Lowest noise OP amps for low frequency, low level AC coupled signals.

[janaf]

I made this graph for some of the lowest noise OP amps available for

- Low frequency i.e 0.1-10Hz
- AC coupled signals
- 1000x gain 10k & 10R

The noise is SPICE calculated.

- LT6018
- OPA828
- OPA189

LT6018: The new king of low noise bipolar land, after 30+ years with LT1028. Needs less than 100 ohms input impedance and a 22000 uF input capacitor it maintains around 6nVrms in 0.1-10Hz. There are a few others like the LT1007 that would be better around 1kOhm.

OPA828: A JFET input amp with very respectable LF noise data. With 1Mohm and 2,2uF in, 50nVrms 0,1-10Hz.

OPA189: An auto-zero (chopper) amp, less than 20nVrms 0.1-10Hz with 100K input impedance. I also plotted quad of these paralelled, 33K in via 67uF should give less than 10nVrms 0.1-10hz.

The LT6018 was modelled with LTSpice. The OPAs where modelled in TINA which has a very handy tool for noise accumulated over frequency.

Any others that I missed, that really challenge these?

I hope you find the graph useful.

[SilverSolder]

How much can we trust SPICE modelling of noise compared to real world measurements?

(This is not a criticism of SPICE which I love to use, I just don't understand how "good" it is on this front)

[chuckb]

Does your noise model include the Johnson Noise (white noise) for R1? With a 1 Meg resistor the Johnson noise density will be something like 130nV/rt Hz. In a 10Hz BW would that be 400nV rms? So, for low noise you really want to stay with a lower resistance.

[RandallMcRee]

Any others that I missed, that really challenge these?

ADA4625 has lower noise for JFET input (0.15 μV p-p, 0.1 Hz to 10 Hz) than opa828...
Also would be interesting to compare the ADA4522-1 which I think is comparable or superior to the opa189 in the chopper realm.

Thanks!
Randall

[SilverSolder]

In tests done by @chuckb of this parish, the ADA4522-1 did not perform as well as OPA189:

https://www.eevblog.com/forum/metrology/low-frequency-noise-of-zero-drift-amplifiers/msg2093356/#msg2093356

[David Hess]

A spot noise graph would show the contribution of 1/f noise versus wideband noise.  The higher 1/f noise of the FET input parts could be suppressed with a parallel chopper stabilized amplifier without introducing more wideband noise although I have only done this in DC coupled applications.

Discrete bipolar and FET input stages are especially suitable for low noise AC coupled applications but there is a lot to be said for the simplicity of using an integrated part.

Note that the non-inverting configuration produces additional distortion products from limited common mode rejection and non-linear input impedance which may be undesirable in low noise applications.  Some operational amplifiers take special steps to avoid the especially pertinacious non-linear input capacitance problem like the LT1169 and parts like this may be desirable in some non-inverting applications despite higher noise.  The alternative is changing the circuit topology to suppress these sources of error.

[janaf]

I'll get back about that when I've done some measurements, for a point here and there.
It would be very time consuming, long measurement times plus the requirement to measure with different cap/resistor values... There's also quite a spread in real world from device to device.... See the graph as ballpark values showing what is realistic to aim for.

How much can we trust SPICE modelling of noise compared to real world measurements?

(This is not a criticism of SPICE which I love to use, I just don't understand how "good" it is on this front)

[RandallMcRee]

In tests done by @chuckb of this parish, the ADA4522-1 did not perform as well as OPA189:

https://www.eevblog.com/forum/metrology/low-frequency-noise-of-zero-drift-amplifiers/msg2093356/#msg2093356

Sure, but chuck's testing (and this simulation) are done with a pretty specific test setup. It does not take much to flip the relative rankings around. So, in my view, best to keep all of the contenders in the mix. For example, the OPA189 has higher bias current than the ADA4522. So, if that is important, and it often is, you might find yourself choosing the ADA4522.

[SilverSolder]

In tests done by @chuckb of this parish, the ADA4522-1 did not perform as well as OPA189:

https://www.eevblog.com/forum/metrology/low-frequency-noise-of-zero-drift-amplifiers/msg2093356/#msg2093356

Sure, but chuck's testing (and this simulation) are done with a pretty specific test setup. It does not take much to flip the relative rankings around. So, in my view, best to keep all of the contenders in the mix. For example, the OPA189 has higher bias current than the ADA4522. So, if that is important, and it often is, you might find yourself choosing the ADA4522.

Understood and agreed.  I guess Chuck's testing was focused on ultimate low noise performance below the usual 0.1Hz test range, which seems interesting for DC measurement applications and is a spec that we don't normally see in datasheets.  Having to deal with the 70pA bias current is definitely a minus for the OPA189, but it seems a pretty unique part if ultimate low frequency noise performance is important to you (as long as other factors don't outweigh this benefit, as you said).
 

