# Does an op amp need an Input Bias Current to operate?

#### Hassan mahmoud

Joined Jan 23, 2016
19
I'm confused about Input Bias Current, is it "side effect" characteristic, where the current "leak" from the op amp. or it's a must to operate the op amp? for example if I have in the datasheet a 1pA input bias current, I should provide a 1pA to the inputs of op amp.
Note: if the type of op amp make a difference, I'm interested in
LTC6078/LTC6079
Thanks

#### kubeek

Joined Sep 20, 2005
5,796
basically it will flow in or out of the input whether you want it or not. You don´t need to provide anything. In some special cases the current needs to be taken care of, but most of the time the current is negligible.

#### Papabravo

Joined Feb 24, 2006
21,254
An ideal opamp will have an infinite input impedance, and thus have zero input bias current. So an ideal opamp does not require, and cannot have any input bias current at all. The higher the input impedance the less the input bias current will be. More input bias current implies a further departure from the operation of an ideal opamp. This means that a virtual ground in a real part won't be exactly ground. It will be very very close, but that is not good enough for some folks; for them,close just doesn't make it.

#### Hassan mahmoud

Joined Jan 23, 2016
19
An ideal opamp will have an infinite input impedance, and thus have zero input bias current. So an ideal opamp does not require, and cannot have any input bias current at all. The higher the input impedance the less the input bias current will be. More input bias current implies a further departure from the operation of an ideal opamp. This means that a virtual ground in a real part won't be exactly ground. It will be very very close, but that is not good enough for some folks; for them,close just doesn't make it.
Thanks Papabravo,
Based on you answer I understand that the input bias current is undesirable characteristic and has nothing to do with the operation of op amp. Can I build a bias circuit that will counteract the input bias current, so I will make the op amp more closely to ideal op amp? I know that it's not easy, but as a theoretical talking, can we achieve this?

#### Hassan mahmoud

Joined Jan 23, 2016
19
basically it will flow in or out of the input whether you want it or not. You don´t need to provide anything. In some special cases the current needs to be taken care of, but most of the time the current is negligible.
In a biomedical circuit you would not neglect this current, Actually, I'm reading a paper that is talking about the Cons of this current.

#### chuckey

Joined Jun 4, 2007
75
The bias current and its temperature stability can have a big effect on an opamps DC output, suppose the bias current changes from 1 to 2 pA over its operating temperature. Then if this current flows through a 10Mohm resistor to earth, then the input voltage would change from 1mV to 2mV, if the gain was set to 100, then the DC output voltage would then change by 100mV. Could be a problem.
Frank

#### Papabravo

Joined Feb 24, 2006
21,254
It is hard to imagine an external countermeasure. In a theoretical context of biomedical device perfection, I would concentrate on either increasing the input impedance of the opamp so as to reduce the input bias current to an acceptable level. Alternatively, I would investigate isolation techniques that would mitigate the adverse effects of those small currents. Does your research indicate what level of input bias current would be negligible? Certainly 1 pA is a very small current, but 1 femtoamp is 3 orders of magnitude smaller. We can go further to 1 attoamp at which point we are almost "counting" electrons. It might be difficult to get much better than that.

Here is an article on how you would even maesure 1 attoampere.
http://www.tek.com/sites/tek.com/files/media/document/resources/2648 Counting Electrons1.pdf

#### Hassan mahmoud

Joined Jan 23, 2016
19
It is hard to imagine an external countermeasure. In a theoretical context of biomedical device perfection, I would concentrate on either increasing the input impedance of the opamp so as to reduce the input bias current to an acceptable level. Alternatively, I would investigate isolation techniques that would mitigate the adverse effects of those small currents. Does your research indicate what level of input bias current would be negligible? Certainly 1 pA is a very small current, but 1 femtoamp is 3 orders of magnitude smaller. We can go further to 1 attoamp at which point we are almost "counting" electrons. It might be difficult to get much better than that.

Here is an article on how you would even maesure 1 attoampere.
http://www.tek.com/sites/tek.com/files/media/document/resources/2648 Counting Electrons1.pdf
FYI, That's what it does say in the paper:

Although the FET input amplifier (LT6078) provides low
noise operation at low power consumption with high input
impedance and low input capacitance, it requires an input bias-
ing network to provide a DC current path to counteract leakage

currents and fix the DC input voltage to a mid-rail level for
maximum output signal range. Using a simple resistive bias
network is impractical from a reliability and noise standpoint.
Although a biasing resistor can be bootstrapped to the required
resistance, minimizing its current noise contribution requires
an impossibly high value (>1TΩ). Instead, input biasing was
accomplished with two back-to-back diodes to V bias through
a 100kΩ resistor,Rb, at DC and provide a path for the am-
plifier’s input bias current in a similar scheme to ...

if you have found something that I misunderstood or something interesting, I will be happy if you share it with me
Thanks !

#### crutschow

Joined Mar 14, 2008
34,672
For bipolar (BJT) op amps the input bias current is indeed required for the input transistors to operate.
It is the input transistor's base current.

For J-FET or CMOS op amps, the input bias current is just a result of the gate leakage current.
These types generally have the lowest bias current so the LTC6078/LTC6079 you referenced likely have about as low an input bias current as you will find in commercial op amps.

