TL072 op-amp voltage follower acts weird

Thread Starter

CuriousScientist

Joined Aug 4, 2021
7
Hi All,
I hope that I am posting to the right forum. After days of searching, reading, experimenting, I still have not been able to find a proper solution which I can tailor to my issue, so it is time to ask you guys.

I made a voltmeter which consists of an Arduino microcontroller and an ADS1115 module that I use in differential mode because I measure negative voltages too. Unfortunately, this ADC module does not have high enough input impedance for the task I want to perform, so it influences the measured circuit. I thought I should build a voltage follower using an op-amp with a high enough input impedance and connect it to the inputs of the ADC.

So, I found this TL072 op-amp and I built a voltage follower with it. Here is an ugly sketch of the relevant part of the follower:
TL072_question.JPG
The input (A) takes the let's say positive input and ''forwards" it to the A0 input of the ADS1115 module. The input (B) does the same thing but it goes to A1. The measured voltage range is roughly +/- 3 V, or even lower. The +/-15 V power is supplied by a XL6007e1-based power supply circuit which is a +/- power supply with a step up converter, so I can power it from the microcontroller or USB directly.

So, after the long introduction, let me tell the issue. First of all, the circuit works very well when the inputs of the op-amp are connected to a voltage source. However, when I remove all the connections from the input (A and B are floating), the op-amp goes bananas and it starts to overheat. Within a minute, I cannot hold my finger on the chip anymore. Furthermore, I could measure ~33 mA current flowing between the output of the op-amp and the input of the ADC which is not good at all. Interestingly, when I do not connect the outputs of the op-amp to the AD converter, despite the fact that I let the inputs of the float, the op-amp does not overheat! So, I am extremely confused with this issue.

Finally, my questions:
1.) Is there any explanation/solution for this issue? Could it be that the 3 kOhm resistor is too small, or am I overlooking something trivial?
2.) Are there any better op-amps than this one for voltage-follower purposes within the range of giga/terra-Ohm input impedance?

Thank you very much for your time!
 

MrChips

Joined Oct 2, 2009
24,612
Welcome to AAC!

I don't know why it is overheating.
Remove the capacitor and the resistor and try again.
Instead, put the 100nF capacitors from +15V to GND and -15V to GND.

btw, that is not the way to design a difference amplifier.
This is the way to do it. Measuring negative voltages is a different issue.

1628105444487.png
 

crutschow

Joined Mar 14, 2008
28,177
Try putting a 10kΩ resistor in series with each of the op amp's negative inputs.
There may be a sneak path from the negative input to the output, when the output goes to the rails when the plus inputs are floating.
 
Last edited:

MrSalts

Joined Apr 2, 2020
714
Your op amp is overheating because it is oscillating and supplying, then extracting, charge to the capacitor.
As a general rule,
1- you don't connect a capacitive load to an op amp output (more than 1nF)
2- you don't leave op amp inputs floating. The charge can drift way beyond the + or - power supply with static charge and damage the op amp. This rule is generally true for all CMOS logic and JFET op amps (and other high impedance input chips.
 

Thread Starter

CuriousScientist

Joined Aug 4, 2021
7
Welcome to AAC!

I don't know why it is overheating.
Remove the capacitor and the resistor and try again.
Instead, put the 100nF capacitors from +15V to GND and -15V to GND.

btw, that is not the way to design a difference amplifier.
This is the way to do it. Measuring negative voltages is a different issue.
Thank you very much! I assembled the circuit without the C and R and I placed the capacitors on the power supply pins. Unfortunately, I still have this overheating issue.

Regarding the circuit you posted, I do not want to build that kind of circuit. The ADS1115 has differential inputs where you can have a pair of the input pins "floating", the voltage across these pins can be between +/- 6 V. So, I want to make the followers for this application. Both followers have to be able to cover this +/- 6 V.
 

Audioguru again

Joined Oct 21, 2019
3,836
When you let the + input of an opamp that has a +15V and -15V power supply float, it might oscillate or float to +15V or -15V. The input protection of the ADS1115 limits the voltage then the outputs of the opamps draw their shorted output current of about 33mA and get very hot.

Why are the opamp power supply voltages so high? Why is there a negative power supply voltage?
The TL07x opamps need a negative power supply voltage because if an input voltage gets within a few voltage of the negative supply then the output suddenly goes as high as it can. Many newer Jfet input opamps like an OPA2134 work fine without a negative supply.
 

