Ipod Touch

Can anyone recommend a good book or web resource where I could learn about improving the bandwidth or noise performance of high gain op amp circuits by using more complex configurations?

I've searched for terms like "compound amplifiers" but the examples I've found show simply how to boost the output power of an op amp, which I'm not interested in. I've also seen one or two circuits where people have stuck JFET front ends on op amps, to get low Ein noise, but monolithic op amp inputs are so good these days I suspect that is not the path to follow any more.

I wondered about adding a gain block inside my feedback loop, so the input op amp didn't have to have a gigantic GBW, but experiments along those lines didn't seem to improve performance. So it looks like I need a medium-to-advanced level text which will show me how to use novel and obscure feedback topologies and what their tradeoffs are. Or perhaps there are search terms I haven't thought of which would widen my horizons a bit.

TIA,

Nemo

My favourite compound amp trick is to run a BF862 follower single ended, driving the noninverting input of a low noise bipolar op amp, e.g. an OPA4898, with a sub-Poissonian current sink in the FET's source lead. The current sink is controlled by a FET op amp that forces the FET to operate at a V_GS of zero (a "snooper loop"). Making a quiet current source mostly requires dropping a volt or two across the sense resistor, e.g. the emitter resistor of a BJT, and bypassing the daylights out of its base. (Use the base bypass to set the loop bandwidth, and pick a snooper with really low input capacitance, e.g. an ADA4817.)

Linearity is improved by cascoding the FET, so V_DS is constant or nearly so. That also reduces the input capacitance, and even reduces the high frequency noise, because it tends to null out the resistive part of the gate-drain feedback admittance as well as the imaginary part.

You get 1.3ish nanovolts of noise, plus very low current noise, the combo is as fast as the op amp, the linearity is fine, and there are no stability worries due to excess phase shift in the FB loop, because the FET is outside it.

The snooper op amp will be noisier than that, but if you make the snooper loop bandwidth narrow enough, its noise gets lost in the 1/f noise of the FET. You can get another few tenths of a nanovolt by using two BF862s in parallel.

Two-path loops are difficult to get right, at least if you need the settling artifacts to be below 1% or so--there are always small whoop-de-doos in the transfer function that wind up dominating the settling tails eventually.

It's possible to apply automatic tweaks, but that's only worthwhile if you're trying to do something hard that you can sell for $$$ or that will make you famous, or if you have some gigantic volume. (Joerg's auto-tweaked 74HC123 monostables come to mind.) (*)

One of Jim Willams's books has a chapter on oscilloscope vertical amplifiers that's well worth reading on this topic--I'm in an airport right now, or I'd look it up. There's also a Tektronix Concepts book on vertical amplifiers, which is a good read too.

Cheers

Phil Hobbs

(Coming to you from the Delta Sky Club in scenic Newark NJ) ;)

(*) An English friend of mine says that the difference between an Englishman and an American is that an American will lie, cheat, and steal for money, whereas an Englishman will do it for a knighthood. ;)

I wonder, if you put all this into a little box, if anyone would buy it.

An AC amplifier with a *lot* of paralleled BF862s in the front end would be interesting, too. What's the square root of 16?

John

Err... 3^1.2618595

John S

I was just today re-reading AN45 by the late great Jim Williams.

Rather poignant actually, as he describes late-night prototyping sessions in between feeding his new-born baby.

Anyway the first two show low noise low drift composite amplifiers.

Also AN21 "composite amplifiers"

Or just read the whole lot, see you in a few weeks :)

Fig 30: a beta-based current limiter!

John

Great leads, some meat to get my teeth into there. Phil's BF862 trick sounds eminently do-able. I've got Jim Williams' article on vertical amps, but the Tektonics Concepts book would never have occurred to me. I've downloaded the LT app notes and am about to go through them. Thanks!

Nemo

Does one have to be a golfer to answer that?

Somewhere between 3 and 5, last time I checked. ;)

I'll have to buy a few more reels.

The problem with doing that in a box is that folks want to hang 6 feet of cable the input!

Cheers

Phil Hobbs

Write a WEB Page on a local PC as a Http Server. THis PC would also be the one taking care of getting the data from the device. Your Ipod can then browse to this page via the IP address of the server machine using your web browser.

Simple enough, I guess.

Jamie

Maybe check this out?

you could also check out Bluetooth.

Best regards, Spehro Pefhany

Classic 3-bottle circuit ;)

Cheers

I done two-bottles-of-Drambuie chip designs... one night, finished by dawn ;-) ...Jim Thompson

Just remembered that I have at least one paper on increasing gain-bandwidth product by using various compound OpAmp configurations.

It's not on a PC, it's on paper, so if you'd like a copy let me know, and sometime in September I'll scan it and send it out.

(I'm doing work at a secure installation, so must be on-site in NY :-( ...Jim Thompson

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

Still in NY? At least the weather isn't bad, but it is our monsoon season. There is supposed to be a great Greek restaurant in either Huntington or Port jefferson, can't remember which. Theres a long wait for a table. Port Jefferson is nice in the summer, like Huntington lots of people mulling around.

Cheers

The natives think it's "hot" ;-)

I'm staying in Melville, but going to Huntington daily, I'll look around... my GPS is due to arrive tomorrow... these cowpaths are hard to find your way through :-(

...Jim Thompson

Wouldn't this require setting up an apache2 web server on the PC? Simple, but not trivial. Most of the time, you would probably set up a full blown WAMP.

There is also windows IIS if you don't want to try apache2. This is potentially easier on a windows box, but I have no first hand experience with IIS.

from

=20

There is no need for such a heavyweight solution. Web pages hosted by nothing more than a pic have been demonstrated.

?-)