Attempt at a 100 V/V gain low signal amplifier - NE5532 offset correction

[Verdefluox]

Hi folks,
this is my first attempt at trying to amplify low signals (I'd start with something in the order of 100's uV).
At my direct availability I have the NE5532 (for low noise) and the OP07 (for precision) sitting around, hence I thought to combinate them. Just for trial, I've set my flat bandwidth to around 15kHz.
While trying different configurations, the easiest I could find in which I could both zeroing the offset and (I hope) maintaining a low noise has been like the following:
https://www.renesas.com/us/en/document/apn/r13an0016-reducing-offset-wideband-amplifiers
which is basically an inverting amplifier with limited BW and zeroing offset correction.
I also attached a similar schematics I came up with. Please note that I cannot evaluate extensively the performance of this circuit, since I would not build it if I already managed to amplify such low signals.
Nevertheless, my primary goal is to get a reasonable low noise, but I'm unsure regarding the OP07 offset correction. I know that the OP07 has higher noise density than the NE5532 (and with same BW means higher pp noise value), but I think I'm missing something: if the OP07 zeroes the offset (with its noise filtered and with 0 dB gain after the LP), doesn't this make the noise amplified by 101 by the NE5532 **the one of the OP07**, hence making useless the use of a low noise NE5532?
In other words: I don't understand at the NE5532 + input which noise gets then amplified, and from this I don't understand if I'm ruining the NE5532 good noise characteristics by using the OP07 **before it**.
Btw, I'm assuming that in general the signals I'm interested in are >> than the noise I'm getting from the circuit. I also performed some calculations but before proceeding further I'd like to ask if this configuration is somewhat valid.
If you have any improvements regarding this circuit please let me know, and I sincerely hope my questions are clear (I know that maybe they're not formulated in the best way possible, but I tried my best).

[Kleinstein]

The circuit is pretty useless, as the input impedance in only 10 ohm. Also the RC values don't work with too little BW and too little filtering for the OP07.

It would make more sense to use the NE5534 in non inverting configuration to start with.

A big question is if the amplifier needs to work all the way to DC or can be DC coupled - this case would be much easier, e.g. with a DC servo.

[Verdefluox]

It would make more sense to use the NE5534 in non inverting configuration to start with.

Well, I obviously considered that before, but I struggled to find a configuration that makes me correct the NE553x offset (I know that manually with a pot is a possibility, but for the sake of learning and maybe improving in the future a precision op amp seemed a good idea). Do you have any suggestions?
I also read that I could combine different op amps with a composite configuration, like this
https://www.engineering.com/story/composite-amplifiers-part-four-using-composite-amplifiers-to-tame-dc-offset-and-noise
but as far as I understand, the noise is determined by the 1st stage (where I would have the precision, not the low noise one).

The circuit is pretty useless, as the input impedance in only 10 ohm. Also the RC values don't work with too little BW and too little filtering for the OP07.

I understand the impedance is low, but I thought that, given the signals I'm interested in are at most in the 10's of mV that shouldn't have been a problem. Also, what's wrong with RC values? I just set them accordingly to the band I'm interested in. If it's all descending from the low values, is basically increase the R values (and also decreasing C) what you're implying? Idk, maybe by x10?

A big question is if the amplifier needs to work all the way to DC or can be DC coupled - this case would be much easier, e.g. with a DC servo.

My bad for not specifying. I'm interested in amplifying DC too.

[magic]

Your circuit is different than the one in the appnote. I don't understand the point of adding R2, and R1 should be connected as shown originally because it's the IN- of U2 which you want to precisely compare against zero and bring it down by tweaking IN+.

R2 and C2 can work to remove U1 noise from U2 input. Alternatively, C2 could simply be a resistor for a constant and predictable attenuation at all frequencies and no worries about stability. U1 works at approximately unity noise gain at higher frequencies, so 10nV/rtHz on its output. Simple 10x division brings it down to irrelevance (5nV/rtHz in the 5532).

