When does an inverting opamp circuit need resistor to ground?

[grizedale]

Hello,

Is it only the old-fashioned opamps that need to have a resistor to ground as in the following circuit of an inverting opamp?......

https://i45.tinypic.com/2rhle77.gif



Is it true that the following opamp (TSV321) would not need this resistor?..........since it would already be very accurate without it?
https://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00063292.pdf

 

[mister_rf]

The resistor R was added to the standard inverting amplifier to cancel out bias currents (output offset).
The compensation resistor causes a current on the positive terminal, equal and opposite to current flowing into the negative terminal. So any DC output offset caused by the inverting input is cancelled by the non-inverting input. The value of resistor should be equal to the parallel combination of Ri and Rf.

 

[FoxyRick]

The datasheet gives a worst-case input bias current of 150nA for that opamp. So, if you had an input impedance of 100kΩ (or the parallel Rf-Ri mentioned by Mister_rf), there could be a voltage of 15mV developed across the input impedance. Not a lot, but if that's a problem then a matching compensation resistor will mitigate that.

There are many opamps with much lower input bias currents and unless you are then dealing with tiny signals or huge input impedances, I doubt that a compensation resistor would be of any benefit. Likewise of the input impedance is lower as it very often is, the offset voltage introduced might be below the level of interest.

Originally, simple bipolar opamps had quite high input bias currents so the issue was much more prevalent that it is now, but I would say it is still something to be aware of, even if only to think, "no - not needed this time."

 

[FvM]

As already clarified in the previous posts, saying the resistor "is needed" misses the point in any case. It's at best suggested or helpful in some cases.

Two general reservations in addition:
- bias current compensation resistor are effectively useless for FET input OPs
- they are only meaningful for OPs with |Ibias| > |Ios|, in other words classical bipolar OPs without internal bias current compensation

For the asked OP with classical bipolar input stage (a kind of LM358/LM324 clone), a compensation resistor can be useful, if it reduces the total offset voltage. The effect should be calculated before placing a resistor. In addition, a bypass capacitor may be needed to avoid additional noise generated by the resistor.

One could guess that an application running well with a cheap TSV321 won't be particularly obligated to low offset and thus most likely can save the compensation resistor as well.

 

[Kral]

grizedale,
To aid in making the decision whether a bias current compensating resistor is required, you can calculate the output voltage error due to bias current as follows:
Let
Ri = The equivalent resistance from the inverting input to ground
= RfRi/(Rf + Ri) for your configuration
Rni = The resistance from the non inverting input to ground
Ib = The bias current
Ve = The output voltage error due to bias current
Then
Ve = RfIb(1-Rni/Ri)
.
From this equation it is apparent that if Rni = Ri, the error will be zero.
If Rni = zero, the error is simply IbRf.
.
Be aware that the plarity of the output voltage error depends on the direction of the bias current which is a function of the input stage of the op amp.

 

[grizedale]

Thanks

I(bias) and i(offset) are less than 150nA for the TSV321, so i can presume that this is an opamp with FET inputs...the datasheet doesnt say

(It may seem a silly question, but why would anyone choose an opamp with a bipolar input stage, since they have higher bias and offset currents and have no advantages over FET based opamps)

 

[Kral]

grizedale,
This appears to be a bipolar input stage op amp. It is listed as a low voltage replacement for the LM324, which is a bipolar. Also, the bias current of 40 nA typical is in the range you would expect for a commodity bipolar input stage op amp.
.
It is difficult to make accurate general statements, but FET input op amps tend to have a much higher input offset voltage drift. Also, FET input op amps tend to have high input voltage noise and low current noise. It's just the opposite with bipolars, which have low input voltage noise and high current noise. As a result, the driver source impedance will affect your choice of op amps from a noise perspective. Low voltage noise FET input op amps are available, but they tend to have high input capacitance,which is not usually a big problem for audio applications, but it is for high frequency and switching applications.

 

[grizedale]

I used to work for a shower company who used the LMC6035 op amp to amplifiy and buffer the signal from a thermistor in the hot water pipe, as part of regulating the shower temperature...............since this was a low bandwidth regulation, were they then wrong to use a cmos based opamp like LMC6035?

LMC6035 datasheet
**broken link removed**

Also, supposing one wishes to amplify the (DC) voltage signal across a current sense resistor in series with a LED load whose current is being regulated..........since the feedback loop is going to crossover at less than 1KHz, would you again say that this is a job for a bipolar opamp?

 

[godfreyl]

For most industrial applications, the "right" choice is the cheapest one that is good enough. In many cases almost any opamp will be good enough. In other cases, certain parameters will be important.

Some opamp parameters that may be important are:
a) High speed (bandwidth and slew rate).
b) Low noise.
c) Low distortion.
d) High output current.
e) Low input DC offset voltage.
f) Low input bias and offset currents.

For example, for a microphone preamplifier, low noise is very important, but for a headphone amplifier, noise is much less important but high output current is very important.

For your LED example, if the voltage to be measured is very small, then low DC offset voltage would be important.

NE5534 is often used for line level audio circuits as it has very low distortion, high enough speed and low enough noise. However it also has high input bias and offset currents, which can be a problem when the source impedance is high. In that case a JFET-input opamp like TL071 would be a better choice.

The old 741 is bad in almost every way by modern standards, but it is still sold and used because in many cases it is good enough.

 

[Kral]

I agree with goddfreyl that the best op amp is the cheapest one that is good enough. “Good enough” is determined at the time the accuracy requirements of the system are determined. In the case of the water temperature problem, the critical parameter would be the required temperature accuracy. For example, the type J thermocouple has a sensitivity of about 52.6 uV/DegC in the region of 40 Deg C. The LMC60357 has a typical offset voltage drift of 2.3 uV/Deg C. Double this for worst case to get 4.6 uV/Deg C. This translates to .087 Deg C water temperature error for each Deg change in ambient temperature. The question is, (“Can you tolerate this error?) If you can, then this op amp is good enough. Out of curiosity, how did you perform your cold junction compensation? How did you compensate for the initial offset voltage of 5 mV?
.
For the LED current regulation problem, input capacitance would not be a problem. To determine error due to offset voltage drift, you would have to compare the scale factor of the sense resistor to the equivalent op amp voltage drift converted to the equivalent mA/Deg ambient to determine whether the error is acceptable. This is similar to the process described above.