Design a Single Supply Difference Amplifer

[akhurash]

Hi, I was wondering if anyone knew how to design a single supply difference amplifier... I don't want a comparator or a difference amplifier chip. I need something that will output an AC wave. The inputs of the amplifier are AC sine waves which have slight difference in amplitudes. I designed one using a schematic I found on Google but I need one that is single supply. When I try to put a DC bias and make the op-amp I'm using a single supply the output is an AC wave but it doesn't seem to work correctly.

The output from the amplifier is going into a synchronous demodulator, so I’m also using a DC Blocking Capacitor at the output of the amplifier. I did to use both a very small cap and a relatively small cap and both didn’t work. I’m putting the cap at the end of the amplifier because I want to block any DC bias.

Any help is greatly appreciated!

 

[berkeman]

Hi, I was wondering if anyone knew how to design a single supply difference amplifier... I don't want a comparator or a difference amplifier chip. I need something that will output an AC wave. The inputs of the amplifier are AC sine waves which have slight difference in amplitudes. I designed one using a schematic I found on Google but I need one that is single supply. When I try to put a DC bias and make the op-amp I'm using a single supply the output is an AC wave but it doesn't seem to work correctly.

The output from the amplifier is going into a synchronous demodulator, so I’m also using a DC Blocking Capacitor at the output of the amplifier. I did to use both a very small cap and a relatively small cap and both didn’t work. I’m putting the cap at the end of the amplifier because I want to block any DC bias.

Any help is greatly appreciated!

Can you post a schematic for what you are doing now?

Also, what kind of specs do you want out of your differential amplifier? Input offset voltages and currents, bandwidth, supply voltage, input and output signal amplitudes, etc.

 

[akhurash]

Thank you for your reply berkeman. For the differential amplifier ideally I would like a very low offset and low current, the bandwidth I think would be around 20kHz because there is no gain in the amplifer but I'm not a 100% sure about this. For this amplifer I can only have one single positive supply voltage. The output signal does to a demodulator and the output from the demodulator goes through a series of LPF so the amplitude at the output of the differental amplifer isn't that important BUT since there is a DC offset (+2.5V) being amplifed because of single supply operation the peak-to-peak cannot exceed 5V.

I have attached a schematic picture the difference amplifer I am working on. It seems to work right now but if you see any flaw in it please let me know. The part I can't get to work right now is the demodulator using single supply op-amps. I have also attached a schematic picture of the demodulator. The output from the non-inverting amplifer seems to be working but the DC level from the inverting amplifer is about +1.5V and it should be +2.5V and I can't figure out why this is happening. The circuit works just fine if I use both negative and positive supply voltages.

 

[es1]

Is it working in lab or in spice?

When you say you want very low offset, what do you mean? The resistor biasing scheme used might be a problem depending on what "low" means (these days when someone says they want a low offset I think sub 1mV. But you are using a 3mV nominal opamp so...). The input resistors might also not be compatible with "low" current. It will also depend on how you spec the current for the input stage, i.e. do you mean just current into the AC coupling cap?

TL071 is likely not the best choice for a single supply circuit depending on what VCC+ is.

For anyone to answer why U3 is misbehaving they will need to know what R7 is connected to.

 

[akhurash]

Thank you for the reply es1. The circuit is working in SPICE; I haven't tested it in lab yet. The AC peak-to-peak is about 300mV for each signal before going through the differential amplifier, so I think 3mV offset would be fine for my application.

Do you have any single supply op-amp in mind that would work well, instead of the TL071? I tried to find a single supply op-amp by looking through Digikey but I didn't find one that I think would work for this application that was also in SPICE (I'm using Multisim). The VCC+ will be 5V max.

Sorry I forgot to label the second picture I uploaded. The R7 is connected to the positive input of U2, which is connected to the input signal. The input signal of the demodulator is an AC wave but it's not a typical modulated signal, the modulated signal is from a sensor, that's why I couldn't use a simplifier demodulator.

Here is a little background information on my application: The differential amplifier is being used to subtract two AC signals, the two AC signals are coming from two sensors. There will be slight changes in the AC signals (mostly due to the how they are being used but also do to the component tolerances). The two AC signals gets subtracted (using the differential amplifier, that's why I used the term difference before but not really sure if that matters) and then goes through the demodulator. The demodulator output is then filter using 3 LPF (I think 2 will work, haven't really tried to simplify this yet) and a DC voltage is obtained based on the difference of the two AC signals. The two AC signals being substracted have difference amplitudes most of the time, there is a very small phase differences between them due to phase shifts in the circuits.

Attached is a picture of what the circuit looks like currently. I put a voltage follower between the difference amplifer and the demodulator with a DC blocking capacitor before the voltage follower. The "Clock" input connect to the switch of the demodulator is actually the a square wave used to drive the circuit.

 

[berkeman]

I'm not understanding the circuit yet, but here are some suggestions for opamps.

When you only have a 5V power supply and want to use opamps, you should be looking at CMOS "rail-to-rail" opamps to get the best input and output voltage ranges. Here is a selector guide at Maxim IC:

http://para.maximintegrated.com/en/search.mvp?fam=op_amp&682=Input/Output|Output Only&tree=amps

And here is a list of the SPICE models that Maxim IC offers, including some of their rail-to-rail opamps:

http://www.maximintegrated.com/design/tools/modeling-simulation/spice/operational-amplifiers/macro/

.

