Differential amplifier for two NTC thermistors

I want to construct a circuit, which multiplies the voltage difference over two NTC thermistors. I made this circuit, but it does not work as intended, since the output doesn’t match the calculated one.

The output, in the given example, should be 1.74V but it is 0.18V so something is defintly wrong.

example.JPG

Hi,
Welcome to the forum.

Please read the first post in any forum entitled how to use this forum.
http://forum.arduino.cc/index.php/topic,148850.0.html

OPs Circuit;

An LM741 is not suited to single ended supply and a supply as low as 5V.
Look up the datasheet for the LM741.

Have you replaced the thermistors with fixed resistors and calculated what your output should be?

Tom... :slight_smile:

Hello Tom, thanks for replying.

I set the thermistors to a value of 900 and 400 to calculate the output, which should be 1.74V but by simulating the circuit, the outut was 0.18V.

And as for the uA741, the datasheet states "The device can operate as a single-supply
or dual-supply operational amplifier depending on the application."

Thanks

I’m unwilling to click on embedded links, so I haven’t seen your schematic (or, whatever it is at the other end of your link), but a few possibilities I can offer:

With each thermistor in it’s own voltage divider arrangement, feed each divider voltage to an Arduino’s analog input, then do the multiplication in code. Then, serially feed the resulting value to a serial input DAC or implement a PWM to analog converter on one of the PWM Arduino outputs (you might need to hack Timer 1 to get a faster PWM period – checkout: https://www.arduino.cc/en/Tutorial/SecretsOfArduinoPWM [know link, so OK :wink: ].

OR

feed each divider voltage to the input of an op-amp configured to convert linear to exponential, then feed those outputs to an op-amp configured as an adder, then do an op-amp exponent to linear conversion.

OR

Use an analog multiplier IC like the AD633 (the datasheet calls this “Low Cost” $14.28 a pop – but it’s less than the other ones I looked at!). Minimum power supply is +/-8V, so that might be an issue.

Hi ReverseEMF.

I am not able to see why I should use PWM or use an op-amp to convert linear to exponential. I understand you didn't click on my link, which is understandable, and I have therefore attached it to the post. I hope this will help you help me.

Thanks

Bassusour:
And as for the uA741, the datasheet states "The device can operate as a single-supply
or dual-supply operational amplifier depending on the application."

Yes, but when used with a 5V single-ended supply, there isn't enough "head room" to do much of anything useful.

The 741 was designed to function with supply voltages more like +/-15 volts. The datasheet has a spec called the Input Common Mode Voltage Range [Vicm], and for the 741, run with a supply of +/- 15 volts, the Vicm is +/-12 volts. In other words, when an input voltage goes below 3 volts + the negative supply rail, then the OP-Amp won't respond to that voltage. Where did I get "3 volts"? By subtracting 15 from 12 = 3. That 3 volt limit also applies to voltages that are 3 volts below the upper supply rail.

So, if the supply rails are: 5 volts and 0 volts, then the input signal needs to be between 5V - 3V and 3V - 0V. In other words, with a 5Volt supply, the 741 can't even see an input voltage.

Also, since a 741 output doesn't swing to "the rails" and in fact, has a best case max voltage swing of 1 volt less than the supply voltage (and worse case of around 3V), that's a loss of 2 volts on the output (because that 1 volt loss applies to the highest voltage swing AND to the lowest voltage swing). So, even if the 741 could respond to an input voltage, when run at 5V, it wouldn't be able to swing it's output much, or even at all!

You need an op-amp that is not only designed with 0 volts as part of it's input common mode voltage range, but also one designed to work at 5V.

change R2 to 47k

ReverseEMF thanks man, that really explained a lot. I have been discussing this with my teacher, who couldn't see the problem. Defintly a +karma for me.

And as for another op amp, what do you suggest?

I am not able to see why I should use PWM or use an op-amp to convert linear to exponential.

Because:

  1. PWM can be used to convert a digital value to an analog voltage by using the digital value to set the Duty Cycle. When the PWM output is sent through a Low Pass Filter, the output is a DC voltage that is proportional to the digital value. Drawbacks are: a tradeoff between ripple voltage and conversion delay (or reduced rate of change) -- this can be reduced by lowering the PWM period -- i.e. the higher the frequency, the easier it is to filter out the "carrier" frequency, and the quicker the filter response can be.

