you, you should be posting on sci.electronics.basic

That's dealt with a few more of the obvious points ...

In my case the signal being amplified was quite big - up to +/-10V - but the mask patterns we were writing were particularly fine, and the point of the work I was doing was not to take advantage of the 1nV per root Herz noise figures of the LT1028. The 3nV per volt of the OP-27 was already being swamped by the Johnson noise of the summing resistors involved, but the LT1028 had a better output stage, so we could use slightly smaller summing resistors, and the lower input noise didn't do any harm. It was a pretty sensitive application.

At that time Fairchild had gone over to a more sensitive electron resist, and on the first of the masks that they'd tried to write for their - then - new 100K ECL logic, the Johnson noise in the original amplifier had left visible bobbles in the lines written. The cleaned up amplifier wasn't all that much quieter, but any residual bobbles were too small to see.

-- Bill Sloman, Nijmegen

The 0th and 1st order terms don't rectify. The biggie is the 2nd order term. "Square law detector" it's called.

Spice this: an emitter follower biased at, say, 1 mA emitter current, base driven from a DC source, emitter bypassed to ground. Note emitter voltage. Now add one mV RMS high-freq AC to the base source. Note change in emitter voltage. It sure ain't zero.

You get 1 mV RMS in a 50 ohm environment from 20 nanowatts of RF. In a higher-impedance, maybe accidentally resonant circuit, one nanowatt can make a millivolt.

In the typical bipolar opamp, the input diff pair runs at low current. At high frequencies, the shared emitter/current source node is too slow to follow the RF down, so it's essentially bypassed.

Simple.

John

Yet getting 1mV from -60dBm input using a biased conventional silicon junction is a bit of exaggeration, isn't it?

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

I'm not sure what you're saying. What I meant was that 20 nW of RF could put 1 mV RMS of RF on an input pin of an opamp, and the input b-e junction can rectify that to produce a DC offset. The resulting DC offset will be in the ballpark of 10 microvolts, plenty enough to cause trouble. It doesn't take 26 or 52 millivolts of RF to rectify.

As a square-law detector, the dc offset is linear on RF power input. So 1 microwatt of RF rectifies to about a 500 uV DC offset. Then multiply that DC offset by the noise gain of the opamp circuit.

In the chambers of the San Francisco Board of Supervisors (the city council) one hears a blip-blip-blipblip sound over the PA system fairly often, and quite loud. You can hear it in person or on TV. It's apparently cell phone RF getting into microphones, and it's been like that for years.

John

r input

ence) of

atter of

llowed

ess of

lar op-

ias

rent

NPN Q1

the

ulate

of

I never suggested that it was protection diode current. It is the bases of the active parts of the circuits that pass the currents.

Yes but it, like I said, is an added problem that a bipolar will have but a MOSFET with the exact same gm character wouldn't have. The MOSFET would not have the bias current on the input pin so it wouldn't have a bias current to vary in this way.

I was only pointing out an error you had made. I was not getting into the argument between you and John Larkin. You made an error and I pointed it out.

Right. That was the discussion. There wasn't any talk of protection diodes I recall. Well, not until just now.

Cheers, James Arthur

Agree entirely. Sorry, I thought you were talking about the rectified DC of 1mV, which looks way too much for 20nW RF input and biased silicon PN junction at any impedance.

Oh, this effect is very typical in the audio, as they have to use cheap single layer PCBs. The audio cable works as antenna, creating hell of RF potential across the board.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

apart =3D

example

everyone.

wrote.

close

no hands

really

may think neural

can only be found within you,

food,

teach anyone.

=20

mean=20

=20

divine=20

But the text of the article seems to indicate that L. Ron Hubbard's Dianetics is the grandaddy about a decade earlier. .

hands

think neural

only be found within you,

anyone.

Or zen buddhism somewhat before that. But Werner made it big, in volume, and more importantly he cornered the market on justifying oblivious, selfish jerks, free of any pretensions to actual enlightenment.

John

Well, some of us might enjoy a little cool weather, but understand!

Charlie

Yes, that is a very common problem in PA systems. Some of it can be prevented by good grounding of the mike cables, and running them well away from the users, but there is no way to completely prevent it when a council critter sets his phone next to the mike... 8-)

Charlie

You could probably just drive over? What, 2-3 hours ?:-) (I used to do PHX all the way into LA in 3 hours. Then they started using airplanes ;-)

...Jim Thompson

It takes 26/52mV before a substantial proportion of the input RF is rectified, and the proportion maxes out at higher signal levels

But not above the 26/52mV where the proportion of the input signal being rectified stops rising, and the DC offset rises with signal level, which is to say with the square root of signal power.

-- Bill Sloman, Nijmegen

lar input

bsence) of

matter of

followed

less of

polar op-

bias

urrent

e NPN Q1

h the

rculate

to

is

ct.

h of

Since the problem isn't bias current, but the large signal transfer function of the amplifier, your conclusion happens to be wrong.

You think I made an error, and you pointed out what you think the error is. In the process, you've demonstrated that you don't know what you are talking about.

-- Bill Sloman, Nijmegen