Discussion in 'Electronic Basics' started by seware, Jun 6, 2005.

1. ### sewareGuest

I'm fairly new to electronics and I am having a bit of a problem
understanding which path to take. I am needing to construct a receiver stage
that accepts a differential input signal. So I have been reading about and
planning to use an op-amp wired as a difference amplifier. One of my
questions is in powering the op-amp... and also understanding enough about
different op-amps to make a good selection. I have +12VDC power and am
uncertain whether I should use a voltage inverter IC, construct a stage to
produce -6 and +6 or what. Then I started reading about single-supply
opamps, but I haven't found anything that fully explains them (at least
enough to make a choice). I know what "rails" are but what is the difference
in "rail-to-rail", "not rail-to-rail" and "outside the rails" which are some
terms I found on various chip makers sites. The last part of my confusion
come from that fact that I intentionally want to clip any negative portions
of the output of the opamp. One place I read said that I could use ground as
the negative supply to the opamp to accomplish this, but another book I have
said that if the input signals go outside the limits of the opamp power
supply, the opamp can lock-up. Which is it and where can I go for a clear
understanding? Thanks all and sorry for rambling.

2. ### BanGuest

Even if you write all that much, there are many points open which permit to
First the input: where does the input come from? Impedance of the driver
stages. What kind of signal? frequency and voltage range, how much common
mode signal to be suppressed? DC accuracy needed, amplification needed?
Then the output: What is the next stage(impedance), what range of voltage,
do you need swing to gnd or have a reference voltage available?
Then the desired function: how precise has to be the rectification and the
clamping of the negative portion?
These are not theoretical questions, but each refer to specific circuit
constallations.
I even doubt a differential amplifier will do the job, rather look for an
"Instrumentation Amplifier".

3. ### John PopelishGuest

I'll put off giving you advice about the application till we clear up
some of the opamp generalities.
All these terms relate the voltage range of input and output signals
to the supply voltages (rails). All opamps have specific limits on
what the input voltages (there are two input terminals) can be and
still have the input pair function as a differential amplifier. This
is called the common mode voltage (the voltage the two inputs have in
common while the differential amplifier amplifies their difference)
range. There is also a different input voltage range (related to the
supply rails), called absolute maximum voltage limits, that prevents
damage to the device. The opamp does not actually care if its rails
are on one side of signal common or straddle it, as long as the
applied input voltages stay within the absolute limits. It will
function as intended as long as the input voltages stay within the
common mode voltage range, with respect to its supply rail voltages.

Ordinary opamps have a working common mode voltage that reaches to
within a few volts of either supply rail voltage (regardless of what
those voltages are with respect to signal common). Single supply
types typically work from at, or slightly below, the negative rail
voltage to within a few volts of the positive supply rail.
Rail-to-rail types have a common mode input range that equals or
slightly exceeds the voltage range between the rails.

The output of an opamp is also limited in its voltage swing by the
voltages applied to the supply pins. Ordinary (not rail to rail
types) typically can swing their outputs within a few volts of either
supply rail. Those designated as single supply types can swing their
output voltage to within millivolts of the negative rail under some
load conditions, but still swing to within only a few volts of the
positive rail. Rail-to-rail types can swing their output voltage to
very nearly either supply rail (under some load conditions).

So you need to explore your range of input signal voltages and what
you can do to bias them in different ways, and your requirement for
output voltage swing, before you are ready to try to fit those
requirements to a combination of opamp types and available (or
practical) supply voltages.
This is a separate problem until you decide how this clipping is to be
done. The two general ways involve saturating the output (which may
have recovery problems and input over voltage problems, when the
negative feedback opens up during saturation) and using some nonlinear
feedback that abruptly changes the gain as the voltage passes through
some particular voltage, but the opamp continues to function as a
closed loop process the whole time.
A definite possibility for some opamp designs.
Start with the signals and work out to what you need to do with them,
to set some boundaries on the problem. Try to keep as much
flexibility in this part of the process as you can imagine, to not
rule out some good solutions before they are explored. Then you are
in a position to begin considering which of the three general kinds of
opamps and their required supplys fit each way the signals could be
processed.

