Low Noise, Low Ibias Op-Amp over temperature

[AeroETech]

Hello everyone,

I've been searching the usual suspect websites for hours, and my eyes are starting to bleed.  Unfortunately, none of the sites have a good filter for Ibias(max) over full temperature range, and it can be very, very different than Ibias max at 25C.

I'm hoping one of you here may have already run across something with the specifications I'm looking for.

Op-Amp.  At least 30V supply range.  Lowest noise possible/reasonable.  Ibias < 1nA over full -40 to 125C range.

Low-ish offset and offset drift are nice-to-have, but low noise is the primary concern.

ADA4523 is a good fit, but the Iq per amplifier is painful.  There will be 25 of these on the board, and that power/heat starts to add up really fast.

So basically, I'm looking for a wide temperature range, low iBias, very low noise amplifier with a more reasonable Iq than ADA4523.

If anyone has any thoughts, I would really appreciate it.

Thank You!
Ben

[Swainster]

Quick disclaimer - I'm a metrology wannabe, i.e. enthusiasm for the subject but no real experience.

That said, some things to consider:

• The impedance and bandwidth of the signal of interest is very important to help the experts make a recommendation when they get here.
• I ordered some ADA4523s to play with back in April - I'm still waiting for delivery. Mouser reckons that I may get them by the end of the year...
• ADA4523 is a zero drift type op amp, so may pump out some noise at the input.
• If you need low noise and low input bias current at the same time, over the whole temperature range, then adding "superbeta" to your search terms may help narrow down the field

[Andreas]

Hello,

Electrometer grade amplifiers with +/-15V supply are rare these days.

So good old

AD549S

would fit all your requirements.

with best regards

Andreas

[AeroETech]

Quick disclaimer - I'm a metrology wannabe, i.e. enthusiasm for the subject but no real experience.

That said, some things to consider:

• The impedance and bandwidth of the signal of interest is very important to help the experts make a recommendation when they get here.
• I ordered some ADA4523s to play with back in April - I'm still waiting for delivery. Mouser reckons that I may get them by the end of the year...
• ADA4523 is a zero drift type op amp, so may pump out some noise at the input.
• If you need low noise and low input bias current at the same time, over the whole temperature range, then adding "superbeta" to your search terms may help narrow down the field

You're right about bandwidth - I absolutely should have included that.  Nothing super high speed - Perhaps 40Khz on the upper end.
I have a bunch of ADA4523s.  Be happy to share a few if you cover shipping.
Even a lot of the "suberbeta" parts, unfortunately, still have a very high Ibias at elevated temperatures.

Thank you for your reply!

[AeroETech]

Hello,

Electrometer grade amplifiers with +/-15V supply are rare these days.

So good old

AD549S

would fit all your requirements.

with best regards

Andreas

Thank you for that suggestion.  I hadn't thought about that oldie-but-goodie.

Unfortunately, 4uV pk-pk of noise from 0.1 to 10Hz and 60nV/rtHz to 100Khz isn't quite what I'm hoping for.  I'm really shooting for much lower than that.

The search continues.

Thank you very much for your reply!

[macaba]

The combination of your requirements is tricky, you will have to compromise one or more specifications to get better than ADA4523.
For example: MCP6V51 (single/dual/quad variants) gives less supply current at the expense of more noise (but perhaps acceptable?).
Wanting a stable input bias suggests to me that you might have high input impedance so be very cautious about using zero-drift amplifiers (charge injection spikes/current noise).

[AeroETech]

The combination of your requirements is tricky, you will have to compromise one or more specifications to get better than ADA4523.
For example: MCP6V51 (single/dual/quad variants) gives less supply current at the expense of more noise (but perhaps acceptable?).
Wanting a stable input bias suggests to me that you might have high input impedance so be very cautious about using zero-drift amplifiers (charge injection spikes/current noise).

You are absolutely correct.  High input impedence source, so of course zero-drift is touchy.

That's an interesting part.  I'll look over the specifications.  Have to admit, microchip is rarely on my list when i'm looking for high performance amplifiers, so I would not likely have come across that one.

Thank You!

[AeroETech]

The combination of your requirements is tricky, you will have to compromise one or more specifications to get better than ADA4523.
For example: MCP6V51 (single/dual/quad variants) gives less supply current at the expense of more noise (but perhaps acceptable?).
Wanting a stable input bias suggests to me that you might have high input impedance so be very cautious about using zero-drift amplifiers (charge injection spikes/current noise).

You are absolutely correct.  High input impedence source, so of course zero-drift is touchy.

That's an interesting part.  I'll look over the specifications.  Have to admit, microchip is rarely on my list when i'm looking for high performance amplifiers, so I would not likely have come across that one.

Thank You!

I was getting pretty excited about that part.  Then I got to input bias.  +/-4nA over the full temperature range.

Still appreciate the suggestion - I'm going to keep that part number in my back pocket.

[macaba]

Assuming you've looked at AD and TI, another example of an op amp from a less-known brand: ISL28110.
(warning: note the difference of input bias current at high temperature on +/-15V supplies between the single and dual channel version!)

