Why does the datasheet only has one input bias current value?

There should be two input bias currents, one flowing into or out of the inverting input and the other, the non-inverting input. Their difference is the input offset current. A typical datasheet only states one value for the input bias current. Is that bias current going into the inverting or the non-inverting input? Or is that an average of the two input bias currents? Or perhaps something else no specified on the datasheet?

If I assume $$\ I_{B+} = 600 \text{nA} \$$ and $$\ I_{B-} = 400 \text{nA} \$$ then I would get an average value of $$\ 500 \text{nA} \$$ and a difference value of $$\ 200 \text{nA} \$$. The calculation matches the values on the specification sheet where input bias current = 500nA and input offset current = 200nA. The leads me to assume that the single input bias current value provided on the datasheet is actually an average.

A similar question was asked here Input bias current in an opamp and its value in a datasheet but I can't find a relevant answer there or elsewhere on the web.

• look later in the datasheet, they may describe how these parameters are measured. Commented Oct 18, 2020 at 11:30
• – Chu
Commented Oct 18, 2020 at 11:33
• @Jasen i did. Was expecting at least a chart somewhere but not a single graph is shown on the entire 12-page datasheet. Guess I need to look up a more modern amp and learn from it
– KMC
Commented Oct 18, 2020 at 11:49
• Do you understand the difference between input bias current and input offset current? Here you have some info about bias current electronics.stackexchange.com/questions/521875/…
– G36
Commented Oct 18, 2020 at 12:05
• @G36 The input bias currents are generated internally in the opamp that adds to the error to the input and amplifies at the output. The offset current is just a redundant figure calculated as the difference between the inv and non-inv input bias. So the datasheet should just state the bias current for each input, and the figure for offset current is just redundant and unnecessary as I could just deduct the two bias current myself. Hence the question...
– KMC
Commented Oct 18, 2020 at 13:15

In a voltage feedback opamp1, the two input circuits are identical, at least up until the stage where their difference is taken. This helps with things like low tempco of offset voltage. The specification for bias current is therefore for each input individually.

On the data sheet you've shown, you would expect each bias current to never exceed 1.5uA over the temperature range, and the two be within 500nA of each other over the range. This is not a specification for the average of the two bias currents.

The important thing to bear in mind is that bias and offset current specify two totally different things about the opamp.

Bias current. Smaller is generally good. But it has to be compromised for noise, speed, common mode range etc with any given technology. The clever designer will use high gain transistors, reduce capacitances, use various circuit tricks (including compensating the bias current with an internal bias source) to minimise the external bias current for a given performance specification.

Offset current. Smaller is generally good. This is a measure of how well one input tracks the other, and stays tracking the other with temperature changes, and with power changes. This is important as if the user can match the two input resistances, changes in bias current will cancel out if the offset stays small. The two input circuits cannot occupy the same physical space, so variations of doping density and temperature across the die will unbalance the inputs. The more expensive opamp will actually interleave multiple transistors from the two input circuits, so even with a temperature and doping gradient across the chip, on average they will still track.

1 - a current feedback opamp has very different inverting and non-inverting inputs.

• If I follow the logic and make an assumption where inv input happens to have bias current of a typical 80nA and the non-inv input has a max bias of 1.5uA. Then the input offset current would be 1.5uA - 80nA = 1.42uA. But the datasheet specifies the input offset current cannot exceed 500nA - the datasheet cannot be correct then. The max input offset current should be the max bias - min bias.
– KMC
Commented Oct 18, 2020 at 13:26
• The logic is that if one input has a bias current of 1.5uA, and the max offset is 500nA, then the other bias current must lie in the range 1uA to 1.5uA. All currents must satisfy all conditions. The only way a typical 80nA can be achieved on one input (we're on the 25C line here) is if the other input is in the range 0 to 280nA (to satisfy max offset). The max input offset current should be the max bias - min bias. there is no min bias specified, so the max-min condition is not a number. Commented Oct 18, 2020 at 14:01
• @KMC not really, offset=maxbias-minbias would hold true only if inverting and non-inverting input biases were INDIPENDENT quantities. But they are instead somewhat correlated. So, take one of them, it shall be within max-min range. The other one is not further away the offset allows. Commented Oct 18, 2020 at 14:03
• @Neil_UK ahhh...i get it...so the offset was NOT a redundant, causational calculation from the bias (I was confused as to why subtracting the max/typ bias values didn't match that of the offset's max/typ). So the "input bias current" specifies any one of the individual input bias, whereas the "input offset current" specifies the boundary of their difference, not just any difference of input bias, for the opamp to operate properly
– KMC
Commented Oct 18, 2020 at 14:23

Because it's the 741; it's the oldest opamp still commercially available, and because it was a challenge to specify these values reliably in the 1960s, when this was invented. Because nobody caring about these specs would subsequentially be able to rely on this, TI / its predessors and competitors simply never cared enough to spec these.

Honestly, if you care about opamp specs at all, the 741 is the wrong choice. Pretty much end of story; it's a dinosaur, and not an impressive one at that – its use case is really limited to

• least-cost designs where bias currents largely don't matter
• 1968 to say, rough guess, 1975 legacy designs that were tuned to work with a µA741 and re-designing it would be a hassle
• people on the internet copying copies of copies of stolen designs from magazines that copied them elsewhere, all the way back to the late 60's/early 70's