BobTPH

Joined Jun 5, 2013
4,034
Putting a differential of ±6V on the inputs of an opamp is totally useless. A difference of a fraction of a millivolt will drive the ouput to one or the other rail.

And leaving a high impedance input floating is equally useless. What do you expect the circuit to do when an input is floating?

Tell us what you are trying to do instead of asking us why what you have done does not work.

Bob
 

Thread Starter

CuriousScientist

Joined Aug 4, 2021
7
Try putting a 10kΩ resistor in series with each of the op amp's negative inputs.
There may be a sneak path from the negative input to the output, when the output goes to the rails when the plus inputs are floating.
Thank you very much! Unfortunately, I was still not able to solve the issue with the 10k resistors.
 

Thread Starter

CuriousScientist

Joined Aug 4, 2021
7
Your op amp is overheating because it is oscillating and supplying, then extracting, charge to the capacitor.
As a general rule,
1- you don't connect a capacitive load to an op amp output (more than 1nF)
2- you don't leave op amp inputs floating. The charge can drift way beyond the + or - power supply with static charge and damage the op amp. This rule is generally true for all CMOS logic and JFET op amps (and other high impedance input chips.
Thank you for the ideas! I tried to look for some capacitance-related info in the datasheet of the ADS1115, but I could not find any info. Do these circuits count as a capacitive load? Also, if I cannot leave the inputs floating, then what would be a good way to handle them? Should I tie them to the GND or should I short them together? Or something else?
 

Thread Starter

CuriousScientist

Joined Aug 4, 2021
7
When you let the + input of an opamp that has a +15V and -15V power supply float, it might oscillate or float to +15V or -15V. The input protection of the ADS1115 limits the voltage then the outputs of the opamps draw their shorted output current of about 33mA and get very hot.

Why are the opamp power supply voltages so high? Why is there a negative power supply voltage?
The TL07x opamps need a negative power supply voltage because if an input voltage gets within a few voltage of the negative supply then the output suddenly goes as high as it can. Many newer Jfet input opamps like an OPA2134 work fine without a negative supply.
Thank you very much! If you mean why I used a +/- 15 V power supply, it is because I had this. I looked up the datasheet of this op-amp and I saw that the max supply is +/- 18 V, so I thought that this PSU will be OK.
 

MrSalts

Joined Apr 2, 2020
714
Thank you for the ideas! I tried to look for some capacitance-related info in the datasheet of the ADS1115, but I could not find any info. Do these circuits count as a capacitive load? Also, if I cannot leave the inputs floating, then what would be a good way to handle them? Should I tie them to the GND or should I short them together? Or something else?
A "typical" DMM has a load to ground of 10Mohm. That would be a good start. Some 5.5 or 6.5-digit precision bench meters have extremely high input impedance (like your existing design) but they tend to hold the previous input because of gate capacitance and then drift all over the place if the inputs are left floating. I think a 10M resistor to ground is a good idea.
 

Thread Starter

CuriousScientist

Joined Aug 4, 2021
7
Thank you very much for all the ideas, suggestions and everything! I am really thankful.

It seems that the mixture of MrSalts, MrChips and crutschow suggestion is the go to solution! I attached a figure to show what I did.
tl072-working.JPG
So, what I did is the following:
1. I pulled down the inputs of the op-amps (AIN and BIN) to GND using a 10M resistor. In fact, I used 1M because I could not find larger at this moment, but it seems to work too. Nevertheless, I will buy some high-value resistors for future work.
2. I used a 10k resistor between the inverting input and the output of the op-amps.
3. I used a 100 nF capacitor between each supply rails and the ground.

With the above configuration, it works properly. I get the proper voltage readings on the ADC and the circuits stay at normal temperatures. Also, now the multimeter does not pick up any current between the output of the op-amp (AOUT or BOUT) and the corresponding input on the ADS1115. Fun notice: If I unplug the 10k resistor, the current jumps back to 33 mA and everything start to get hot.

Once again, thank you for all the comments and help!
 

Thread Starter

CuriousScientist

Joined Aug 4, 2021
7
You cannot unplug the 10kΩ resistor and expect it to work.
Instead, try putting a jumper from output to inverting input in place of the resistor.
Sorry, of course yes, I did a silly mistake here as I opened the circuit between the output and the inverting input by removing the 10k resistor just like that. And of course if I open that part of the circuit, it stops being a voltage follower anymore and it goes on a rampage. Nevertheless, the circuit I posted about an hour ago works well, so thank you again for your inputs!
 
Top