emi with op-amps

[john]

I have a circuit with a LM358 installed. The trouble is when I place a
two way radio near the circuit the LM358 output changes, (by about
30%). I have placed a JFET input op-amp in the circuit and there are
no problems. I'm told that this is a common problem with BJT input
op-amps. The input offset voltage on the JFET op-amp is too high so
I'm thinking about using a CMOS op-amp instead. Am I likely to have a
similar problem with a CMOS op-amp?

This is NOT a layout problem because the JFET op-amp works fine. The
problem is caused by the op-amp itself. Does anybody know the
mechanism behind the EMI susceptability on the various types of
op-amps. Any intellegent comments greatly appreciated.

 

[John Woodgate]

I read in sci.electronics.design that john <johnshadie473@yahoo.com.au>
wrote (in <e7047e.0403150236.5d6fb405@posting.google.com&gt about 'emi
with op-amps', on Mon, 15 Mar 2004:

I have a circuit with a LM358 installed. The trouble is when I place a
two way radio near the circuit the LM358 output changes, (by about
30%). I have placed a JFET input op-amp in the circuit and there are
no problems. I'm told that this is a common problem with BJT input
op-amps. The input offset voltage on the JFET op-amp is too high so
I'm thinking about using a CMOS op-amp instead. Am I likely to have a
similar problem with a CMOS op-amp?

Much like the JFET, I expect, but will it really give you lower offsets?

This is NOT a layout problem because the JFET op-amp works fine. The
problem is caused by the op-amp itself. Does anybody know the
mechanism behind the EMI susceptability on the various types of
op-amps. Any intellegent comments greatly appreciated.

Well, it's rectification of the r.f. at the junctions of the input
devices, and the JFETs are less efficient rectifiers at low input
voltages than the bipolars. Note that there is no 'threshold' signal
level below which rectification does not occur; it just changes from
square-law to linear law over some range of levels.

A 47 pF ceramic cap directly between the + and - inputs of your LM358
may solve the problem anyway. You may need to experiment with the value,
but remember that many caps are self-resonant well below UHF. Don't try
100 nF!
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

 

[CBarn24050]

Any pn junction will rectify a strong rf field. Most chips are unaffected by rf
because of their small size compared to the rf wavelength, if your circuit is
suseptanble then it is almost certain to be a layout problem.

 

[DJ]

On 15 Mar 2004 02:36:24 -0800, johnshadie473@yahoo.com.au (john) wrote:

I have a circuit with a LM358 installed. The trouble is when I place a
two way radio near the circuit the LM358 output changes, (by about
30%). I have placed a JFET input op-amp in the circuit and there are
no problems. I'm told that this is a common problem with BJT input
op-amps. The input offset voltage on the JFET op-amp is too high so
I'm thinking about using a CMOS op-amp instead. Am I likely to have a
similar problem with a CMOS op-amp?

This is NOT a layout problem because the JFET op-amp works fine. The
problem is caused by the op-amp itself. Does anybody know the
mechanism behind the EMI susceptability on the various types of
op-amps. Any intellegent comments greatly appreciated.

The BJT chip has PN junctions (mainly the b-e junctions of the input
transistors) which act as detectors and rectify the RF. The JFET does not have
the same junctions already biased to act as effective RF detectors (the gate
junction is strongly reverse biased). You should expect a CMOS amp to be
similarly immune, though there are never guarantees with this kind of thing!

Your deduction that this is not a layout problem is a mute point. If you had
nothing but BJT amps available you would have to find a layout, or at least a
design, that was capable of withstanding the RF. That might involve RF
bypassing, chokes, shields or whatever.

dj

 

[Spehro Pefhany]

On Mon, 15 Mar 2004 11:00:48 +0000, the renowned John Woodgate
<jmw@jmwa.demon.contraspam.yuk> wrote:

A 47 pF ceramic cap directly between the + and - inputs of your LM358
may solve the problem anyway. You may need to experiment with the value,
but remember that many caps are self-resonant well below UHF. Don't try
100 nF!

The cap reduces the phase margin, not that a '358 with typical
feedback resistors will notice.

Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com

 

[Fred Bloggs]

john wrote:

I have a circuit with a LM358 installed. The trouble is when I place a
two way radio near the circuit the LM358 output changes, (by about
30%). I have placed a JFET input op-amp in the circuit and there are
no problems. I'm told that this is a common problem with BJT input
op-amps. The input offset voltage on the JFET op-amp is too high so
I'm thinking about using a CMOS op-amp instead. Am I likely to have a
similar problem with a CMOS op-amp?

This is NOT a layout problem because the JFET op-amp works fine.

It's obviously a layout problem for the '358.

The
problem is caused by the op-amp itself. Does anybody know the
mechanism behind the EMI susceptability on the various types of
op-amps. Any intellegent comments greatly appreciated.

