Op Amp Filter With Gain

[gmv]

Hello,

A couple questions regarding a relatively simple
home built op amp circuit. I am having a terrible
problem with noise.

All stages are simple RC single order butterworth filters.
The stages are as follows:

0. 50 feet of cable between the 450ohm coil sensor and the amplifier
1. Unity Gain Buffer Amp
2. Unity Gain High Pass Filter FL=160 seconds
3. 28.6dbv Low Pass Filter for Equalization FH=6 seconds
4. 21.7dbv Low pass Filter FH=2.4 Hz
5. 21.7dbv Low Pass Filter FH=2.4 Hz
6. 14.9dbv Low Pass Filter FH=2.4 Hz
7. Unity Gain High Pass Filter FL=160 Seconds Output into 5kohms.

The gains are not even simply because I
have been playing with those and are totally
open for changes. But the frequencies or periods are
solidly locked into place simply because that
is what I need.
The gain can be 83dbv to 90dbv overall.

Using a DC gain of 87dbv in a frequency range
of 100seconds to 1Hz and am having a terrible
problem with noise from sources other than
the desired signal. The op amps used are all OP177G.

[Question #1]
I. Is it important for the + and - inputs
to be resistively balanced to a high
precision & accuracy ?

[Question #2]
II. Major Noise sources seem to be RF, Hot/Cold, Electrostatic
in nature. What is the best way to guard against
all of these things ?

Right now everything is breadboarded and I do
not want to hard wire everything until I can beat
this noise problem.

I would post a diagram but since this site
does not allow binaries...oh well.

GMV

 

[John Woodgate]

[Question #1]
I. Is it important for the + and - inputs
to be resistively balanced to a high
precision & accuracy ?

Leaving aside various strange statements in your article, the short
answer is yes, but it's more complicated than that. Post your schematic
to alt.binaries.schematics.electronic or to a web page and tell us here
that you have done so.

[Question #2]
II. Major Noise sources seem to be RF, Hot/Cold, Electrostatic in
nature. What is the best way to guard against all of these things ?

The only general answer is unhelpful - design it properly. There is a
whole list of things: this is just page 1:

- put it in a grounded metal box;

- use proper screened connectors and ground the screens directly to the
box;

- use a twisted pair for your long cable, preferably a shielded twisted
pair;

- put an RF filter directly on the back of the connector that feeds this
long cable.

 

[Rene Tschaggelar]

Hello,

A couple questions regarding a relatively simple
home built op amp circuit. I am having a terrible
problem with noise.

All stages are simple RC single order butterworth filters.
The stages are as follows:

0. 50 feet of cable between the 450ohm coil sensor and the amplifier
1. Unity Gain Buffer Amp
2. Unity Gain High Pass Filter FL=160 seconds
3. 28.6dbv Low Pass Filter for Equalization FH=6 seconds
4. 21.7dbv Low pass Filter FH=2.4 Hz
5. 21.7dbv Low Pass Filter FH=2.4 Hz
6. 14.9dbv Low Pass Filter FH=2.4 Hz
7. Unity Gain High Pass Filter FL=160 Seconds Output into 5kohms.

The gains are not even simply because I
have been playing with those and are totally
open for changes. But the frequencies or periods are
solidly locked into place simply because that
is what I need.
The gain can be 83dbv to 90dbv overall.

Using a DC gain of 87dbv in a frequency range
of 100seconds to 1Hz and am having a terrible
problem with noise from sources other than
the desired signal. The op amps used are all OP177G.

[Question #1]
I. Is it important for the + and - inputs
to be resistively balanced to a high
precision & accuracy ?

[Question #2]
II. Major Noise sources seem to be RF, Hot/Cold, Electrostatic
in nature. What is the best way to guard against
all of these things ?

Right now everything is breadboarded and I do
not want to hard wire everything until I can beat
this noise problem.

I would post a diagram but since this site
does not allow binaries...oh well.

It would be better to have the amplifier at the sensor.
Nevertheless, the cable to the sensor should be twisted
pair and shielded.
Since your source appears to be low impedance, you could
have the first stage with a gain of 1000.
Those 90dB can be achieved with 2 stages.
The amplifier can be breadboarded, but has to be shielded.


Rene

 

[Ian Bell]

Hello,

A couple questions regarding a relatively simple
home built op amp circuit. I am having a terrible
problem with noise.

