Looking, QST construction article about measuring noise of a resistor

Don't you keep cryogens around? Freeze the resistor!

I think you can synthesize a room-temp active resistor that has below Johnson noise.

Seems like you could connect N opamps to one shared resistor, vary N, very R, and do some math, to separate the resistor Johnson noise from the opamp current-induced noise from the opamp voltage noise.

But you know the Johnson noise.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

" snipped-for-privacy@krl.org" wrote in news: snipped-for-privacy@googlegroups.com:

The first article I looked at in that mag was the one about making in inband gated synch video decoder for the early analog video "scrambling" methods. I think it was like Feb '82. I know it was a February issue.

Horn Construction article,

Mikek

I guess there is an optimum drain current (hence Vgs) for gate current noise. Zero gate current may not correspond to zero gate current noise!

Vibrating reed electrometers have pretty low gate current!

Most people don't have much to say about electronics.

The NMR people like cooled jfets for their preamps. They have lots of liquid helium, and like to keep their amps around 20K.

This is the opposite, FTMS preamp

jfet opamps in a vacuum with no cooling, and 1Meg gate resistors. It was designed by a couple of chemists.

What signal source needs such low-noise gain? Something that doesn't last, can't be signal averaged?

FTMS uses lots of signal averaging, but an ion in orbot eventually collides with some residual gas molecule and gets wrecked, so averaging time is limited.

Spice is great for rainy weekends too.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

John Larkin wrote in news: snipped-for-privacy@4ax.com:

It should be called Larkin noise, because you sure are guilty of spewing enough of it here despite claiming to be all about pure electronics, you have a way with weilding and using the shit pot stirring stick.

There are a couple of better opamps around, like AD8067 or LT1793. Maybe something even better.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

If you had a name, we could name something after you.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Yabut, long wires going into cryogen adds C. (I know there are work arounds for that too.)

Right like with clocks, if you have three of them you can compare them in pairs and figure out a unique number for each.

That is a lot of work, and measuring amp voltage noise is easy.

This says the same thing... pick the right resistor.

George H.

I'm sure there is a kickstarter lifestyle. Just invent something like that every few months and live the good life. Many involve graphite or nano somethings.

There's no legal limit to how many companies you can simultaneously be the CEO of.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Good catch. Thanks.

The bootstrap gain is nominally 1.0, but actually slightly less, which helps to insure it doesn't oscillate. This is to remove most of the bootstrapped sensor's capacitance from the amplifier's e_n-Cin noise, i_n = e_n Cin 2pi f. See AoE-III section 8.11 and page 538. The bootstrap gain value isn't critical. A source follower running at 3mA, with 4.7k load, and emitter follower should be above 0.95. If it's only 0.8, then 80% of that noise source is removed.

--
 Thanks, 
    - Win

radiator-in-the-world

How, in Gawd's name, do they expect to get away with such bullshit!

Oh, hang on, I know. It's the result of an education system designed to produce "Scam Fodder". :-(

--
Johnny B Good

Sure. A big resistor with positive feedback to the other end is the classical method, and at AC you can do it with transformers and so on. Good amplifiers are quieter than room-temperature resistors.

A good diode or a diode-connected transistor has an effective noise temperature of T_j/2, which gives a useful method for making not-too-high frequency noise measurements.

Yup. A resistor from the summing node to the inverting input and another one in the noninverting input works, at least at frequencies where the resulting gain peaking isn't important.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

If you have a lot of BF862s in stock, they parallel very nicely if you use them near I_DSS. I have about 30,000 of them for my retirement savings. ;)

The BF862's 1/f corner is near 1 kHz, but milliamp for milliamp it's hard to beat. The current-production CPH3910 is about 1 dB quieter, but its I_DSS is inconveniently high. I don't have enough experience with it to know how well it parallels.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

I'm open to suggestions.

The DUT impedance must be low against 2 or 3 Ohms, or such a low noise amplifier would not make any sense from the beginning.

The input impedance would have to be 10 or 100 times larger, so that it does not make the DUT noise collapse.

For cross correlation, one would have 2 amplifiers in par. and at

2*50 Ohms for example, the noise current would make heavy xcorr / averaging pointless.

The lower frequency corner of the amplifier would not be required by f-3dB but by the fact that the noise of the bias resistor must be effectively shorted through the low impedance DUT.

At f-3dB, Xc and Rin are equal; that's a far cry from shorting. And the non-shorted noise rises twice as fast as 1/f, so it is easy to spoil the noise corner of the amplifier.

A real guru (Scott Wurcer of AD) saw in 100 ms that the input noise of my first 20*ADA4898 was rising much too fast below 50 Hz.

The 20 10u WIMA foil caps in the first release were not enough; now there is a 4700uF wet slug tantal. $120. :-( AVX is much cheaper than Vishay.

The fat tantalum has proved to be dangerous for delicate DUTs; I have zapped a LED with stored charge. Now there are shorting relays that are sequenced properly by a BBB.

FET amplifiers make that easier. Their 22 Meg bias resistor may produce more noise voltage absolutely - but the shorting is easier without the sqrt, so the situation is a win after all.

BTW, there is another fail in cross correlation. For measuring the added phase noise of an amplifier or sth. like that, a clean oscillator signal is split (sp?) with a Wilkinson divider and one path gets the DUT. The 2 paths then are cross-correlated.

Now, there is a 200 Ohm resistor typically in the Wilkinson splitter and its noise goes into both outputs. But it anti-correlates because at each instant one output has the + side of the resistor and the other the - side.

cheers, Gerhard