I dunno- what is "suscept-ability"?

 

[R.Legg]

johnshadie473@yahoo.com.au (john) wrote in message news:<e7047e.0403150236.5d6fb405@posting.google.com>...

I have a circuit with a LM358 installed. The trouble is when I place a
two way radio near the circuit the LM358 output changes, (by about
30%). I have placed a JFET input op-amp in the circuit and there are
no problems. I'm told that this is a common problem with BJT input
op-amps. The input offset voltage on the JFET op-amp is too high so
I'm thinking about using a CMOS op-amp instead. Am I likely to have a
similar problem with a CMOS op-amp?

This is NOT a layout problem because the JFET op-amp works fine. The
problem is caused by the op-amp itself. Does anybody know the
mechanism behind the EMI susceptability on the various types of
op-amps. Any intellegent comments greatly appreciated.

Are two-way radios likely to be 'placed' near the IC? What is 'near'?

The susceptibility is increased if higher impedance, longer tracks are
routed to the input pins. Copper tracking length to input pins, from
passive gain-setting components, should also be minimized. SMD bodies,
with ground plane are less susceptible.

I suspect that the device change merely resulted in a more acceptible
degree of visible interference. What fet input device did you use that
ran on the single supply of the 358? There are higher-speed op amps
that have a similar CM input range that show more obvious response to
layout quality - 34071-2 is a good example. Improved local supply
decoupling can also improve immunity, strangely enough, especially if
inputs are near ground.

While I've seen small caps placed across input pins, by people who
claim 'they work', I've never seen it done in a gain-stage that was
visibly stable under a simple inspection of their output with an
oscilloscope. App notes only recommend series RC in that position,
never a straight C. The added pick-up area produced by the extra parts
is generally self-defeating.

If the 'place'ment of a two-way radio means the physical layout of
this circuit permanently in the assembly, it should be partitioned
physically and in it's supply and connections, from LF analog cctry.
Reorietation of field-generating components might help. The single
part-swap cure is a band-aid that is not guaranteed to work the next
time it happens in the same location.

RL

 

[Bill Sloman]

johnshadie473@yahoo.com.au (john) wrote in message news:<e7047e.0403150236.5d6fb405@posting.google.com>...

I have a circuit with a LM358 installed. The trouble is when I place a
two way radio near the circuit the LM358 output changes, (by about
30%). I have placed a JFET input op-amp in the circuit and there are
no problems. I'm told that this is a common problem with BJT input
op-amps. The input offset voltage on the JFET op-amp is too high so
I'm thinking about using a CMOS op-amp instead. Am I likely to have a
similar problem with a CMOS op-amp?

This is NOT a layout problem because the JFET op-amp works fine. The
problem is caused by the op-amp itself. Does anybody know the
mechanism behind the EMI susceptability on the various types of
op-amps. Any intelligent comments greatly appreciated.

This is a layout problem, in the sense that the components connected
to the inputs of the op amp are acting a receiving aerials for the RF
radiated by your two-way radio. A more compact layout, built on top of
a ground plane, with some by-pass capacitors to shunt the RF away from
the op amp inputs, would be a lot less susceptible to the RF.

The other responses talk about "rectification" and suggest that the
diodes associated with the LM358 inputs are acting as rectifiers.

This probably isn't true. If you expose as long-tailed pair to enough
RF, and this applies both to BJT and FET long-tailed pairs,
NPN/N-channel bases/gates look as if they follow the upper envelope of
the RF, while PNP/P-channel bases/gates look as if they follow the
lower envelope.

If you think about the way a long-tailed pair actually works, the
reason is obvious enough.

"Enough" RF is enough to drive the voltage difference between the two
side of the long-tailed pair outside the region where the transfer
function is more or less linear - more than about 50mV for BJTs and
more than about a volt for FETs, whose transconductance is a lot
lower.

MOSFETs are just another sort of FET, so a CMOS-input op amp would be
expected to act like a FET-input op amp in your circuit.

------
Bill Sloman, Nijmegen

 

[John Woodgate]

I read in sci.electronics.design that Bill Sloman <bill.sloman@ieee.org>
wrote (in <7c584d27.0403151041.38b1fe43@posting.google.com&gt about 'emi
with op-amps', on Mon, 15 Mar 2004:

The other responses talk about "rectification" and suggest that the
diodes associated with the LM358 inputs are acting as rectifiers.

This probably isn't true. If you expose as long-tailed pair to enough
RF, and this applies both to BJT and FET long-tailed pairs, NPN/N-
channel bases/gates look as if they follow the upper envelope of the RF,
while PNP/P-channel bases/gates look as if they follow the lower
envelope.