#### Hassan mahmoud

Joined Jan 23, 2016
19
For bipolar (BJT) op amps the input bias current is indeed required for the input transistors to operate.
It is the input transistor's base current.

For J-FET or CMOS op amps, the input bias current is just a result of the gate leakage current.
These types generally have the lowest bias current so the LTC6078/LTC6079 you referenced likely have about as low an input bias current as you will find in commercial op amps.

#### Papabravo

Joined Feb 24, 2006
21,254
FYI, That's what it does say in the paper:

Although the FET input amplifier (LT6078) provides low
noise operation at low power consumption with high input
impedance and low input capacitance, it requires an input bias-
ing network to provide a DC current path to counteract leakage
currents and fix the DC input voltage to a mid-rail level for
maximum output signal range. Using a simple resistive bias
network is impractical from a reliability and noise standpoint.
Although a biasing resistor can be bootstrapped to the required
resistance, minimizing its current noise contribution requires
an impossibly high value (>1TΩ). Instead, input biasing was
accomplished with two back-to-back diodes to V bias through
a 100kΩ resistor,Rb, at DC and provide a path for the am-
plifier’s input bias current in a similar scheme to ...

And this is the link of that paper: http://ieeexplore.ieee.org/xpl/logi...5226856/5226857/05226885.pdf?arnumber=5226885
http://ieeexplore.ieee.org/xpl/logi...5226856/5226857/05226885.pdf?arnumber=5226885

if you have found something that I misunderstood or something interesting, I will be happy if you share it with me
Thanks !
Unfortunately that paper is not available to the general public. I refuse to support the IEEE for variety of reasons and will therefore have to forgo their collective wisdom.

#### ian field

Joined Oct 27, 2012
6,536
I'm confused about Input Bias Current, is it "side effect" characteristic, where the current "leak" from the op amp. or it's a must to operate the op amp? for example if I have in the datasheet a 1pA input bias current, I should provide a 1pA to the inputs of op amp.
Note: if the type of op amp make a difference, I'm interested in
LTC6078/LTC6079
Thanks
They all need bias current - but on modern MOS types it can be as small as a few pico-amps.

Even some bipolar types which have an emitter follower structure on the inputs, it can be fairly small.

#### Hassan mahmoud

Joined Jan 23, 2016
19
They all need bias current - but on modern MOS types it can be as small as a few pico-amps.

Even some bipolar types which have an emitter follower structure on the inputs, it can be fairly small.

#### ian field

Joined Oct 27, 2012
6,536
He's also correct - but you can't make an omelette without breaking a few eggs. If you want to influence the inputs, its going to take the transfer of some energy - no matter how tiny.

#### crutschow

Joined Mar 14, 2008
34,672
All op amps have a small bias current which varies from nA to pA depending upon the op amp design.
It's just the source of that current that can be different.
You always need to account for that current in your design (e.g. a capacitor in series with an input requires a resistive path to ground or to some bias voltage for the bias current and to establish the DC voltage for that input).

#### #12

Joined Nov 30, 2010
18,224
When you get an answer that doesn't speak to you well, use the answers you understand.

#### #12

Joined Nov 30, 2010
18,224
It is hard to imagine an external countermeasure.
I believe I saw an input bias current compensator in an Application Note from National about 40 years ago. Of course, that was when we only had BJT inputs. That circuit probably has much less usefulness now that we have j-fet and mosfet inputs.

#### ian field

Joined Oct 27, 2012
6,536
I believe I saw an input bias current compensator in an Application Note from National about 40 years ago. Of course, that was when we only had BJT inputs. That circuit probably has much less usefulness now that we have j-fet and mosfet inputs.
Bootstrapping emitter follower input amplifiers used to be a popular way of maximising input impedance on discrete component designs, but I don't know of any op-amps with access to the connections you'd need to get at.

Its almost positive feedback, but since the follower has less than unity voltage gain; oscillation is pretty unlikely. you can do a smidge of positive feedback on an op-amp - but if it bursts into song, you have to delve into the math of poles and zeros.............and worse!

#### dannyf

Joined Sep 13, 2015
2,197
I'm confused about Input Bias Current, is it "side effect" characteristic
a bjt opamp will need to have a dc path to operate -> it has to have input current. In theory, input bias current, with sufficiently high beta, can approach zero so the answer to your question will depend on your perspective -> how low of a level of input bias current will you consider to be "zero"?

fet opamps does not need input (bias) current to operate, though they all have small levels of input bias current.

#### ian field

Joined Oct 27, 2012
6,536
a bjt opamp will need to have a dc path to operate -> it has to have input current. In theory, input bias current, with sufficiently high beta, can approach zero so the answer to your question will depend on your perspective -> how low of a level of input bias current will you consider to be "zero"?

fet opamps does not need input (bias) current to operate, though they all have small levels of input bias current.
With JFET input stages, you're reverse biasing a PN junction, so the bias current and leakage current amount to pretty much the same thing.

With MOS input stages, the gate is insulated from the channel by a very thin silicon dioxide layer - the leakage current is orders of magnitude lower, but there's no such thing as a *PERFECT* insulator.

You probably can't measure the JFET leakage without laboratory instruments, let alone MOS leakage - but in some applications this stuff matters.

Its probably best not to forget that generally; MOS inputs have more shunt capacitance than bipolar types - not a problem at DC, but starts loading the signal as frequency rises.