4. ### ChrisGuest

Hi, Seware. I think I understand what you want, although your
description is a little hazy. You have a single 12VDC supply and two
input signals, both of which go below GND. You want to use an op amp
to amplify the difference between the two signals, but you only want
the positive excursions of the output signal -- in other words, you
want the output to act like a rectifier of the difference between the
two signals, with unspecified gain.

I don't think there are any rail-to-rail op amps which can amplify the
difference of two signals, both of which are below the negative rail of
the single supply. One thing you might do is use an op amp which has
an output which can go down to the negative rail with a pulldown, and
do something like this (view in fixed font or M\$ Notepad):

` VCC
` +
` | ___ ___
` '-|___|-o-|___|-.
` R1 | R2 |
` .--------' ===
` | GND___
` | ___ .--|___|--.
` o--|___|---. | Rf |
` | | | VCC |
` V1 | ___ | | + |
` o---|--|___|---o | |\| |
` | '---o--|-\ | Vout
` | | >---o----o
` V2 | ___ .---o--|+/ |
` o---|--|___|---o | |/| .-.
` | | | | | |Rp
` | ___ | | === | |
` '--|___|---' .-. GND '-'
` | | |
` | |Rf |
` '-' ===
` | GND
` |
` ===
` GND
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

This looks a little like the difference amp you know from school.
There are a couple of changes. The voltage divider R1 and R2 sets up a
DC value which should bring your input signal up out of the weeds so
it's in play. You sum that DC level with your input voltage to provide
a composite input your difference amplifier can work on. Note that
you're going to have to get out your DC electronics book to figure out
appropriate resistor values to get a composite Ri at a DC voltage level
that's useful to you. Remember also that, to keep it simple, Ri and Rf
for V1 should equal Ri and Rf for V2. Then you can use your standard
difference amplifier equation (if you don't remember, see National
Semiconductor AN31 for op amp basics).

The only other quirk is Rp, which is a pulldown necessary for most op
amps that go down to the negative rail (or within a couple of
millivolts).

This difference amplifier will obviously saurate at negative output
(0V) as V1 exceeds V2, which I guess is what you're looking for. And
you won't have to worry about amplifying inputs below the negative
specification and expense as your problem demands.

If this doesn't fill the bill, or if you want more help, you'll have to
throw a little more light on your problem:

* What are the maximum and minimum frequencies of interest? You've
got me half thinking this is RF the way you phrased this.

* What's the maximum signal amplitude? If your signal is DC or low
frequency AC, what's the maximum and minimum excursion of the voltage?

* What's the output impedance of your signal source? Can your signal
source source and sink current?

* Is your 12VDC supply floating, or is it referenced to the input
signals?

* What's at the output of the gain block you want? What's the load?

* What's an acceptable level of error or distortion?

Hope we've been of service, Guv'nor.

Good luck
Chris

5. ### ChrisGuest

VCC
+
| ___ ___
'-|___|-o-|___|-.
R1 | R2 |
.--------' ===
| GND___
| ___ .--|___|--.
'--|___|---. | Rf |
| | VCC |
V1 ___ | | + |
o------|___|---o | |\| |
'---o--|-\ | Vout
| >---o----o
V2 ___ .---o--|+/ |
o------|___|---o | |/| .-.
| | | | |Rp
___ | | === | |
.--|___|---' .-. GND '-'
| | | |
| | |Rf |
| '-' ===
'--------. | GND
| |
| ===
| GND
VCC |
+ |
| ___ | ___
'-|___|-o-|___|-.
R1 R2 |
===
GND
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

And, as inevitable when posting before the second cup of coffee, a
minor revision. Use two identical voltage dividers R1 and R2 for
_each_ input as above, instead of sharing. Other than that, OK, I
guess. Sorry for the confusion.