Bill, the damn things eagerly demodulate AM RF signals. If that isn't a
rectification effect, what is it? A right-wing plot?
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

 

[Bill Sloman]

John Woodgate <jmw@jmwa.demon.contraspam.yuk> wrote in message news:<y$Zpp9In7fVAFwHj@jmwa.demon.co.uk>...

I read in sci.electronics.design that Bill Sloman <bill.sloman@ieee.org
wrote (in <7c584d27.0403151041.38b1fe43@posting.google.com&gt about 'emi
with op-amps', on Mon, 15 Mar 2004:
The other responses talk about "rectification" and suggest that the
diodes associated with the LM358 inputs are acting as rectifiers.

This probably isn't true. If you expose as long-tailed pair to enough
RF, and this applies both to BJT and FET long-tailed pairs, NPN/N-
channel bases/gates look as if they follow the upper envelope of the RF,
while PNP/P-channel bases/gates look as if they follow the lower
envelope.

Bill, the damn things eagerly demodulate AM RF signals. If that isn't a
rectification effect, what is it? A right-wing plot?

It is certainly detection going on and it may look like rectification,
but none of the diodes involved are switching the RF current the OP is
picking up.

You may choose to call it "rectification", but it isn't the
rectification you see in a half-bridge or full-bridge diode rectifier,
and there is no significant current flow through PN junctions - with a
high enough RF level you could see the same effect in a long-tailed
pair built with MOS-FETs where there would be zero current flowing
through the gates,

-----
Bill Sloman, Nijmegen

 

[R.Legg]

John Woodgate <jmw@jmwa.demon.contraspam.yuk> wrote in message news:<y$Zpp9In7fVAFwHj@jmwa.demon.co.uk>...

I read in sci.electronics.design that Bill Sloman <bill.sloman@ieee.org
wrote (in <7c584d27.0403151041.38b1fe43@posting.google.com&gt about 'emi
with op-amps', on Mon, 15 Mar 2004:
The other responses talk about "rectification" and suggest that the
diodes associated with the LM358 inputs are acting as rectifiers.

This probably isn't true. If you expose as long-tailed pair to enough
RF, and this applies both to BJT and FET long-tailed pairs, NPN/N-
channel bases/gates look as if they follow the upper envelope of the RF,
while PNP/P-channel bases/gates look as if they follow the lower
envelope.

Bill, the damn things eagerly demodulate AM RF signals. If that isn't a
rectification effect, what is it? A right-wing plot?

DC operation can be affected without actual rectification taking
place.

As the device output can only respond to a disturbance with low
frequency, simply dithering the input DC over a small range of its
nonlinear range gives an appearance of demodulation on the output,
without any guarantee that the sign or sense will be accurately
preserved, as might be expected in an AM rectifier-demodulator.

You can think of it as rectification, if you want. It's certainly
easier to explain it that way.

RL

 

[Tony]

On 15 Mar 2004 15:08:10 -0800, bill.sloman@ieee.org (Bill Sloman) wrote:

You may choose to call it "rectification", but it isn't the
rectification you see in a half-bridge or full-bridge diode rectifier,
and there is no significant current flow through PN junctions

You're right in principle, but as "active rectifiers" can behave this way too,
the distinction is at best blurred. It's too late to draw a distinction based on
efficiency or whether gain is added, and the use of "envelope detection" is
inappropriate, since most power supply rectifiers do this too.

Tony (remove the "_" to reply by email)

 

[R.Legg]

Bill, the damn things eagerly demodulate AM RF signals. If that isn't a
rectification effect, what is it? A right-wing plot?

Some quick references,

"Susceptibility of Integrated Circuits to RFI"
V. Pozzolo, P. Tenti, F. Fiori, G. Spiazzi, S. Buso,
CPES Annual Seminar'02

http://www.dei.unipd.it/~pel/Articoli/2002/Cpes/Cpes_1.pdf

"Susceptibility of Integrated Circuits to RFI:Analysis of PWM
Current-Mode Controllers for SMPS"
M. Citron, M. Corradin, S. Buso, G. Spiazzi
IECON'02.

http://www.dei.unipd.it/~pel/Articoli/2002/Iecon/Iecon2002_2.pdf

RL

 

[Terry Given]

"R.Legg" <legg@magma.ca> wrote in message
news:e715b5cc.0403151840.22b40103@posting.google.com...

Bill, the damn things eagerly demodulate AM RF signals. If that isn't a
rectification effect, what is it? A right-wing plot?