Also OPA140 (though it specifies 3nA max which feels like an arbitrary n-sigma limit, there's no typical value stated, and no charts in the datasheet so might be worth characterising it)

Edit: I should point out this could be an XY problem - we may not be able to give a better answer without knowing the full circuit.

[chuckb]

The LT1012 has less than 100pa bias over -55C to 125C (Superbeta) with 0.5uVpp low frequency noise. The current draw is 0.5ma.
update -
Or the LT1008 for higher speed.

[Kleinstein]

The LT1012 has less than 100pa bias over -55C to 125C (Superbeta) with 0.5uVpp low frequency noise. The current draw is 0.5ma.
update -
Or the LT1008 for higher speed.

There is however a catch: 125 C operation is only for the obsolete metal case version.

There are generally not that many OPs specfied up to 125 C at all.

The specs for the maximum leakage / bias can be quite loose / unspecific. So many parts will actually perform quite a bit better than the guarantied specs. If deparate is many be worth considering selected parts.

For the rather high power ADA4523 one would have to consider self heating. So the actual temperature may be considerably higher than for a lower power OP-amp that could run closer to ambient.

With the microchip AZ OP-amps one has to keep in mind the slightly different definition used for the bias current:
The normal way is to consider the larger of the 2 input currents (or the average of the absolute values) as the bias.
Microchips seems to use the average of the 2 input currents (including sign) for the bias.  With the AZ op amps the 2 input currents tend to be of opposite sign and largely cancel out, especially at lower temperature. This different definition gets obvious with offest current specs way higher than the bias.
For the high temperature bias this is less of an issue, as this is more like leakage current of the protection diodes.

[Conrad Hoffman]

Can anything be gained using a composite design with a discrete front end?

[Kleinstein]

Discrete JFET may get a low bias in the sub nA range, but only the smaller type like J201 or simiar. So this would not be that low in noise. One may still need to screen and adjust the offset.
Chances are the low frequency noise is relatively high. A larger JFET and some input current compensation may work, but would be quite some effort.

There is a chance to get low bias by selecting super beta OP-amps, like the OPA207, ADA4077 - the maximum bias is still to high but typical values look good. So it is not about finding a few extra good ones, but more like making sure not to have especially poor ones.

[Gyro]

...
There is a chance to get low bias by selecting super beta OP-amps, like the OPA207, ADA4077 - the maximum bias is still to high but typical values look good. So it is not about finding a few extra good ones, but more like making sure not to have especially poor ones.

That would be the traditional way of doing it. With a temperature compensated bias current compensation network, old school DVMs used to be able to get sub 50pA (and 0.2uV/'C).


EDIT: I was thinking of the old Nat Semi / TI AN29... https://www.ti.com/lit/pdf/snoa624  (Fig 16).

[TimFox]

With automatic test equipment, it takes a long time to measure a very low bias current, so the manufacturers tend to guarantee a value that is quite a bit larger than the typical value, since they can verify that larger current.
As discussed above, the super-beta op amps tend to have lower bias currents than JFET models at high temperature.
You may have to socket the components and reject ones that are much worse than typical.

[chuckb]

The Analog Devices Web site lets you select low bias opamps then sort on operating temperature. Stay away from JEFTS if you want high temp, low bias operation.
https://www.analog.com/en/parametricsearch/11090#/d=s14|s25|2687|4500|4501|4502|300|4101|4108|2839|2840|s3

For full 150C specified operation, 50ua operating current and less than 1na bias you can look at the ADA4097 or ADA4098 for lower noise.

Or ADA4077, higher operating current.
or ADA4098, don't use the "Over the Top" feature, the bias currents increase 1000x.

Several choppers are specified to 125C.
LTC2057
ADA4522

Very few signal filter capacitors will have low leakage at 125C.
TDK or AVX COG caps will have pa leakage current at 60C. They are available up to 0.22ufd.
Next look at Panasonic ECQ PE Caps for up to 10ufd. Use a cap with a voltage rating of 10X (or more) of your operating voltage. This will keep the high temp leakage current low.

For some tested caps see -
https://www.eevblog.com/forum/metrology/film-and-ceramic-capacitor-leakage-current/msg3190618/#msg3190618

I will be updating this list soon with the results of the last year of capacitor leakage testing.

[miro123]

There are plenty of Opamps that will satisfy your criteria, the problem is IC availablity.
I have few question
- what is the source impedance. - ADA4523-1 is optimized for low impedance source - 1pA/sqrt(Hz)
- what type of noise must be low, 1/F noise , white noise or both?

As kleinstein mentioned - many Opamps has lower Ib as specified.
Short list
- OPA182 /2182 4182 - 3x lower current consumption and 6x lower current noise
- opa140/ 2140 - if ultimate DC performance is not required
- opa 196/2196/4196 - great RRIO
l have been using those components in different projects, unfortunately TI delivery time now is >40..60weeks

[AeroETech]

Thank you very much to everyone who has replied.  You've all given some good ideas.

I'm going to tear through the datasheets and decide which one to prototype up.

Appreciate all the input!