 

[berkeman]

I'm still not understanding the switch figure for a synchronous demodulator... It's left over from your original thread asking about demodulators:

https://www.physicsforums.com/showthread.php?t=635380

Could you explain how you are trying to "synchronously demodulate" the output of your difference amplifier stage?

 

[akhurash]

What I'm trying to synchronously demodulate is an AC signal from a magnetic sensor. The magnetic sensor outputs an AC signal based on the external field it is sensing; when the external magnetic field changes the sensor output also changes. The DC voltage that is obtained after this AC signal is demodulated and filtered will be unique to the field strength the sensor is sensing. For my application the two sensors are sensing the same field but they are far apart, so the strength of the field on one sensor increases while on the other it decreases (when the field is changing). What I'm trying to do is build a hardware that will compare these two signals and give me an absolute DC voltage in reference to the strength of the field on one of the sensors.

My application is similar to a fluxgate magnetometer. It’s similar in the sense that it is sensing a magnetic field and the output signal looks similar to that of a magnetometer. If I was using only one sensor I wouldn’t need a difference amplifier, I could just put the output through the synchronous demodulator and get the DC voltage corresponding to the field strength. Since my sensor works similar to the fluxgate magnetometer I’m assuming that the output of the sensor is modulated and I need to synchronously demodulate it (like in fluxgate applications). I tested the sensors in lab too see the output with different fields exposed to them. To be honest I'm not completely sure about the signal processing of a fluxgate magnetometer, that's why I can't explain the synchronous demodulator to well. I do know you need one to get a DC voltage that corresponds to a certain magnetic field.

What I could do is use two different circuits (without the difference amplifier) and compare the two DC voltages from the sensors using a microcontroller and compare it in software instead of hardware but for me it will be easier to do it with the hardware.

 

[berkeman]

What I'm trying to synchronously demodulate is an AC signal from a magnetic sensor. The magnetic sensor outputs an AC signal based on the external field it is sensing; when the external magnetic field changes the sensor output also changes. The DC voltage that is obtained after this AC signal is demodulated and filtered will be unique to the field strength the sensor is sensing. For my application the two sensors are sensing the same field but they are far apart, so the strength of the field on one sensor increases while on the other it decreases (when the field is changing). What I'm trying to do is build a hardware that will compare these two signals and give me an absolute DC voltage in reference to the strength of the field on one of the sensors.

My application is similar to a fluxgate magnetometer. It’s similar in the sense that it is sensing a magnetic field and the output signal looks similar to that of a magnetometer. If I was using only one sensor I wouldn’t need a difference amplifier, I could just put the output through the synchronous demodulator and get the DC voltage corresponding to the field strength. Since my sensor works similar to the fluxgate magnetometer I’m assuming that the output of the sensor is modulated and I need to synchronously demodulate it (like in fluxgate applications). I tested the sensors in lab too see the output with different fields exposed to them. To be honest I'm not completely sure about the signal processing of a fluxgate magnetometer, that's why I can't explain the synchronous demodulator to well. I do know you need one to get a DC voltage that corresponds to a certain magnetic field.

What I could do is use two different circuits (without the difference amplifier) and compare the two DC voltages from the sensors using a microcontroller and compare it in software instead of hardware but for me it will be easier to do it with the hardware.

So you are driving a sine wave into the sensors, and will use that same sine wave for synchronous demodulation? Are the received sensor signals very noisy? Is that why you need to use synchronous demodulation instead of just using envelope detection to find out the amplitude of the sensor output waveforms?

And I still have the same question about the switch representation and opamps that you are using to try to do the demodulation. Shouldn't you be using a Gilbert Cell type of device to do the mixing/multiplication/demodulation?

 

[akhurash]

The sensor are driven by a very small AC sine wave (about 100mV, I tried to use 10mV but the noise was high and couldn't get a accurate output signal, this was done in the lab). The sine wave is obtained by using filters and filtering a square wave (the square wave is outputted from a microcontroller). The square wave is also feed into the switch of the demodulator, so the frequency will be the same and the phase will be similar; there is a small phase shift between the square wave at the demodulator and the sine wave which is driving the sensors. The sensors are also feed a DC current which makes the sensor approach saturation, they are not completely saturated. In this state the sensor are able to detect external magnetic fields and the output changes when that field changes. I haven't built this exact circuit but I did do some testing in the lab using the sensors and the DC current and AC voltage and I looked at the output. The DC output changed when I applied an external magnetic field to the sensors. The received sensor signals don’t seem to be very noisy using the 100mV peak-to-peak AC sine wave. I'm using the synchronous demodulator because when I searched about the fluxgate magnetometer that's the type of demodulator others were using. Since I haven't tested the demodulator in lab I don't know if the envelop detection will work. I did try this in SPICE. I used an envelope detector instead of the demodulator and the circuit didn't work properly. I used a variable inductor to model the sensor and when I used the synchronous demodulator the circuit seemed to work in SPICE; when I say it seemed to work I mean I got the output I expected and the value of the output DC voltage was changing when I changed the inductances of the inductor.

I wasn't aware of the Gilbert Cell type device but I'm looking into it now and I will use that for the demodulator.