Good luck
Chris

6. ### ChrisGuest

` VCC
` +
` | ___ ___
` '-|___|-o-|___|-.
` R1 | R2 |
` .--------' ===
` | GND___
` | ___ .--|___|--.
` '--|___|---. | Rf |
` | | VCC |
` V1 ___ | | + |
` o------|___|---o | |\| |
` '---o--|-\ | Vout
` | >---o----o
` V2 ___ .---o--|+/ |
` o------|___|---o | |/| .-.
` | | | | |Rp
` ___ | | === | |
` .--|___|---' .-. GND '-'
` | | | |
` | | |Rf |
` | '-' ===
` '--------. | GND
` | |
` | ===
` | GND
` VCC |
` + |
` | ___ | ___
` '-|___|-o-|___|-.
` R1 R2 |
` ===
` GND
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

And, as inevitable when posting before the second cup of coffee, a
minor revision. Use two identical voltage dividers R1 and R2 for
_each_ input as above, instead of sharing. Other than that, OK, I
guess. Sorry for the confusion.

Good luck
Chris

7. ### ChrisGuest

` VCC
` +
` | ___ ___
` '-|___|-o-|___|-.
` R1 | R2 |
` .--------' ===
` | GND___
` | ___ .--|___|--.
` '--|___|---. | Rf |
` | | VCC |
` V1 ___ | | + |
` o------|___|---o | |\| |
` '---o--|-\ | Vout
` | >---o----o
` V2 ___ .---o--|+/ |
` o------|___|---o | |/| .-.
` | | | | |Rp
` ___ | | === | |
` .--|___|---' .-. GND '-'
` | | | |
` | | |Rf |
` | '-' ===
` '--------. | GND
` | |
` | ===
` | GND
` VCC |
` + |
` | ___ | ___
` '-|___|-o-|___|-.
` R1 R2 |
` ===
` GND
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

And, as inevitable when posting before the second cup of coffee, a
minor revision. Use two identical voltage dividers R1 and R2 for
_each_ input as above, instead of sharing. Other than that, OK, I
guess. Sorry for the confusion.

Good luck
Chris

8. ### SteveGuest

Sorry for the lack of info. I'll try better.
Basically I have a two-wire inductive pickup that is used for RPM count of a
motor shaft. The AC voltage on the two wires are of nearly equal magnitudes
and opposite polarity. Differential signal right? The voltage range I see on
my oscope for each wire is +- 8 V. Freq ranges from 0 - 300 Hz (18000RPM).
I'm not sure what the impedance of the pickup is, but I think I have a
datasheet on it at home. I currently have available a 12VDC battery and a
buck-boost circuit that outputs 5VDC. So my desired output of this stage is
a TTL voltage edge on the positive pulses of the composite signal. My use of
op amps is limited outside of simple signal amplifying... so reading all
that I could find, I surmised that using a differential op amp setup would
allow me to create a composite signal that I could then rectify to
eliminate the negative swings. I just don't know enough about op amps to
make an educated guess as to how best to power the thing, what type of opamp
to use, etc. Do I require pre-filtering for transients or this what CMR is
for? I'm sure there are circuits available out in www-land that I could
study and learn from but I have not found exactly what I am looking for yet.

9. ### ChrisGuest

Sorry - it looks like the diagram got munged -- Google burped. Let's
try again:

` VCC
` +
` | ___ ___
` '-|___|-o-|___|-.
` R1 | R2 |
` .--------' ===
` | GND___
` | ___ .--|___|--.
` '--|___|---. | Rf |
` | | VCC |
` V1 ___ | | + |
` o------|___|---o | |\| |
` '---o--|-\ | Vout
` | >---o----o
` V2 ___ .---o--|+/ |
` o------|___|---o | |/| .-.
` | | | | |Rp
` ___ | | === | |
` .--|___|---' .-. GND '-'
` | | | |
` | | |Rf |
` | '-' ===
` '--------. | GND
` | |
` | ===
` | GND
` VCC |
` + |
` | ___ | ___
` '-|___|-o-|___|-.
` R1 R2 |
` ===
` GND
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

Or actually, if you're using a dual or quad op amp, it might be better
to feed the voltage divider into one of the op amps set up as a voltage
follower and save the two resistors.