[Kleinstein]

For the low frequency part the OPA827 is still a little better than the OPA828. Another option is the OPA140.
For the low current noise JFET based OPs there is the option to use a few of them in parallel.

From some point on the AZ OPs can be lower noise, especially if the bias current is not a real issue in a pure AC application.

The BJT based OPs are low voltage noise, but the current noise also usually goes up with an 1/f contribution. Quite often the 1/f cross over for the current noise is higher than for the voltage noise. So the optimum source impedance is even lower in the low frequency range. So in the LF range even 100 Ohms is still on the high side for the LT1028.
The AD8676 seems to be a relatively good BJT based OP, with not that much low frequency current noise.

However one problem with the data for the current noise is that there are two noise currents that are partially correlated. So tests with the same resistor at both inputs may give results that can be misleading in cases where the high impedance is only at one input.

AZ OPs noise, especially the current noise may depend on the capacitance at the input.

[janaf]

I did do the ADA4522, ADA4622 and they ended up a bit higher than OPA189. I didn't plot all devices I calculated, only the best in this particular setup. 

[janaf]

Discrete bipolar and FET input stages are especially suitable for low noise AC coupled applications but there is a lot to be said for the simplicity of using an integrated part..

That's pretty much why I made this graph. As far as I know some discrete input AC coupled JFET amp can achieve sub 10nvrms 0.1-10Hz while having (edit) 100Meg+ (edit) input impedance (Levinzon and others) But quite a lot of work, calibration, stability issues......would go into that.

[David Hess]

However one problem with the data for the current noise is that there are two noise currents that are partially correlated. So tests with the same resistor at both inputs may give results that can be misleading in cases where the high impedance is only at one input.

I should have mentioned that for bipolar input parts.  Low noise operational amplifiers intended for audio applications do not include input bias current cancellation partially because of the noise it adds.  So even under ideal conditions in an AC coupled application, a discrete bipolar design can achieve lower noise than an integrated part which includes bias current cancellation.

Is there an equivalent to any of the lowest noise bipolar precision parts which does *not* include bias current cancellation?  I have looked a couple times but never found any.  I suspect that market is too small so discrete designs are used for that kind of performance.

[janaf]

For the low frequency part the OPA827 is still a little better than the OPA828. Another option is the OPA140.

You are right, OPA827 and OPA140 are both a little better than OPA828 in this setup, see pic.

[Kleinstein]

However one problem with the data for the current noise is that there are two noise currents that are partially correlated. So tests with the same resistor at both inputs may give results that can be misleading in cases where the high impedance is only at one input.

I should have mentioned that for bipolar input parts.  Low noise operational amplifiers intended for audio applications do not include input bias current cancellation partially because of the noise it adds.  So even under ideal conditions in an AC coupled application, a discrete bipolar design can achieve lower noise than an integrated part which includes bias current cancellation.

Is there an equivalent to any of the lowest noise bipolar precision parts which does *not* include bias current cancellation?  I have looked a couple times but never found any.  I suspect that market is too small so discrete designs are used for that kind of performance.

For the precision parts the bias cancellation is kind of a requirement. The extra circuit is not a big deal anymore - so hard to find some.
Even the low noise audio amplifiers (e.g. OPA1611) tend to include bias cancellation as otherwise the bias would be rather large.
Too keep the current noise reasonable it may take super beta transistors, and these usually require a kind of bootstrapped cascode to use them. So not that attractive in discrete form.

It looks like even in the low frequency range good JFETs are better than BJTs, when it comes to high impedance - so there is likely not that much market for discrete BJT solutions and such BJT pairs. So if discrete it's more like JFETs anyway.

With lot's of effort a discrete build copper amplifier may also be an option, as it can use a JFET based amplifier instead of the usually CMOS based in AZ OP chips. So no more need for a high chopper frequency to get low noise, especially if a low BW is OK.

[janaf]

Added OPA27 which has input bias cancellation and it performs very well here! Interesting, didn't know about this....

OPA1611 was mentioned, I had it in there originally but removed it as it wasn't top notch in _this_ application. However it seems one of the very best in terms of THD. I also saw somewhere that it's virtually identical to OPA211. Anyway, I put the OPA1611 back in th plot. 

[Inverted18650]

I'll get back about that when I've done some measurements, for a point here and there.
It would be very time consuming, long measurement times plus the requirement to measure with different cap/resistor values... There's also quite a spread in real world from device to device.... See the graph as ballpark values showing what is realistic to aim for.

How much can we trust SPICE modelling of noise compared to real world measurements?

(This is not a criticism of SPICE which I love to use, I just don't understand how "good" it is on this front)

It may take a full year (or more) to test these all in real world environments, with all the variables, but that would be a great paper!
I am retired and almost interested enough to give this a go. Seems the TI and Analog Devices clans are split pretty evenly among their respective sides. Only real world data will due. The question then becomes, what is a fair and practical test that can provide repeatable data within for each DUT?

edit*

Adding: we need a basic building block circuit, with say ten (10) variations (for caps, etc), that we can plug and play each of the 20 or so DUTs and record the data for accurate comparisons. If this were the case, I believe I would love to be involved.