All stages are simple RC single order butterworth filters.
The stages are as follows:

0. 50 feet of cable between the 450ohm coil sensor and the amplifier
snip

[Question #2]
II. Major Noise sources seem to be RF, Hot/Cold, Electrostatic
in nature. What is the best way to guard against
all of these things ?

Have you determined if the noise is entering via the sensor and cable?
Remove the cable, terminate the inpout with 450 ohms and check the noise
again.

ian

Ian

 

[gmv]

[Question #1]
I. Is it important for the + and - inputs
to be resistively balanced to a high
precision & accuracy ?

Leaving aside various strange statements in your article, the short
answer is yes, but it's more complicated than that. Post your schematic
to alt.binaries.schematics.electronic or to a web page and tell us here
that you have done so.

The Scematic that has been built on breadboard
is now posted to alt.binaries.schematics.electronic
under subject "Problem:How to eliminate Electrical & Thermal noise"
It is a very simple circuit with no special RFI filtering.
I need to make it proper on breadboard before going all the way
and putting it on a circuit board in a box.
This circuit seems sensitive to every little thing
that might disturb it such as static build up and
thermal changes. It is a proximity detector even though
I want it not to be. Proximity to the electronics.
I can narrow the bandwith X10 and it works ok but
I need the wide bandwidth. The only other alternative I
can think of is to build three different narrow band
circuits and sum them all at the output but that would increase
costs at least 3X expected.

[Question #2]
II. Major Noise sources seem to be RF, Hot/Cold, Electrostatic in
nature. What is the best way to guard against all of these things ?

The only general answer is unhelpful - design it properly. There is a
whole list of things: this is just page 1:

- put it in a grounded metal box;

- use proper screened connectors and ground the screens directly to the
box;

- use a twisted pair for your long cable, preferably a shielded twisted
pair;

- put an RF filter directly on the back of the connector that feeds this
long cable.

--
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

 

[Mac]

[Question #1]
I. Is it important for the + and - inputs
to be resistively balanced to a high
precision & accuracy ?

Leaving aside various strange statements in your article, the short
answer is yes, but it's more complicated than that. Post your schematic
to alt.binaries.schematics.electronic or to a web page and tell us here
that you have done so.

The Scematic that has been built on breadboard
is now posted to alt.binaries.schematics.electronic
under subject "Problem:How to eliminate Electrical & Thermal noise"
It is a very simple circuit with no special RFI filtering.
I need to make it proper on breadboard before going all the way
and putting it on a circuit board in a box.

Well, breadboards sometimes don't work very well. I don't particularly
trust them when high frequencies or high gains are involved. You don't
have high frequencies, but you do have fairly high gains.

If you go through ExpressPCB (http://www.expresspcb.com/) you can get 3
pcb's for $56. They have design software you can download for free. The
software also allows you to place an order with the company.

This circuit seems sensitive to every little thing
that might disturb it such as static build up and
thermal changes. It is a proximity detector even though
I want it not to be. Proximity to the electronics.
I can narrow the bandwith X10 and it works ok but
I need the wide bandwidth. The only other alternative I
can think of is to build three different narrow band
circuits and sum them all at the output but that would increase
costs at least 3X expected.

[snip]

My other suggestion is to simply break your circuit up into components,
and test each component separately to see if it is acting as you expect.
Also try running the circuit with a short cable just to see if that long
cable run is part of the problem.

Good luck


--Mac

 

[John Woodgate]

The only other alternative I can think of is to
build three different narrow band circuits and sum them all at the
output but that would increase costs at least 3X expected.

I'm sure it's possible to solve the problem without doing that. But you
MAY need to go straight to a printed board in a box: it may not be
possible to make a breadboard immune to all the disturbances you list.

 

[Kevin Aylward]

Hello,

The gains are not even simply because I
have been playing with those and are totally
open for changes. But the frequencies or periods are
solidly locked into place simply because that
is what I need.
The gain can be 83dbv to 90dbv overall.

No one seems to have addressed this irritating point, so I will.

Arhhh.... this is finger nails down a blackboard. Its way worse than
using db incorrectly to mean an absolute level.

dbv is *not* gain. Gain is specified simply by db (or dB). It is a
ratio. dbv is a unit used to specify absolute level, not a ratio.


