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

On Apr 10, 10:03 pm, John Larkin
<jjlar...@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
On Fri, 10 Apr 2009 10:21:39 -0700 (PDT), bill.slo...@xxxxxxxx wrote:
On Apr 9, 8:01 pm, John Larkin
<jjlar...@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
On Thu, 9 Apr 2009 08:27:51 -0700 (PDT), bill.slo...@xxxxxxxx wrote:
On Apr 9, 5:54 am, John Larkin
<jjlar...@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
On Wed, 08 Apr 2009 22:56:40 GMT, James Arthur

<bogusabd...@xxxxxxxxxxx> wrote:
Jan Panteltje wrote:
On a sunny day (Wed, 08 Apr 2009 09:59:47 -0700) it happened Joerg
<notthisjoerg...@xxxxxxxxxxxxxxxxxxxxx> wrote in
<bE4Dl.15236$8_3.12...@xxxxxxxxxxxxxxxxxxxx>:

bill.slo...@xxxxxxxx wrote:
MOSFETs typically offer one or two orders of magnitude less
transconductance than bipolar transistors, so that where the bipolar
input stage produces a substantially non-linear ouput for input
signals over 26mV (at room temperature) effectively "rectifying"
larger input excursions, the MOSFET input stage stays linear for input
signal ranges of up to about a volt.

Joerg has never got stuck into the mathematics of what is going on, so
he still thinks that high frequency noise is being rectified by the BE
junction, and consequently can't believe that the gate of a MOSFET can
"rectify" a (bigger) signal in exactly the same way.

He doesn't just think that, he's experienced it. Or rather, a client has
and now it's gone :-)

Oh, and it's got nothing to do with 26mV.

--
Regards, Joerg

I could be wrong, but I think this depends as much or more on the circuit
topology then on the transfer curve of the semiconductor in question.

The topology was a long-tailed (differential) pair.

The word 'common mode' comes to mind.
Although Bill Slowman may have a point somewhere mathematically,
nobody is driving RF into the base of a transistor with the emitter grounded.

Parasitic capacitance at the emitter node makes that pretty
much the case at UHF, right? You'd be driving RFI into the
base(s), with the emitter(s) quasi-grounded.  The transistor's
live, making it a pre-biased detector.

If you've got equal antennae at the inputs then you'll rectify
a common mode signal.  If unequal, a normal mode voltage.

Even if the induced signal is common-mode balanced, it's likely still
a problem: op-amps' CMRRs generally suck @ 1GHz, so common mode RFI
shows up at the output, peak-detected & low-pass filtered.

No 26mV required.

Indeed not. Lots of people will sell you zero-bias diode detectors
that operate in their square-law region. A typical sensitivity is 0..5
mV DC out per microwatt of RF in, which is about 7 millivolts RMS for
a 50-ohm detector. Such a detector outputs about 5 uV DC at -50 dBm RF
in, about 700 uV RMS. 5 uV is plenty enough to trash a
thermocouple/load cell/microphone amp.

Diodes don't quit detecting below 26 mV.

You-know-who might consider googling "square-law diode detector" and
"zero bias diode detector". Adding bias, as an opamp front-end does,
just improves RF detection a bit.

There's no abrupt change in detection efficiency at 26mV, but above
26mV the pre-biased silicon diode is converting an appreciable
proportion of the incoming RF into DC (most of it when you get up to
50 or 100mV or above), while below 26mV the diode looks more and more
like a resistor and a declining proportion of the incoming RF gets
converted to DC.

It's basic math. Diode current is exponential on voltage. An
exponential can be approximated by a power series. Only even-order
terms rectify. The dominant even-order exponent is 2. So diodes act
like square-law detectors at low levels. And a millivolt of RF is
plenty enough to cause trouble.

I'm surprised I have to explain stuff this simple.

But you said...

"My point is that small RF currents won't get rectified; if they are
big enough to get rectified they are going to show up as a more-
than-26mV voltage excursion between the base connection to the input
transistors."

so now you're weaseling.

No. I'm clarifying what I said earlier, and filling detail.

Obviously, what I should have said is that only a proportion of the
sub-26mV signal gets rectified, and that proportion decreases as the
signal declines below 26mV, but it's hard enough to get you guys to
read short declarative sentences without raising the level of
difficulty with qualiflying clauses.

Your own responses have been getting progressively more intellectual
content as the discussion goes on, and to that extent you too are
"weaseling".

As long as you quit repeating the "26 mV" nonsense, I've done a
service to humanity. RF power meters and crystal radios can continue
to work without fear of your disapproval.

As long as you fail to appreciate that the 26mV is a way of referring
the reader back to the Ebers-Moll relationship, I've still got some
work left to do.

--
Bill Sloman, Nijmegen
.

Relevant Pages