Note that NE5532 noise increases at low frequencies and it likely exceeds OP07 below some point, maybe a few Hz.

[Verdefluox]

Your circuit is different than the one in the appnote. I don't understand the point of adding R2, and R1 should be connected as shown originally because it's the IN- of U2 which you want to precisely compare against zero and bring it down by tweaking IN+.

Oh damn, I'm so sorry. I drew the circuit quickly just for the sake of giving a quick schematic and I made a mistake. Should be corrected now. Btw, I added R2 on U1 just to have a flat response within my BW of interests.

Simple 10x division brings it down to irrelevance (5nV/rtHz in the 5532).

Hold on, I think I lost the 10x division you're talking about. What are you referring to? Adding a simple divider?

[David Hess]

You have the right idea but a better example from Jim Williams is shown below from Linear Technology application note 21 - Composite Amplifiers.

Besides preventing overdrive and improving stability, the attenuator at the output of the LT1012 divides its noise contribution.  The 10k - 300pF combination sets the break point for the noise contribution between the LT1012 and LT1022 but this does not matter with the attenuation.  There is no reason to include your R2.

Your operational amplifier selections will work fine, but adjust the impedances accordingly.

[Kleinstein]

There indeed seems to be no easy way for a non inverting version.

The low impedance is a problem as the gain than depends on the source impedance. With only 10 Ohm this would be a 10% error for a 1 ohm source impedance.
One can however go higher, but resistance also adds to the noise. More than some 1 K is thus not attractive for low noise.

[blackdog]

Hi,

First this, always use the "A" version of the NE5534 or the NE5532 so you will have the best noise behavior.
You can also drive a NE5534A on one of its trim connections with an offset opamp....

But it might be even better for your 100x amp to get an LT1037a, it has low noise which is at better than the NE5534a,
low offset voltage and a 1/f corner of 2Hz, the open loop bandwidth is enough for your 15KHz bandwidth.
Then you can do it with one opamp.

But since I don't quite know what your source is and/or what needs to be driven it's hard to advise further.

What David Hess shows of the Jim Williams application is also a possibility and there are many others.
Which one is the good one, usually the KISS principle applies.

But if you want to gain knowledge, try different circuits, make a list of how each circuit works, if you run into something you don't understand, then there is plenty of help on this forum. 

Kind regards,
Bram

[David Hess]

There indeed seems to be no easy way for a non inverting version.

Sure there is.  I have used it.  Use the same configuration but drive an offset null pin.  It requires a pair of RC time constants with one around the integrator and another at the non-inverting input.

[Kleinstein]

If the OP-amp had pins for an offset trim, getting a low offset is easy. With the offset trimmed also the NE5534 (kind of single version of the 5532) has reasonable low offset drift - not as good as the OP07 with a trim, but should still be OK as most of the offset drift is proportional to the offset. So the offset trim also improves the drift.

[Verdefluox]

Use the same configuration but drive an offset null pin.  It requires a pair of RC time constants with one around the integrator and another at the non-inverting input.

This actually seems a good solution, and I discarded it since I had no NE5534 at hand (I read the AN21), but maybe since I have to buy something anyway I can go with better opamps directly. Anyway, something like the 1st attachment can work (apart for some additional caps)? Some questions:
- what a cap at the NON INVERTING input?
- the RC of the integrator should be compatible with my bandwidth, or it doesn't matter since it is about the offset alone?
- the offset resistors are taken almost arbitrary in this case. I went with values based on comparison with the LT1028 circuit, but since in the datasheet of the NE5534 suggest to use 100k pot for offset correction, I don't know if I should increase them or if these are fine.
- as in 2nd attachment, I thought to add a LP filter to attenuate better the HF. Unless there's something else I'm missing I don't think is a bad idea, what do you think?

In the end, maybe coming back to a non inverting configuration in the end pays more, also given that I'll have a lot less problem with the input impedance.