Kevin Aylward
[email protected]
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

http://www.anasoft.co.uk/NewBeginning.mp3

"quotes with no meaning, are meaningless" - Kevin Aylward.

 

[John Woodgate]

I read in sci.electronics.design that Kevin Aylward <kevindotaylwardEXTR
[email protected]> wrote (in <[email protected]
t>) about 'Op Amp Filter With Gain', on Sun, 7 Mar 2004:

No one seems to have addressed this irritating point, so I will.

Arhhh.... this is finger nails down a blackboard. Its way worse than
using db incorrectly to mean an absolute level.

dbv is *not* gain. Gain is specified simply by db (or dB). It is a
ratio. dbv is a unit used to specify absolute level, not a ratio.

Quite right. But if you look at the schematic, you will find that this
point pales into insignificance compared with everything else.

I am deferring any response to the OP pending Fred Bloggs posting his
instant solution. The OP is starting from a LONG way back.

 

[gmv]

No one seems to have addressed this irritating point, so I will.

Arhhh.... this is finger nails down a blackboard. Its way worse than
using db incorrectly to mean an absolute level.

dbv is *not* gain. Gain is specified simply by db (or dB). It is a
ratio. dbv is a unit used to specify absolute level, not a ratio.

Ohkee Dhokee I just wanted somehow to say
it was [20*log] and not [10*Log].
But you still do not say how to handle
noise on a breadboard circuit.
Is this going to be another pissing contest in the snow.
It really irritates me when people get
too precise and accurate when it is not needed.

In the true military tradition:
The only things I NEED-TO-KNOW that are
accurate and precise are how to keep
500 to 900 Mhz radio waves out of my
breadboard circuit and how to keep
electro static in close proximity to
the breadboard from having any effects
upon the class A baseline.

I can see the binaries people clean house
like everyone else should be.

NOTE:
******************************************************
If you can not address Electrical or thermal noise issue
then keep your opinions to yourself.
******************************************************

 

[Ben Bradley]

I looked at you schematic on ABSE. There's no reason for the two
resistors in series with the signal source, and they only add noise.
The first thing I'd do is short them out and see if that lowers the
noise level enough to make it acceptable.
If not, I'd replace the input and most of the gain with a
TI/Burr-Brown in-amp like the INA103 or INA163, using a balanced input
(both sides of the signal source going to the in-amp, each with about
a 10k (matched) resistor to ground for providing bias current), and
balanced, shielded cable. You can get most of your gain (60dB) or
perhaps all of it in the in-amp, then run it through a multipole LPF
from the Active Filter Cookbook. Pick the four-pole filter, plug in
the gain you want, and put in the appropriate values.
Run your electrolytic capacitor ground connections to the V- supply
(well filtered and regulated, of course) so they will be biased and
not dry up and lose capacitance because of a lack of voltage on them.
You perhaps can't do this directly with the active filter circuit, but
there's a way around it.

NOTE:
******************************************************
If you can not address Electrical or thermal noise issue
then keep your opinions to yourself.
******************************************************

Are you trying to tell others what and how to post on Usenet? This
is a futile activity.

 

[Ian Bell]

gmv wrote:

snip

NOTE:
******************************************************
If you can not address Electrical or thermal noise issue
then keep your opinions to yourself.
******************************************************

I suggested you determine if the noise is entering thru the sensor and/or
cable. Did you do this?

Ian

 

[Rene Tschaggelar]

No one seems to have addressed this irritating point, so I will.

Arhhh.... this is finger nails down a blackboard. Its way worse than
using db incorrectly to mean an absolute level.

dbv is *not* gain. Gain is specified simply by db (or dB). It is a
ratio. dbv is a unit used to specify absolute level, not a ratio.

Ohkee Dhokee I just wanted somehow to say
it was [20*log] and not [10*Log].
But you still do not say how to handle
noise on a breadboard circuit.
Is this going to be another pissing contest in the snow.
It really irritates me when people get
too precise and accurate when it is not needed.

In the true military tradition:
The only things I NEED-TO-KNOW that are
accurate and precise are how to keep
500 to 900 Mhz radio waves out of my
breadboard circuit and how to keep
electro static in close proximity to
the breadboard from having any effects
upon the class A baseline.

It was mentioned.
A shielded twisted pair cable.
A shielded box with Ferrite through hole filters.

Rene