Some quick references,

"Susceptibility of Integrated Circuits to RFI"
V. Pozzolo, P. Tenti, F. Fiori, G. Spiazzi, S. Buso,
CPES Annual Seminar'02

http://www.dei.unipd.it/~pel/Articoli/2002/Cpes/Cpes_1.pdf

"Susceptibility of Integrated Circuits to RFI:Analysis of PWM
Current-Mode Controllers for SMPS"
M. Citron, M. Corradin, S. Buso, G. Spiazzi
IECON'02.

http://www.dei.unipd.it/~pel/Articoli/2002/Iecon/Iecon2002_2.pdf

RL

Once again, terminological inexactitude rears its ugly head....lol

congratulations gentlemen, a fine series of postings, and all about
electronics (shock, horror).

 

[Winfield Hill]

Bill Sloman wrote...

It is certainly detection going on and it may look like rectification,
but none of the diodes involved are switching the RF current the OP is
picking up.

You may choose to call it "rectification", but it isn't the
rectification you see in a half-bridge or full-bridge diode rectifier,
and there is no significant current flow through PN junctions - with a
high enough RF level you could see the same effect in a long-tailed
pair built with MOS-FETs where there would be zero current flowing
through the gates,

It's a DC component resulting from the inherent non-linearity of the
amplifier. Some modulated RF is presented, and a noisy DC input-
offset voltage, perhaps with some demodulated signal, is the unwanted
result. Surely it's fair to call that process rectification.

Thanks,
- Win

whill_at_picovolt-dot-com

 

[Dave VanHorn]

It's a DC component resulting from the inherent non-linearity of the
amplifier. Some modulated RF is presented, and a noisy DC input-
offset voltage, perhaps with some demodulated signal, is the unwanted
result. Surely it's fair to call that process rectification.

Thanks,
- Win

I don't know if Win will remember, but I had a severe problem with this a
couple of years ago.
We talked about it, and I ended up going with not only a FET op-amp, but a
precision, rail to rail, low offset current (aka "slow as snail snot in
january") op-amp, which solved my problem.

I had to operate, at signal levels of a few mV, within an inch of an antenna
radiating 6W pulses in the 915 MHz band.

Once the right opamp was found, the rest fell in place.

 

[Frank Miles]

In article <Fd6dnc8IkpWzisrdRVn-hQ@comcast.com>,
Dave VanHorn <dvanhorn@cedar.net> wrote:

It's a DC component resulting from the inherent non-linearity of the
amplifier. Some modulated RF is presented, and a noisy DC input-
offset voltage, perhaps with some demodulated signal, is the unwanted
result. Surely it's fair to call that process rectification.

Thanks,
- Win

I don't know if Win will remember, but I had a severe problem with this a
couple of years ago.
We talked about it, and I ended up going with not only a FET op-amp, but a
precision, rail to rail, low offset current (aka "slow as snail snot in
january") op-amp, which solved my problem.

I had to operate, at signal levels of a few mV, within an inch of an antenna
radiating 6W pulses in the 915 MHz band.

Once the right opamp was found, the rest fell in place.

Care to divulge what opamp it was? I had a similar problem, used an opamp
with a similar description (except I'm not sure what 'low offset current'
means with a FET op amp -- perhaps you meant low offset voltage?), and
had fits with offsets, ended up with some ugly shielding around the opamp
in addition to the input filtering. (FWIW: AD854x series -- ok, not
particularly low offset voltage).

-frank
--

 

[Dave VanHorn]

Care to divulge what opamp it was? I had a similar problem, used an opamp
with a similar description (except I'm not sure what 'low offset current'
means with a FET op amp -- perhaps you meant low offset voltage?),

Brain fart. Yes.

and

had fits with offsets, ended up with some ugly shielding around the opamp
in addition to the input filtering. (FWIW: AD854x series -- ok, not
particularly low offset voltage).

One of the national CMOS amps, 6066 or something like that.

 

[Bill Sloman]

Winfield Hill <Winfield_member@newsguy.com> wrote in message news:<c370ms018v8@drn.newsguy.com>...

Bill Sloman wrote...

It is certainly detection going on and it may look like rectification,
but none of the diodes involved are switching the RF current the OP is
picking up.

You may choose to call it "rectification", but it isn't the
rectification you see in a half-bridge or full-bridge diode rectifier,
and there is no significant current flow through PN junctions - with a
high enough RF level you could see the same effect in a long-tailed
pair built with MOS-FETs where there would be zero current flowing
through the gates,

It's a DC component resulting from the inherent non-linearity of the
amplifier. Some modulated RF is presented, and a noisy DC input-
offset voltage, perhaps with some demodulated signal, is the unwanted
result. Surely it's fair to call that process rectification.

Rectification is an odd word. It can also mean correction, repair or
restoration.

Are you aware that distilled spirits (brandy, whisky and so forth) are
also called rectified spirits?

I've always understood electronic rectification to mean separation,
and if the RF current input isn't being switched to separate positive
and negative components, I don't like using the word rectification.
Detection doesn't have the same sort of implication.

------------
Bill Sloman, Nijmegen