[blackdog]

Hi,

Verdefluox, your last diagram with the offset correction via the balance connections doesn't go well that way....
Look in the datasheet how the NE5534 is constructed, with your proposal to put 100 Ohm in parallel to the collector resistors a large part of the openloop gain disappears.
So you put 100 Ohm in parallel to 12K....

Again as advice, look at the LT1037, good noise behavior enough bandwidth and also reasonably low bias currents.

The LT1028 is clearly better because of its low noise behavior, but only with very low source impedances.
Again, the bias current largely determines the overall noise behavior if your source impedance exceeds 50 ohms.

As I indicated without all the info it is difficult to give good advice.

Greetings,
Bram

[Verdefluox]

Look in the datasheet how the NE5534 is constructed, with your proposal to put 100 Ohm in parallel to the collector resistors a large part of the openloop gain disappears.
So you put 100 Ohm in parallel to 12K....

Oh right. I honestly went with those values because I didn't have a direct comparison with the LT1028 (in the datasheet I don't see an internal schematic), so I assumed that for a full range adjustment I had to lower everything, but honestly I don't know. In this case, which values should I use? I assume that I have somehow to maintain that value, hence use something greater than 12k (like 50k?), so that in parallel doesn't change much. But in that case, I think that in order to maintain the magnitude order, I guess that R5 should go up quite a bit, hence I'm not sure about that too. Can you suggest any value?

EDIT: My bad, saw the LT1028 now, but unfortunately I don't see the resistors split in two in the NE5534. Guess I'll have to go up then.

As I indicated without all the info it is difficult to give good advice.

I don't know if this is what you're looking for, but I thought about this project because I wanted to see the my power supplies ripples (since I have a scope with not that high resolution). In general, however, I thought about it as a general purpose amplifier.

[David Hess]

- what a cap at the NON INVERTING input?

I usually see the same R1 and C1 at the non-inverting input of the OP07, but with the capacitor going to ground.  This might only depend on what kind of amplifier the OP07 is and how it handles high frequencies.  If the OP07 was a chopper stabilized amplifier, then the RC network at the non-inverting input is required to maintain a low AC input impedance at its non-inverting input.

- the RC of the integrator should be compatible with my bandwidth, or it doesn't matter since it is about the offset alone?

The RC constant determines the frequency break point between which operational amplifier is controlling the output.  It matters because it also determines which operational amplifier is contributing noise to the output, since the OP07 has lower low frequency noise than the faster amplifier.  In practice this is only going to matter below 10 Hz unless the fast operational amplifier has a lot of low frequency noise.

There is one RC time constant where low frequency output noise will be lowest.  It can be calculated, but in practice I would only use this as an estimate and find it empirically because too many datasheets have questionable noise specifications.

- the offset resistors are taken almost arbitrary in this case. I went with values based on comparison with the LT1028 circuit, but since in the datasheet of the NE5534 suggest to use 100k pot for offset correction, I don't know if I should increase them or if these are fine.

When I last did it with the LT1028 and a chopper stabilized amplifier, I first measured the sensitivity of the offset null input of the LT1028, and then scaled the resistance network so that over the output range of the low frequency operational amplifier, the LT1028 had a reasonable offset adjustment range.

For example with +/-15 volt supplies, I might want the low frequency amplifier to have a maximum output of +/-7.5 volts for a change in offset voltage of +/-200 microvolts to cover the entire range of possible offset values.

I was working with the non-inverting version of this circuit because I extended it to make a fully differential version using twice as many operational amplifiers.  Performance with the LT1028 was incredible, but of course only with a low resistance source.  I was able to measure the resistance of the source from its Johnson noise to within a couple of ohms, although this was not the intended use.  Incidentally the noise specifications of the Linear Technology datasheets were right on, but I would sure not trust anybody else.

[blackdog]

Hi,

Try building this first to do some measurements on the noise and interference behavior of Power Supply's.

Build it in a metal box, it can be powered from 2x a 9V battery.
And definitely don't use SMPS near this or any other preamplifiers.

