I have an application in which I am producing a millivolt signal as a linear value in proportion with the capacitance of a small probe. I am getting a stable signal from the probe, however the range I am monitoring is slight. My voltage output varies from 200 mv to 230mv depending upon the condition of the probe. This behavior is exactly what I want, however, now I need provide this signal as an input to a PLC. I think I can amplify the signal with an external op-amp circuit, however if I amplify into roughly a 10VDC maximum signal, my span is still pretty small. Does anyone know of a way to amplify the span of a signal, basically create a non-linear analog which could be produced through circuitry rather than a processor? I would appreciate any help I can get.

Thanks in advance

 

if the signal is fast changing, it is easy to deal with: you can either dc block it, or use a servo to null it.

if the signal can be dc, it is tougher. I would use a chopper amplifier to null the dc. do a google on that and you will see.

 

What is the desired voltage range for the PLC input signal?

 

Indeed! What you are looking for is a LogAmp. In the general configuration a Log Amp COMPRESSES the signal, but if you use one in the feedback loop of a conventional amplifier, you have an EXPANSION AMPLIFIER....just what you need.


Analod Devices makes a variety of logamps, or you can roll your own with conventional amps with a transistor feedback.

Eric

 

http://www.electronics.dit.ie/staff/ypanarin/Lecture Notes/DT021-4/6LogAntiLogAmplifiers.pdf


Excellent material on logamps n such

eric

 

Can't you input the signal into an opamp and reference the other input to 200 mV? Then set the gain to get the output range you want.

 

An instrumentation amplifier (IA) will amplify the difference.
The IA is laser trimmed to reduce offset, gain and common mode errors.

I like the AD623 which comes in a DIP8 or SO8 package.
It is around $5 and runs of a single 3V-12V supply.
You can adjust the gain from 1 to 1000 with an external resistor.
Analog Devices has a number of other devices if the AD623 does
not meet your requirements.

(* jcl *)

 

One thing to consider is the capacitance of the connection between the probe tip and the amp input. Ideally you need to get the connection as short as possible.

This is why such probes are often constructed with UHF Mosfets / instrumentation amps directly mounted on the probe tip.

 

Thank you for all of the responses. I will study the situation a little further and I seem to be, as a result of the answers I received, a bit smarter about the situation. Thanks again.

 

Not sure if this has been said, but there is a configuration of an opamp that amplifies the difference in voltage of both of the inputs, so you'd just voltage divide your supply to 200mv, add that as the differential in the amplifier, then you can amplify the 0-30mv with a much higher usable range within 0-10v.

 

Let me see I really appreciate all of the helpful advice I receive from this group. I have put the AD623 into a circuit and created a ref. voltage so I can utilize only the span of my low-high sensor output. I am using a +5v power supply, and I seem to be having a little trouble. I have a 1.2K resistor as my RG. My output seems to work, although not as I had suspected. Shouldn't the output from the AD623 be 0v if the S+ & the S- are the same? Now, once the S+ increases even slightly higher than the S- ref. voltage, the output shoots up to over 1V, and maxes out at about 1.6. So my sensor is actually producing a change of approximately .6V instead of the 50mv which it produces if it is not amplified. Is this accurate? Now, my question is, could I get even more gain by inputing the output of the AD623 into the S+ of an additional op-amp? Please see the attached schematic for reference.

Thank you.

 

The output of the IA should be close to zero when there is zero volts on the input.
Remove your sensor and short the pins together. What do you read for an
output voltage?

A 1.2K resistor gives you a gain of 84 (1 + 100K/1.2K). A 50mV difference
should yield 4.2V (84 * 0.05V). Since you are getting only a small fraction of
this gain something is wrong. Have you verified your inputs at the IA?
Power supply OK? Noise? Bad connections? Are you sure you picked a 1.2K
resistor from the parts cabinet

You said your sensor is producing a change of 0.6V instead of 0.05V Is the sensor value changing when the instrument amp is connected?

Your schematic looks correct. For noisy environments the datasheet recommends additional filtering. Hard to see if your environment is noisy by looking at my
computer screen

(* jcl *)

 

I am simulating my sensor response right now using a variable output power supply. I am using 2 voltage divider circuits for S+ & S- circuits with trim pots in each circuit so that I can set the S- = to S+ when it is simulating low, as a calibration. The other is for simulating the increase in the S+ as my sensor would provide. I adjusted the first trim pot to be .200v at S-. Then I adjusted the 2nd pot to start at .200v and rise to .250v. I calculated a rough gain I wanted at around 85. I didn't want to use 100 and try to reach a full 5v just in case it wouldnt swing all the way to 5. I figured I could use a max signal just between 4 -4.5v. If I got the amp to produce 0v when the S+ and S- where the same, and 4.25v when the S+ was at .250v, my job would be done. Alas, no rest for the wicked. I am sure I have a noisy circuit, I am also equally confident that I really don't know enough about what I am doing, but I appreciate the help greatly.

 

Are you using one of those white protoboards or did you build the circuit on vector board?
Everytime I have built an analog circuit on a protoboard I end up troubleshooting the
setup not the circuit

The best way to get 0V is with a wire. Disconnect your pots and resistors and
short the inputs.What do you get for an output?

Since this part is only 8 pins I would use vectorboard, flea clips and vector pins.
For some pics see -- http://tinyurl.com/5foeou

(* jcl *)

 

Let me see I really appreciate all of the helpful advice I receive from this group. I have put the AD623 into a circuit and created a ref. voltage so I can utilize only the span of my low-high sensor output. I am using a +5v power supply, and I seem to be having a little trouble. I have a 1.2K resistor as my RG. My output seems to work, although not as I had suspected. Shouldn't the output from the AD623 be 0v if the S+ & the S- are the same? Now, once the S+ increases even slightly higher than the S- ref. voltage, the output shoots up to over 1V, and maxes out at about 1.6. So my sensor is actually producing a change of approximately .6V instead of the 50mv which it produces if it is not amplified. Is this accurate? Now, my question is, could I get even more gain by inputing the output of the AD623 into the S+ of an additional op-amp? Please see the attached schematic for reference.

Thank you.

it is tough to know how "slightly" is your "slightly". but Rg=1.2 means a gain of 80x. that means the max voltage differential between S+ and S- is 4.5v/80=60mv.

so if your "slightly" is more than 60mv, what you observed is the right behavior.

otherwise, something is wrong.

But without your being able to quantify "slightly", nobody can tell you for sure.

 

My sensor signal changes between .200 volts and .250 volts at 0 and full span respectively. When I say slightly, let's call it 1 mv. My S- is tied to .200v in an attempt to cancel out the bottom end of the sensor signal. So, when my sensor is reading 0, it is producing no voltage differential across S+ & S-. The maximum differential I will have across S+ & S- is .050mv at maximum sensor output. My output signal from the amp varies from ~1.2 to 1.6 volts at 0 and full span respectively.

 

that means your gain is 4.5v/0.05v=80. so your amp is doing exactly what you wanted it to do.

the problem you are experiencing on the lower end is likely caused by unstable 200mv reference. and that can be caused by a host of issues, both electrical or mechanical. if I were you, I would put a rc filter on that leg and power it through a real voltage reference (tl431 or a bandgap).