Re: Hardening a simple op-amp circuit against EMI

bill.sloman@xxxxxxxx wrote:
On Mar 17, 7:36 pm, Joerg <notthisjoerg...@xxxxxxxxxxxxxxxxxxxxx>
wrote:
D from BC wrote:
On Tue, 17 Mar 2009 08:52:33 -0700 (PDT), Russell Warren
<russandheat...@xxxxxxxxx> wrote:
I've got a simple op-amp circuit that is misbehaving in an EMI
environment. It looks like below (hopefully the ascii drawings work
out)...
___10k_______
| |
GND--------|\ 20k
| \____|___
--100----------|+/
|/
The numbers are resistor values and the + input comes from a unity
gain buffer.
On first inspection my first obvious thought was to limit the
bandwidth to avoid EMI issues (circuit only actually needs kHz'ish
bandwidth) so I put a simple RC filter on the input as follows:
___10k_______
| |
GND--------|\ 20k
| \____|___
--100----------|+/
| |/
1u =
|
GND
But this still had EMI issues and was probably bad because hanging a 1
uF cap off the input is a bad idea? Not that I have any fundamental
basis for making that statement.
Consulting an EMC design book which has been *extremely* helpful in
the past (EMC for Product Designers by Tim Williams), one of the
things it recommends is putting low value resistors at each of the
inputs, right at the pins. It also indicates that the values should
be no larger than a few hundred Ohms. ie: adding 200 Ohms to each pin
of the circuit above:
___10k_______
| |
GND----200-|\ 20k
| \____|___
--100------200-|+/
| |/
1u =
|
GND
I'd like to understand what is going on here. My questions are:
1. Is hanging a large capacitance on the input of an op-amp a bad
idea? If so, why? I saw some rumblings somewhere about CMRR issues,
but don't get it.
2. With the op-amp having a high input impedance, what is the reason
for limiting the resistors at the input pins to "up to a few hundred
Ohms"?
I'm trying to simulate the impact of it all in Switchercad, but I'm a
hack and my attempts have not been that illuminating. I've spent some
time trying to work out how to simulate and understand when an
amplifier circuit is stable or not, but haven't been able to fully
grasp phase margin and certainly not how to simulate it properly.
Any help is much appreciated! I have an extreme dislike for doing
anything because "that's just how you do it" and want to understand
what is going on underneath.
Russ
Is is conducted noise, capacitively coupled noise, RF pickup noise or
magnetic induction noise?
Ok, Russell, I won't be able to see your posts and replies because you
use gmail (maybe use a better address?). But I want to alert you to
something that is rarely known in the trade:

Opamps with bipolar transistors in the input can cause RF to be
rectified. A cell phone transmitting in the GHz range can cause that
trouble even with an opamp that has only 10MHz or so GBW. This is
because the BE junctions at IN+ and IN- rectify RF and that rectified
signal shows up in your signal. With GSM you'd have pretty hard
"rat-tat-tat" style bursts when the phone connects to a new tower or
checks in once in a while. CMOS opamps suffer much less from this
behavior because there is no BE junction that is slightly conducting,
there are usually only substrate diodes that are sufficiently reverse
biased (as long as your input swings stay withing the rails).

With strong EMI none of the external methods such as ferrite beads and
capacitors help much. I had one such case where the EMI went straight
through the SO-8 plastic smack dab into the chip. So I recommended a
CMOS opamp to that client or, as an alternative, shielding. We then
tried both and both fixed it. Since they really liked the BJT-based amp
for audio noise level reasons they opted for shielding.

You can express this more formally, by pointing that the long-tailed
pair at the input of a bipolar op amp goes non-linear if the voltage
between the inputs goes above about 26mV.

At low frequencies - within the op amps bandwidth - negative feedback
stop this from happening, but at higher frequencies, there isn't any
feedback.

The long-tailed pair on the input of a FET- or MOSFET-input op amp
goes non-linear in the same way, but only at signal levels around a
volt.

It explains the rectifier action ...


Happens much earlier. Some hundred microvolts from a cell phone and your low noise pick-up circuit is toast. Rectification happens the instant RF hits a forward biased diode path such as a BE junction of a properly biased BJT amp stage, whether diff or not. Ye olde LM324 and his brethren can't regulate that away via feedback.

--
Regards, Joerg

http://www.analogconsultants.com/

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

Relevant Pages

  • Re: Hardening a simple op-amp circuit against EMI
    ... Is is conducted noise, capacitively coupled noise, RF pickup noise or ... because the BE junctions at IN+ and IN- rectify RF and that rectified ... Since they really liked the BJT-based amp ... At low frequencies - within the op amps bandwidth - negative feedback ...
    (sci.electronics.design)
  • Re: Hardening a simple op-amp circuit against EMI
    ... bandwidth to avoid EMI issues (circuit only actually needs kHz'ish ... for limiting the resistors at the input pins to "up to a few hundred ... Is is conducted noise, capacitively coupled noise, RF pickup noise or ... Since they really liked the BJT-based amp ...
    (sci.electronics.design)
  • Re: Troubleshooting experience and intuition
    ... >> 15-20 amp automotive fuses in people's amps. ... It's often the elemental things like that (fuse check) ... that was the end of the noise - duh. ... plate resistors, but I ignored that since the ...
    (alt.guitar.amps)
  • Re: Amp noise Pt. 2 (w/pic)
    ... Static noise that sounds like the ocean with some popping and hissing ... starts about 5 minutes after amp is on. ... Noise responds to volume knob ... noise commonly associated with resistors. ...
    (alt.guitar.amps)
  • Re: Which tube expert can do custom works?
    ... OK, I think I follow you this far, for the case of the mic pre example you are discussing, although I don't think Henry Pasternack would care much for your use of the term "output noise". ... For example if we had a mic pre with 10 dB less gain, and 10 dB less "output noise", it isn't obvious to me that we couldn't mix at a level of -10 dBu and still achieve the desired 96 dB S/N. ... Let's say we have 10 channels to mix together and we use voltage mixing as described earlier by Graham and for the sake of argument say we use 10K bus resistors. ... One channel is fed via a 10K resistor and the other 9 off 10K resistors load the end of this 10K resistor so the bus signal from this channel is 10 times lower than the mic amp output so the mix amp needs 20dB of gain to make up the mix bus loss. ...
    (rec.audio.tubes)