Here we go again: the black magic of grounding.

Thread Starter

bararu

Joined Jul 27, 2008
4
Hello all, this is my first post here and I swear that I used a search button and read a whole bunch of articles regarding grounding issues.

Background:

I'm trying to measure an analog signal between 20 and 30 kHz into a PC via Digital Acquisition(DAQ) Board. Before A/D conversion signal is filtered and amplified through my home-built filter/amplifier board. This filter/amplifier board is largely based on 741 OpAmps and requires +12 and -12 voltages to operate.

Problem description:

Once the filter/amp board is connected to laboratory (earth grounded) power supply - everything works fine. However, it is supposed to be powered by two 12Volt batteries and be a mobile system, then when it is connected to two 12V batteries filter board's output becomes noisy and drops down significantly. What happens next is very puzzling for me: When I connect "Common ground" between batteries to "Earth ground" from 110VAC wall plug, everything becomes normal.


Can I somehow improve my filter output without connecting it to wall earth ground??? It has to be a mobile solution.

Here's AC-ground output:


Here's "wrong" battery-ground output:




Here's the setup:
 

SgtWookie

Joined Jul 17, 2007
22,210
Connect the chassis ground of the computer to the "virtual ground" between the two 12v batteries using a good-sized conductor.

It's likely that you have a ground path from the computer through the filter board and then eventually to the virtual ground between the batteries. This will cause a great deal of noise.

Ditch the 741 opamps. Look at TL071's, LF351's and the like. 741's are ancient, and have many drawbacks. There are far better opamps available for mere pennies more. Even the TL071s are ancient, but will have far broader bandwidth and far lower noise than the 741's.

Now if you want a really decent opamp, look at something like Linear Technology's LT1007's.
 

Thread Starter

bararu

Joined Jul 27, 2008
4
Connect the chassis ground of the computer to the "virtual ground" between the two 12v batteries using a good-sized conductor.
Good point, tried that. Made absolutely no difference.


Ditch the 741 opamps.
I wish I could, this board was designed and made by someone else. This thing has to work in five days, so I really got no time to rebuild it.:mad:
 

Wendy

Joined Mar 24, 2008
22,146
The op amps Wookie suggests are drop in replacements, which makes that part easier. Can you get the parts? This thread discusses the alternatives.

It strikes me as your signal source could have some significance. Could you post the complete schematics?
 

John Luciani

Joined Apr 3, 2007
477
Without a schematic and board layout it will be difficult to help troubleshoot
this problem.

I would start by measuring the noise on the power supplies. Solder some testpoints
into onto your filter board, remove the ground lead of your scope probe and rest the
probe against the testpoints. For a picture of scope probe placement see http://www.luciani.org/eng-notes/ee-notes/ee-notes-index.html

I would check your amplifier grounding and decoupling. A good article is

"An IC Amplifier User’s Guide to Decoupling, Grounding,
and Making Things Go Right for a Change" by Paul Brokaw

http://www.analog.com/static/imported-files/application_notes/135208865AN-202.pdf


(* jcl *)
 

studiot

Joined Nov 9, 2007
4,998
Your displays go up to 70Khz, though you mention 30 Khz.

Either way the 741 will be running out of gain steam at these frequencies.

Not sure what you input/output , gain and gain margin levels are like for your filter, but is it up to the job?

Earths in computers are usually noisy at millivolt signal levels because the data signals in the pc are in the region of a few volts. Also the psu is a SMPS flinging around some hefty currents.
 

Thread Starter

bararu

Joined Jul 27, 2008
4
Your displays go up to 70Khz, though you mention 30 Khz.
It is really just a MATLAB scale that goes up to 70kHz, signal range is between 20kHz and 30kHz.


It strikes me as your signal source could have some significance. Could you post the complete schematics?
Absolutely, here's complete schematics for a single channel:



Not sure what you input/output , gain and gain margin levels are like for your filter, but is it up to the job?
The input signal range is 20mV peak to peak, I'm amplifying it to 1V peak to peak which roughly gives us Gain=50; However, I must mention that I'm varying gain through a simple potentiometer.


What completely puzzles me is the fact that I have good results with "Earth" ground but not with "Floating ground".
 

Thread Starter

bararu

Joined Jul 27, 2008
4
Without a schematic and board layout it will be difficult to help troubleshoot
this problem.

I would start by measuring the noise on the power supplies. Solder some testpoints
into onto your filter board, remove the ground lead of your scope probe and rest the
probe against the testpoints. For a picture of scope probe placement see http://www.luciani.org/eng-notes/ee-notes/ee-notes-index.html

I would check your amplifier grounding and decoupling. A good article is

"An IC Amplifier User’s Guide to Decoupling, Grounding,
and Making Things Go Right for a Change" by Paul Brokaw

http://www.analog.com/static/imported-files/application_notes/135208865AN-202.pdf


(* jcl *)

John, you made many good points, and indeed our laboratory power supply had some noise on its ground at 50Hz that we attributed to normal AC noise, however it does not really interfere with our circuit.

My issue is battery power which has no measurable noise whatsoever (Tektronics 350Mhz Scope).

Thanks for a very good article.
 

John Luciani

Joined Apr 3, 2007
477
Have you measured the power supply noise at the op-amps?

Also I do not see any decoupling caps on any of the op-amps.
The ADI article gives hints on how to decouple a variety of op-amps.

(* jcl *)
 
Top