This is a good and fun way to get started with measurement amplifiers.
If you want more, it gets complex quickly, often you need multiple filters for a good estimate.
But you can also use the FFT function of a scope or if you have a good geluidskaart with some software it is also very pleasant measuring.
This of course together with e.g. the measuring amplifier as shown in the picture.





Greetings,
Bram

[Verdefluox]

I usually see the same R1 and C1 at the non-inverting input of the OP07, but with the capacitor going to ground.  This might only depend on what kind of amplifier the OP07 is and how it handles high frequencies.  If the OP07 was a chopper stabilized amplifier, then the RC network at the non-inverting input is required to maintain a low AC input impedance at its non-inverting input.

Am I wrong or you're basically meaning a LP filter?

When I last did it with the LT1028 and a chopper stabilized amplifier, I first measured the sensitivity of the offset null input of the LT1028, and then scaled the resistance network so that over the output range of the low frequency operational amplifier, the LT1028 had a reasonable offset adjustment range.

For example with +/-15 volt supplies, I might want the low frequency amplifier to have a maximum output of +/-7.5 volts for a change in offset voltage of +/-200 microvolts to cover the entire range of possible offset values.

So you're basically telling me to respect this proportion and try to scale them up/down (possibly up, because of the parallel resistors)? The only doubt now for me is what practical limit R5 should be. Considering that in any way goes parallel to the 12k resistors inside, should not be a problem if for that I go 10x (or even more) higher for example, right?

Btw, I was thinking of an alternative: from the datasheet (attachment 1), I see that the OP07 is characterized at 0 dB gain, hence I'm assuming that I should be able to drive it as a unity-gain buffer. In that case, I could sacrifice another one to make an input buffer and going back to the inverting configuration, which is easier and should isolate the circuit from the source via the buffer. Something like my 2nd attachment, then I could adjust the values if needed. I mean, I know that right at the input I'm using the theoretically higher noise opamp, but its output doesn't get amplified, no?
EDIT: Please ignore the buffer pin nomenclature ("Vos1" and "Vos2"), they are obviously distinct from the other one. It's just for reference.

[Kleinstein]

If one buys new parts one could as well get a single OP, that can do DC all the way to 15 kHz with low noise, like the LT1037, OP37 or LT1028.
Especially with a manual offest trim one would likely not need an extra low frequency OP-amp.

The combination of OP07 as unity gain butter would add the noise of the OP07 and is thus not as attractive.
A NE5534 with additional OP for the low frequency part can work and would be mainly a thing if the 2nd OP is an AZ type to get very low DC drift. The resistors suitable for the DC trim are given in the datasheet and should not go lower - one may go higher to reduce the trim range to the actual needed range.

With the given parts of NE5532  (no offset adjustment pins) and OP07 the choices are more tricky. One could use the inverting configuration with some 1 K and 100 K as gain setting resistors and than the circuit from the LT appl. note. The divider after the OP07 helps to reduce the op-amps noise and also mulpies the integrator time constant, so one can get away with not too large a resistor and capacitor and still get a cross over of some 10-50 Hz, as aabout the point where the OP07 noise gets lower than the NE5532 noise. If in doubt run the circuit through a simulation (e.g. LTspice).
One could try a non inverting version. Here the OP07 part would however effect the low frequency gain and one thus need well tuned resistors/capacitors to compensate and get a flat frequency response and the same gain for DC and the higher frequencies.

[Vovk_Z]

You may consider use just one OP37 or two OP37 in series (what frequency range do you need?). Non-inverting OP37 amplifier is quite capable, I only not sure about its noise capabilities (they say Low Noise, 80 nV p-p (0.1 Hz to 10 Hz), 3 nV/÷Hz @ 1 kHz). They are faster than OP07 and are precision type with trim too. They are available almost as OP07 I guess and cost not too much.

[blackdog]

Hi,

The elephant in the room.
A gain of 100x at a symmetrical supply of 5V and you have to take into account 2x 1.5V commonmode at the output of the NE5534 opamp.