Good luck
Chris

10. ### sewareGuest

Sorry for the lack of info. I'll try better.
Basically I have a two-wire inductive pickup that is used for RPM count of a
motor shaft. The AC voltage on the two wires are of nearly equal magnitudes
and opposite polarity. Differential signal right? The voltage range I see on
my oscope for each wire is +- 8 V. Freq ranges from 0 - 300 Hz (18000RPM).
Pickup resistance is155 ohms +- 20%. I currently have available a 12VDC
battery and a buck-boost circuit that outputs 5VDC. So my desired output of
this stage is a TTL voltage edge on the positive pulses of the composite
signal. My use of op amps is limited outside of simple signal amplifying
(again an elec. newbie)... so reading all that I could find, I surmised that
using a differential op amp setup would allow me to create a composite
signal that I could then rectify to eliminate the negative swings. I just
don't know enough about op amps to make an educated guess as to how best to
power the thing, what type of opamp to use, etc. Do I require pre-filtering
for transients or this what CMR is for? I'm sure there are circuits
available out in www-land that I could study and learn from but I have not
and thank you for your time.

Steve

11. ### John PopelishGuest

This information is very helpful. First of all, if you check the
leads from the pickup to the case with an ohm meter, I think you will
find that the coil is isolated from ground, so you can connect the
pair of wires with freedom, as far as ground is concerned. This kind
of signal source tends to produce an increasing voltage with
increasing frequency, but also to produce a fairly constant current
into a short circuit, over the operating frequency range.

I suggest that you connect a pair of signal diodes (i.e. 1N4148)
across the pickup leads, to limit the peak voltage to +- 1 volt,
regardless of frequency. Bias one input of a single supply opamp
(i.e. one half of an LM358 http://www.national.com/ds/LM/LM158.pdf or
1/4 of an LM324 http://www.national.com/ds/LM/LM124.pdf )
to about 2 volts above the negative rail with a voltage divider or a
resistor in series with two diodes. Connect the diode clamped pickup
pair to the two inputs. The bias for the second input passes through
the pickup. You may want to add a capacitor to ground from the
divider biased input, to soak up noise spikes that may couple into the
pickup from other noise sources.

This means that one input stays around +2 volts and the other swings
from +1 to +3 volts, well within the common mode voltage range for a

The output will be a TTL compatible signal that is a square wave
(limited by the slew rate of voltage change of the opamp output of
about .3 volts per microsecond) representation of the pickup
frequency. Note that6 on page 11 of the second data sheet, they
recommend a 240 ohm resistor to ground on the output for best TTL
compatibility, but you may not need this.

If this circuit produces too much noise at zero frequency, since it
has a gain of something like 1 million, there, you may want to add
some positive feedback to lock the output up once the input falls
below some minimum frequency. To do this, you add a series resistor
to the + input, and a feedback resistor as shown at the top of page 13
of the second data sheet. The ratio of the two resistors will
determine the minimum operating frequency.

12. ### sewareGuest

Thank you very much John.

13. ### Rich GriseGuest

You're not looking for an opamp at all. You already have 8Vp-p. If your
sensor has only two wires, and neither of these are connected to the case
ground, then you can simply ground one of them, clamp the other to the
+5V rail and to ground with a couple of diodes, and drive the input of a
schmitt trigger inverter, such as the 74HC14:

+5V
|
V DIODE (e.g., 1N4148)
--- cathode
------- | |\
| |--------+----+-----------| >o------- Out
| sensor| | |/
| |---, V 1/6 74HC14
------- | ---
| |
GND GND

This is assuming, of course, that your sensor is simply an inductor,
and there's a magnet on the shaft.

The output will be "better than" TTL levels - HC parts give a
practically rail-to-rail output.

Have Fun!
Rich

14. ### JeffMGuest

seware wrote [witout context]:
I notice that you're relatively new to Usenet.
I would have sworn that you post from Google Groups--but nope.

Folks on Usenet like to see a bit of the previous post
included with (actually, above) the text of your post.
This is called **context**.
If you observe how most folks post, you'll see what I mean.

Most newsreaders blockquote the entire previous post for you
so that you can easily snip out those portions which don't apply,

Rich Grise (pointedly) remarks on context here:
you can easily see why I figured that you posted from there.

The guidelines for Usenet are here:
http://66.102.7.104/search?q=cache:...gh+at-*-top-*-*-message+do-not-*-*-*-original
(Worth scanning--especially the parts I've highlighted.)