[SilverSolder]

Perhaps a board with reed relays to choose between various fixed caps and resistors, so the testing could be automated if desired?

That way it would be possible to subject each amp to the same GPIB battery of tests...

[Andreas]

Needs less than 100 ohms input impedance and a 22000 uF input capacitor it maintains around 6nVrms in 0.1-10Hz. There are a few others like the LT1007 that would be better around 1kOhm.

Hmm,

the question for me is: how usefull are source impedances below 1K Ohm. (when building a LNA for noise measurements).
Typical Zeners (1N829) have 10-20 Ohms zener impedance.
The unbuffered LTZ1000 is also in the some Ohms range.

So if you load such a zener by 100 Ohms you already measure up to 20 % lower noise than actual available.

So typical I want to stay at 1KOhms as input impedance which is the domain of LT1007/LT1037/OP27/OP37.
A large part of the LNA noise floor is also contributed to (the AC-part of) leakage currents of the input coupling capacitor.
(if not using foil capacitors with thousands of uF).

with best regards

Andreas

[Inverted18650]

Very nice catch.
As we move foward designing the test circuit, our guidelines will include a minimum 1k input imp. to ensure noise floor tolerance for every DUT.

[Kleinstein]

Some BJT based OPs need a low source impedance to get a very low noise level or more accurate there best noise figure. However this does not mean that the amplifier is loading the source with this impedance. The input impedance of the amplifier circuit can be higher.

Ideally the resistor to ground after the input capacitor is larger by something like a factor of 10 than needed to get the low frequency limit at 0.1 Hz. This reduces the loading to the source and also avoids much of the resistor noise in the transition region. The exact low frequency cut of that has to be set in a later stage, behind the initial amplification. This could also be in the digital domain, if the signal goes to an ADC anyway.

Still it is problematic to have OPs that need a low impedance source, as it needs a huge capacitor, that can represent there own problems.

Using Spice models to compare OP performance is a little tricky - not all models are really accurate when it comes to details like current noise.
Usually the data-sheets are still a little more reliable.  From the DS performance the OPA27 is not that good.

Testing low noise in real life is not that bad. There is another parallel thread on such tests on AZ OPs. It's mainly setting up an amplifier for 2 input resistors (one near zero and one suitable for the OP, so that current noise dominates). The behavior with a different source impedance can than be calculated to a good approximation. Only AZ OP may need extra test with small capacity to ground, as there can be effects from short input current spikes, that makes them possibly less predictable. The noise test can be done with DC coupling, so no need to wait for a coupling cap to settle at the input.

To get noise data down to 0.1 Hz one might want to take something like 10 times the minium 10 second interval to average over some random parts like popcorn noise. Still this would be something like a few minutes to measure and maybe 30 minutes for thermal settling.  Just looking directly for peak to peak noise is not very effective. The better way today is more like an FFT to get the noise spectrum.

[SilverSolder]

The 0.1Hz lower limit of noise testing seems too high a frequency, somehow...  one gets the sense that this limit was chosen because it was possible to measure with reasonable equipment, rather than what we really need to know.

For example, if you have an instrument that you want to stay stable for hours or days on end, you really need to know the noise performance down to the milliHertz or even microHertz range.

[janaf]

It may take a full year (or more) to test these all in real world environments, with all the variables, but that would be a great paper!
...If this were the case, I believe I would love to be involved.

Glad this caught on!

To provide comparison, I made a new plot with a line I called the "0110 Axel line".
It's a spline on log-log diagram, with just three points, preliminary values at
 
100R 4nVrms
100K 10nVrms
1G 500nVrms

The 0110 Axel could be a baseline to relate to and plot results against. Op-amp results below the line would be Breaking News, or errors 

I also extended the OPA868 and at very high input resistor values, it beats the OPA827 by about as much as the OPA827 is better at low resistances. So the OPA828 remains in the plot.

[janaf]

On a parallell note, any leakage through the input cap would destroy the results. For high impedance inputs, tiny cap values require v-e-r-y low leakage while for the lowest noise amps, low input impedance requires large caps. Which E-lyths
are good enough?

I once stared doing measurements on various caps but results where not very clear due to
- leakage through the voltmeter, even it has 1 Gohm input, that was way too much.
- temperature variations on the cap made results very shaky.

I am thinking of setting this up again. would use
- Mechanical relay MUX
- A fA leakage input amp buffer (LMC660?) for the DMM, repeatedly checked vs a reference.
- A temperature controlled box.

Attaching a pic of a good 200uF polypro. It's BIG.

[TiN]

I have measured few wet-slug electrolytic caps, same as Jim used in his preamp.
They shown leakage <4 nA at 10V bias. Only problem is the cost.