  2. Because multiplication can be done by converting the values to exponential form, then adding the exponents, and then taking the log of the value. This is an old Analog Computer trick, and was done because it's a lot easier to ADD voltages with Op-Amps than it is to multiply voltages. And, it's fairly easy to do the linear to exponent and back to linear conversions. It just takes a lot of components :wink: Also, probably good to use higher voltages and dual supplies, and as such is probably WAY over kill for your application :wink: But, it's a nice bit of esoteric information, no?

BTW: Here's the multiplier circuit I was talking about:

Also, here's the circuit and math for the one that it appears you are using:

Notice how, if the two input voltages are the same, any changes in the thermistor resistances will have no effect on the output, since they are multiplied by zero, and the expected output is 0, not 1.74V (actually the calculated output is negative zero :wink: )

Also, even if you applied different voltages to the thermistor legs, you would get a voltage on the output that is a ratio of the two thermistor resistances, not a multiplication. Also, you need to set the input voltages to something that is within the input common mode range. Most op-amps have an input range that is a few volts below the upper rail. And even with an op-amp designed to see all the way to the upper rail, it's still not a good idea to use input voltages so close to the upper limit, since things like the input offset ratio can place the effective input voltage outside that range. Better to use a voltage closer to the midpoint between the upper and lower rails. Consider the full range of possible voltage swings when designing an op-amp circuit.

Here's something that was bouncing around in my ancient memory (from many decades ago), and by golly they're still available! It's a Balanced Modulator/Demodulator called the MC1496.

And, they are less than a dollar!

The drawback is: this IC isn't really designed to be a DC multiplier, but I scanned the datasheet and it looks like it will work with DC levels. Arrange the thermistors, each in a voltage divider arrangement, and the divider outputs to the signal and carrier inputs. It will probably take some experimentation, but heck, that's how I learned electronics -- by scratching my head, over and over, until I figured it out!! I don't have much hair left (probably for different reasons) but I do have a fair amount of electronics and programming experience under my proverbial belt, so it was worth it :wink:

BTW: Amplitude Modulation is a form of analog multiplication.

Hello again! I’ve been scratching my head, perhaps with more hair left than you :wink: , and found this site (trust me it’s safe) and made the following circuit. However, I am a bit curious to how the gain is calculated?

It is a problem for my existing circuit, which you can see in my first post, that the output will be 0, if R24 is equal, or lower, than R23. This is a problem, because it is very likely that that will happen.

R6 and R7 are the thermistors in the diagram.

Your help is greatly appreciated :slight_smile:

I want to construct a circuit, which multiplies the voltage difference over two NTC thermistors.

It is not completely clear what you want to do. "Multiplies the difference" implies a four quadrant multiplier, which is quite a complicated analog circuit. Since your understanding of op amp circuit design is very limited, why not just do everything with the Arduino?

All you need are the two thermistors, their respective resistors and two analog inputs.

This is for a school project, and I honestly just wanted to include some more electric components to show understanding. I also just find it too easy to just send it through to analog inputs. If you didn't notice, I thought this was a simple fix, but it turned out to be much more complex.

formula can be ok, but value of the components maybe are different than on the paper. U7 pin 3 - should be 2.5 V you have about 4V, use separate voltage divider 2 resistors 100k, or remove R17, google - Wheatstone Bridge Differential Amplifier.

I want to construct a circuit, which multiplies the voltage difference over two NTC thermistors.

I don't really understand what you want to do...

But if you wanted to find the difference between two temperatures the Arduino is really good at math! :wink: So you could find the temperatures the "usual way" and then find the difference (or multiply, etc.) in software.

It is common to use a differential amplifier for a thermocouple, because thermocouples put-out a very-small voltage and a differential amplifier is more immune to noise and ground loops. For the same reason, pro microphone amplifiers are also differential (they call it "balanced" connection in audio terminology).

Or if you want the temperature difference you could make a voltage divider with just the two thermistors. (I'm not sure how you'd calibrate that but it should work.)

The 741 has a very limited common mode range and output voltage swing on a 5volt supply (not even listed in the datasheet).
I doubt if you can make your circuit do what you want with that dinosaur.
Leo..

Two thermistors in one package = strange aragment ?

I honestly just wanted to include some more electric components to show understanding.

You don't have that understanding, because you can't even explain the goal of the project.

Show off something that you do understand, or expect to engage in some very serious study.

Bassusour:
And as for the uA741, the datasheet states "The device can operate as a single-supply
or dual-supply operational amplifier depending on the application."

But not from a single 5V supply, and definitely not rail-to-rail - you want a modern opamp
that works from a single 5V supply - ie a low voltage device that's rail-to-rail on input and output.

The 741 was ground breaking in its day, its frankly a poor choice these days.