That leaves you with a maximum of about 7V toptop and then divide this by 100x gain.
Then the output clips at 70mV Top Top DC at the input, so how sensible is it to have a DC coupled input.

Sincerely,
Bram

[Verdefluox]

If one buys new parts one could as well get a single OP, that can do DC all the way to 15 kHz with low noise, like the LT1037, OP37 or LT1028.
Especially with a manual offest trim one would likely not need an extra low frequency OP-amp.

Obviously. Initially I also thought about an higher BW (100k), but I resized my goal for the sake of learning/building something valid within its (limited) specs. Sorry, I don't want to insist on using this components and reject newer parts out of nothing, but before buying something new I wanted to see if I could get away with something widely available and well known.

With the given parts of NE5532  (no offset adjustment pins) and OP07 the choices are more tricky. One could use the inverting configuration with some 1 K and 100 K as gain setting resistors and than the circuit from the LT appl. note. The divider after the OP07 helps to reduce the op-amps noise and also mulpies the integrator time constant, so one can get away with not too large a resistor and capacitor and still get a cross over of some 10-50 Hz, as aabout the point where the OP07 noise gets lower than the NE5532 noise. If in doubt run the circuit through a simulation (e.g. LTspice).

So basically something like this (see attachment)?
Out of curiosity, in this case couldn't I also try to use 10k/100k, hence splitting the design in two stages? I'm thinking about this because
- with an inverting (simpler), I get an increase of the input impedance
- the following stages can also have their offset adjusted manually (since the hardest part has already been done) and being also non-inverting (so I have both x10 and x100, both with the same polarity). Something like the offset corrections described here:
https://www.analog.com/media/en/training-seminars/tutorials/MT-037.pdf

That leaves you with a maximum of about 7V toptop and then divide this by 100x gain.
Then the output clips at 70mV Top Top DC at the input, so how sensible is it to have a DC coupled input.

For now I just ignored the supply (5V is just an arbitrary value, I can also use 2x9V batteries). Regarding the output clipping, didn't think about that! Mind however that my goal was to see clearly signals in my scopes in the range of mV-100's of uV. I never thought about amplifying something more than 10 mV, but sorry for missing this specs in my previous posts.

[Kleinstein]

The last circuit is alread relatively good. I thing R2 shoudl be about 1 x larger to get less noise from the OP07 and move the cross over lower.

One could change to 2 stages with a gain of 10. This can help with the bandwidth, but it would not really allow for a higher input resistane with the inverting configuration. The 1 K resistor gives a noise of some 4 nV/sqrt(Hz) and thus about as much as the NE5532. In addion a larger resistor would also make the current noise more relevant.


For a more general use I would really prefer a higher input impedance and thus a non inverting configuration.  The extra resistor towards the FB divider can substiture for an offset adjustment, though it may be wrong polarity to use in the same way as the real offste trim part. It may need another OP to fix that, but there should also be a non inverting solution with an extra OP to correct the LF part.

[David Hess]

I usually see the same R1 and C1 at the non-inverting input of the OP07, but with the capacitor going to ground.  This might only depend on what kind of amplifier the OP07 is and how it handles high frequencies.  If the OP07 was a chopper stabilized amplifier, then the RC network at the non-inverting input is required to maintain a low AC input impedance at its non-inverting input.

Am I wrong or you're basically meaning a LP filter?

That is right; it is a low pass filter.

So you're basically telling me to respect this proportion and try to scale them up/down (possibly up, because of the parallel resistors)? The only doubt now for me is what practical limit R5 should be. Considering that in any way goes parallel to the 12k resistors inside, should not be a problem if for that I go 10x (or even more) higher for example, right?

Yea, you have to take a close look at exactly how the internal offset null circuit is configured.  For the LT1028, the offset null pins have a resistance of about 130 ohms to the positive supply, so anything done to drive them looks like a current in or out, and that works just fine.  When I made my measurements, I got a scale factor of microamps in or out of the offset null pin for a given change in input offset voltage.  Then I designed the external network to drive that current in or out, but it comes down to just a resistor or resistor divider.

