New OPA891 op-amp - bias current

[floobydust]

New op-amp from TI OPA891 has some really good specs for ultrasound and audio applications:
0.95nV/√Hz voltage noise
180MHz unity gain bandwidth
105V/μs slew rate
200mA output current drive

But I'm puzzled by the input bias current spec. is 9uA typ. and 20uA max. - which seems huge. NPN BJT input stage.
Am I missing something or is the datasheet out to lunch?

[coppercone2]

it looks like they don't care about DC too much for this op amp on the applications.

[Weston]

Ooh, a nice part. Thanks for sharing.

Looks like it would also be good for a rogowski coil integrator. I was using the OPA2822 before, but this part has lower voltage noise and a lower 1/F corner frequency while being able to more effectively drive coax cable.

[magic]

It's only twice worse than uncancelled bias current of LT1028, according to the published schematic of the latter. Could be explained by lower β on this high-speed process, or higher Rbb (maybe the transistors had to use less area to keep parasitics at bay) necessitating even more ridiculous input stage bias current.

This part clearly omits bias cancellation and most likely doesn't have a bootstrapped input stage cascode (wide common mode input range, compare with OPA211) so no superbeta trickery this time.

[Zero999]

0.95 nV/√Hz f > 10kHz so not specified at DC.

The, large, low beta, high GBWP, input transistors must require a decent current to drive.

[floobydust]

TI OPA891 datasheet: "These amplifiers are built using a 36V, complementary bipolar process with NPN and PNP transistors that possess an f_T of several GHz."
You can sort by I_B at Digi-Key, it appears speed is not a factor though. Bias-cancellation is there or not I guess.
Previous TI datasheet rev had an "oops nA not uA" edit lol thought slim chance it might apply.

I was pitting the part against an LT1115 plus LT1010 buffer, as in Walt Jung's excellent solutions in the LT1115 datasheet. I_B 50nA typ. (at room temp) so the bias cancellation circuit is there. Same value for LT1028/1128.

[David Hess]

But I'm puzzled by the input bias current spec. is 9uA typ. and 20uA max. - which seems huge. NPN BJT input stage.
Am I missing something or is the datasheet out to lunch?

The input bias current of 9 microamps and input offset current of 30 nanoamps, which means that it lacks input bias current cancellation.  If it had input bias current cancellation, then the input bias current would be more like +/-30 nanoamps, which is comparable to other similar parts from Analog Devices and Linear Technology.

Input bias current cancellation adds to input noise so is left out when low noise is more important, like with audio applications.