Summing analog signals

Thu Mar 11, 2010 1:44 am   

Can anyone give me some pointers for mixing DC-20KHz analog signals
from different sources? For example, generated noise, sinewaves, etc.

I have read about it and tried various circuits, resistor networks, a
summing op amp, but it does not appear so straightforward. Adding a
signal causes a voltage drop in the sum, etc.

As a novice, there is obviously something I am overlooking. Any
adfvice would be appreciated.

Is there a sure-fire cirucit I can use for everything in this type of
application?

Richard Holmes

Thu Mar 11, 2010 2:07 am   

"Richard Holmes"


** Adding a signal source affects a resistor summing network - just adding
a signal does not.

A "summing" op-amp is not affected by either.

Eg: http://www.ecircuitcenter.com/Circuits/opsum/opsum.htm



...... Phil

Thu Mar 11, 2010 3:05 am   

On Thu, 11 Mar 2010 00:44:06 GMT, richardholmes_at_trentcor.com (Richard
Holmes) wrote:


---
Yes, use an inverting opamp as a current to voltage converter: (View in
Courier)

V+
+----|----+
| | [R]
IN1>---[R1]--+-+---|-\ |
| | >--+-->
IN2>---[R2]--+ +--|+/
. | |
. GND V-
.
|
INn>---[Rn]--+


JF

Thu Mar 11, 2010 3:58 am   

Richard Holmes wrote:

S1--R1---+
S2--R2---+----Vo
S3--R3---+

Where + is all connected is the basic idea. The problem is that you cannot
draw any current from Vo without distorting the results. This is called
loading the output. You can, for example, connect a scope probe to Vo and
expect to see S1 + S2 + S3.

Now because we don't want to "load the output" we have to one of any number
of methods to allow us to effectively load the output(i.e., still gives us
Vo but let us draw a current from it without changing Vo).

To do this you must use some type of buffering/amplification technique. A
simple op amp buffer will work but will not provide amplification. You'll
essentially get Vo out of the buffer but be able to drive larger loads than
the resistor network will.

http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/opampvar.html

The op amp buffer is at:
http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/opampvar2.html#c1

So you connect Vo to Vin of the buffer and then Vo of the buffer is your Vo
but you can now drive larger loads(up to the op amps spec).

You can also implement buffers and amplifiers with bjt's and fets with
effectively the same result(op amps are more precise).

Here is what you are trying to accomplish:
http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/opampvar5.html#c1


Make sure you get the resistor values right. If you have


S1--R1---+
S2--R2---+----Vo
S3--R3---+

Then the sum is

(R1*S1 + R2*S2 + S3*R3)/(R1 + R2 + R3)

which is known as a weighted sum. Generally you'll want R1=R2=R3. Also, The
R's should not be too large. Something in the 10k's should work. Nothing
below 1k in general.

Thu Mar 11, 2010 6:03 am   

On Wed, 10 Mar 2010 20:05:21 -0600, John Fields
<jfields_at_austininstruments.com> wrote:


It should be noted that the output signal will be the sum of the INn (where
'n' is the number of the input) times the ratio of R/Rn. It's a very powerful
circuit.

Thu Mar 11, 2010 7:20 am   

Richard Holmes wrote:

you can add more inputs without any effect at the output o the other
signals - ASSUMING the op-amp is in the linear range AND the frequencies
aer well within the gain-bandwidth.

Thu Mar 11, 2010 7:26 am   

On Wed, 10 Mar 2010 20:05:21 -0600, John Fields
<jfields_at_austininstruments.com> wrote:


So, assuming two input signals of 1Vpp and 2Vpp, and a single supply
op amp, the following circuit will work?

CA 3130

+12V
+-----|----+
| | |
| | [Trim pot] gain

[R2] |
|___|
|
GND
Not shown:

1. R1 between +12V and non-invert IP.
2. 120pf cap pins 1 + 7

Sorry, I don't have the ASCII circuit program.

Many thanks,

Richard Holmes

Thu Mar 11, 2010 7:54 am   

"Richard Holmes"


** I see your pain.

A summing amp is inherently an * inverting stage* - ie positive inputs
generate negative outputs.

If the input signals are DC coupled, ground referenced and swing positive or
both ways - then you MUST use a suitable op-amp with dual supplies.

The schem you posted will only work with AC coupled input signals - ie there
is a cap between the sources and each input and the ouput will have a +6
volt fixed bias.



..... Phil

Thu Mar 11, 2010 2:22 pm   

On Thu, 11 Mar 2010 06:26:35 GMT, richardholmes_at_trentcor.com (Richard
Holmes) wrote:



One of the most often used is over at
http://www.tech-chat.de/aacircuit.html

Hasn't been updated in a while but it's quite useful.

--
Rich Webb Norfolk, VA

Thu Mar 11, 2010 2:49 pm   

Richard Holmes schrieb:


Hello,

using bipolar input signals with a single supply op amp is not a good
idea, especially if you want bipolar output signals too.

Bye

Thu Mar 11, 2010 9:24 pm   

On Thu, 11 Mar 2010 14:49:16 +0100, Uwe Hercksen
<hercksen_at_mew.uni-erlangen.de> wrote:


How about DC offsetting the input signals?

Richard Holmes

Fri Mar 12, 2010 12:50 pm   

"Richard Holmes"



** You will have to offset them in the negative direction.

Do read what has been posted.



...... Phil

Fri Mar 12, 2010 6:41 pm   

On Mar 12, 6:50 am, "Phil Allison" <phi...@tpg.com.au> wrote:

Hmm, probably a silly idea with a mistake. But if you could reference
the signals to some voltage half way between the supply voltages (call
it common). You could then use the summing op-amp circuit. Then
invert back about the common point and remove the DC offset at the
end... sounds like a lot of trouble just to add two voltages with
single supply opamp.

George H.