[magic]

Circuit simulators can be very helpful in such situations.

For instance, I found that with your value of R1 there was additional noise at all frequencies, which I suppose makes sense because U1 voltage noise present at its inverting input mixes with the input signal through R1. Simply increasing R1 solves this problem. At low frequencies I found it helpful to increase C1, but it appears to be true that one can decrease R3 instead and the effect is almost the same (not shown, download LTspice and try it yourself ).


BTW, regarding this simulation:
LT1001 is Linear's OP07, very similar noise.
U2 is an ideal opamp, so noiseless.
The blue plot shows everything minus resistor noise, because simply plotting V(U1) doesn't work
This is total output noise, so divide by 100 to get input-referred.
At high frequencies the main feedback divider is the dominant factor, about on par with NE5532.

[Kleinstein]

It makes indeed sense to have R1 > R4, as there would otherwise noise current couled back to the fast amplifier. Too large a value for R4 would however add to the low frequency noise. 

[Verdefluox]

Yea, you have to take a close look at exactly how the internal offset null circuit is configured.  For the LT1028, the offset null pins have a resistance of about 130 ohms to the positive supply, so anything done to drive them looks like a current in or out, and that works just fine.  When I made my measurements, I got a scale factor of microamps in or out of the offset null pin for a given change in input offset voltage.  Then I designed the external network to drive that current in or out, but it comes down to just a resistor or resistor divider.

I understood your concept. Retaking the old non-inverting schematics (attachment, still the one with wrong values): in this case the only thing that doesn't convince me is the following: let's say that I change R4, R5, R6 and increase their value. Respecting some proportion with AppNote21 (without any calculations for now), I guess that I should go with R4,R6~15/20 kOhm at least, while for R5 I guess I would go for 100's of kOhm, if not approaching MOhm (to respect some proportion). I imagine that this is about the expected offset range that I want to correct. Now, even before doing calculations my doubt is: if such high values for the R5 value are involved, doesn't this somehow interfere with one of the two opamps (either the OP07 output, or the NE5532 input)?

I ask this because, in an almost stupid way, I know these old opamps don't work very well when they have such high resistors around, but I don't know if that is only when talking about the "commonly used pins" (+,-,out). Usually one doesn't touch the offset correction pins with "strange" configurations like these.
It's also true that when comparing the pure LT1028 datasheet and the correction offered by AN21 we fall in a similar situation (we want ~100's Ohms, but we are also putting 30k there), hence it shouldn't matter that much.
EDIT: For now, in all of this I'm quite neglecting the fact that, noisewise, I'm somehow introducing a 100 kOhm/MOhm resistor, but maybe it can be reused somewhere else.

Sorry if my questions here are quite convoluted, I see that there are also a lot of side-things in all of this.
Also, I think this is important because if I can go higher with R5 values without major drawbacks I could reconsider entirely this configuration.

Circuit simulators can be very helpful in such situations.

For instance, I found that with your value of R1 there was additional noise at all frequencies, which I suppose makes sense because U1 voltage noise present at its inverting input mixes with the input signal through R1. Simply increasing R1 solves this problem. At low frequencies I found it helpful to increase C1, but it appears to be true that one can decrease R3 instead and the effect is almost the same (not shown, download LTspice and try it yourself ).


BTW, regarding this simulation:
LT1001 is Linear's OP07, very similar noise.
U2 is an ideal opamp, so noiseless.
The blue plot shows everything minus resistor noise, because simply plotting V(U1) doesn't work
This is total output noise, so divide by 100 to get input-referred.
At high frequencies the main feedback divider is the dominant factor, about on par with NE5532.

Don't exactely know what to say but thank you. I didn't actually noticed that LTSpice had a noise analysis, but mind that I **barely** learned how to do an AC analysis, so..