Hardening a simple op-amp circuit against EMI

My charity doesn't extend spending time on irrelevant examples. We are talking about the long-tailed pairs at the inputs to op amps, which is a rather different situation.

If you can manage to get that into yur head, you might then find something useful to say, but it seems unlikely.

-- Bill Sloman, Nijmegen

I'm sorry, it's clear I don't understand the relevance of your explanation.

Above, you wrote: "The relationship between voltage and current is still exponential for voltages below 26mV, so you can still get some rectification, even if it is then only a small proportion of the RF current that is getting rectified. In Jeorg's case, that small proportion was enough to create difficulties."

The e-b junction in ordinary bipolar op amps has a considerable forward bias voltage across it--much more than

26mV--and rectifies RFI by diode action. That, in turn, produces a DC offset, which I modeled.

Which input junction of your long-tailed pair is forward biased with 26mV, and what kind of transistor does it use? A magic one?

Cheers, James Arthur

A Slowman transistor. Never heard of it? No problem, you're just ignorant ;-)

...Jim Thompson

--
| James E.Thompson, P.E.                           |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
       How severe can senility be?  Just check out Slowman.

e e

The inputs to an op amp are the bases of the two transistors forming the long tailed pair.

For small voltage differences between the two inputs - less than 26mV

- the consequent difference between the two collector currents is pretty close to a linear function of the voltage difference, and any high frequency content in the input signal averages to zero

Once the voltage difference gets much above 26mV, the transistor that is biassed on starts getting most of the tail current, and NPN inputs start following the upper envelope of any high frequency contnet in the input signal.

As Jeorg poits out, there is still some non-linearity below 26mV, so you can still see some rectification of high frequency content.

Happy now?

-- Bill Sloman, Nijmegen

me

F s r

ve

t

Jim used to know this stuff.

-- Bill Sloman, Nijmegen

Aren't they hand carved out of solid blocks of silicon by European welfare queens?

--
http://improve-usenet.org/index.html

Goggle Groups, and Web TV users must request to be white listed, or I
will not see your messages.

If you have broadband, your ISP may have a NNTP news server included in
your account: http://www.usenettools.net/ISP.htm

Thanks for the clarification--your reasoning's clear.

But, yours is not the mechanism that detects RFI. The mechanism is RFI being rectified single-ended at each input, by the b-e junction. The b-e junction is forward-biased, ready, fast, and willing.

RFI doesn't flow between the inputs; that path doesn't matter.

Cheers, James Arthur

No. Most opamp input diff pairs run at low current and the emitters (and current source) have significant capacitance hence long time constants. The transistor that gets high-freq RF on its base has, in effect, a capacitor to ground on its emitter. At hundreds of MHz, it's a nice diode peak rectifier.

Where do you get the 26 mV thing from? There's serious rectification going on at much lower p-p swings.

Another real-life problem is that real systems, in boxes and on pc boards, tend to have ghastly high-Q resonances all over the spectrum, which magnify modest RFI into serious voltages; narrowband peaks like

50:1 are not unusual.

"Some" is enough to cause serious problems in many, many actual products. I can't use a lot of cheap electronics in my house because the ambient RF fields make problems. A naiive thermocouple signal conditioner design *will* get into trouble when people with cell phones stroll by.

John

ome

RF

is

or

ive

it

e s

Why do you think that?

Pity it doesn't do what you think it does.

In your dreams.

-- Bill Sloman, Nijmegen

Any current going through the base to emitter is base drive, and has the usual effect on the collector current; the long tail drives the (NPN) emitter up towards the positive rail, turning off the base current - or at least bringing it back to what the op amp designer had in mind.

Ebers and Moll. The "serious rectification" is your hypothesis, not an established fact.

Funny how difficult it is to get these high-Q resonances when you actually want them.

Sure. I've seen it.

-- Bill Sloman, Nijmegen

Theory, plus of actual measurement of real circuits.

Well, obviously differential RFI /could/ flow if your layout permitted it. Don't do that.

But that's not the usual case, nor the one we were discussing, nor does it produce the interference seen in real life, for the very reasons you argue. So, that's not it.

Cheers, James Arthur

Sorry, but in my life it is an established fact, evidenced on lots of scope plots. For many of my clients is also was. "Was" because we established serious rectification as the root cause of numerous EMI problems, fixed it, and they were able to forget about it and move on with the projects.

Calculations on paper and SPICE, performance in the real world, two different things.

My first encounter with this was at age 16 or so. Just had my ham radio license and a high power endorsement, built a huge amp, and promptly a neighbor called. I was surprised they let me (being a kid) open their freaking expensive electric organ. The complaint was a distinct "pock ... pock ... po-po-pock ... pock" when the light in my window was on. Morse code transients. Aha! Looked at the circuit board and almost freaked out. This manufacturer had no clue about grounding strategies. So I took a sawtooth pod to find the most sensitive areas. No chance to place ferrites or shield anything, this circuit board was an octopus of traces and wires sprouting to just about everywhere. So I placed small capacitors over the BE contacts -> bingo, problem gone. Back then I didn't know what on earth could cause that, now I know.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

He's heard it reported from three guys who actually design electronics, has seen it modeled (he likes models), and he still doesn't believe it.

He just wants to argue.

James Arthur

Maybe we should re-topic the thread to "Hardening of the minds" now :-)

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

I'm unable to really decode your ASCII schematic to see where the feedback loop around the amp is, but... in an environment with magnetic fields the area in the feedback loop (PCB trace wise) determines how big of a signal is induced by the magnetic fields.

Many years ago when there were 60Hz filament transformers everywhere, people learned real quick to make any loops be very low in cross sectional area... twisted pairs and lead dressing. This is often ignored by newbies today.

Tim.

Wouldn't it be way cooler to start one on Liberating the mind? There's certainly enough hardening of the minds to go around several times around, I seem to notice. >:->

Cheers! Rich

If it's important enough to you to bother, just set your newsreader for reading in a fixed width font, like courier.

Hope This Helps! Rich

Anything like a Slow(man)blow(s) fuse? ;-)

Sloman couldn't blow a picofuse.

--
http://improve-usenet.org/index.html

Goggle Groups, and Web TV users must request to be white listed, or I
will not see your messages.

If you have broadband, your ISP may have a NNTP news server included in
your account: http://www.usenettools.net/ISP.htm