[SOLVED] Deviation from expected behavior of sallen key opamp

[Audioguru]

Re: Deviation from expected behaviour of sallen key op amp

Your new resistor values of 5.1M are much too high for a lousy old LM358 opamp since its input bias current is typically 45nA and is positive-going (PNP input transistors) so the offset voltage is typically 10.2M x 45nA= 0.46V. If the opamp has the maximum input bias current of 250nA then the offset voltage will be 10.2M x 250nA= 2.55V.

 

[albbg]

Re: Deviation from expected behaviour of sallen key op amp

As already said by Audiogu the input bias current of the op-amp generates significant offset when large resistors are used. You could try to mitigate the problem inserting in the negative feedback a resistor of the same value of the series resistors you are using (i.e 5 .1Mohm). I mean: remove the short between output and inverting input of the op-amp and place between these two points a resistor of about 5.1 Mohm. The problem should disappear. In this case however you'll be prone to noise so you can insert a capacitor in parallel with this new feedback resistor. The value of the C have to be low enough not to significantly modify the transfer function.

 

[crutschow]

Re: Deviation from expected behaviour of sallen key op amp

...................
You could try to mitigate the problem inserting in the negative feedback a resistor of the same value of the series resistors you are using (i.e 5 .1Mohm). I mean: remove the short between output and inverting input of the op-amp and place between these two points a resistor of about 5.1 Mohm. The problem should disappear.
.............................

The added resistor should be equal to the total input resistance or 10.2 Mohm (the bias current path to ground) to get proper cancellation.

But note that cancels only the bias current common to both inputs, leaving the bias offset which is 30nA maximum.
This still leaves a maximum offset voltage of 0.3V.

 

[albbg]

Re: Deviation from expected behaviour of sallen key op amp

The added resistor should be equal to the total input resistance or 10.2 Mohm (the bias current path to ground) to get proper cancellation.

Sorry, I've understood the sum of the two was 5.1 Mohm and not each of them.

 

[D.A.(Tony)Stewart]

Re: Deviation from expected behaviour of sallen key op amp

When you wish to use high impedances, choose a better OpAmp such as the CMOS types with pA input bias currents or even much less.

If you are presently stuck with an LM358 bipolar Op AMP then recalculate the RC values to stay with ceramic XR7 or better and R with a tolerable input offset.

Keep in mind there are many suppliers of LM358 types and different specs for input bias current , Iib and offset current Iio , which also depends on rated temperature range. The wide temp are not worse at room temp, just a wider temp limit. Then there are differences in typ. and max.

E.g. ST Electronics rate the LM358A with Iib=20 typ, 50nA max @25'C
and Iio is around 10%~20% of that , namely 2 nA typ, 10nA max.

So ST chip with 0 Ohm from Vin(-) to ground the Vin offset created from Iib drop , V=IR= 10.2M* 50nA= 500mV(max) or =10.2*20nA=200mV (typ)

But with 10M added between Vin(-) and gnd. ( close to 10.2M) the offset created by mismatch is now reduced to 0.2MΩ * 50nA=10mV

10mV OK? with ST parts

If not use 1% parts with a lower value R. ( 5.1+5.1=10.2 thus 2% off vs 10M but 1% 10M increases "error stackup" to 3%.

 

[Audioguru]

Re: Deviation from expected behaviour of sallen key op amp

Sorry, I've understood the sum of the two was 5.1 Mohm and not each of them.

Then all of our offset voltages should be halved.

 

[mamech]

Re: Deviation from expected behaviour of sallen key op amp

As already said by Audiogu the input bias current of the op-amp generates significant offset when large resistors are used. You could try to mitigate the problem inserting in the negative feedback a resistor of the same value of the series resistors you are using (i.e 5 .1Mohm). I mean: remove the short between output and inverting input of the op-amp and place between these two points a resistor of about 5.1 Mohm. The problem should disappear. In this case however you'll be prone to noise so you can insert a capacitor in parallel with this new feedback resistor. The value of the C have to be low enough not to significantly modify the transfer function.

thank you for you advice. I did as you said and the DC offset disappeared, the real problem now is noise. please look at this:




I put 22pf parallel to the new resistance as you said, but nothing improved regarding noise. but now I think that I am on the right track and I understood the causes of all of these problems. thanks God.

- - - Updated - - -

When you wish to use high impedances, choose a better OpAmp such as the CMOS types with pA input bias currents or even much less.

If you are presently stuck with an LM358 bipolar Op AMP then recalculate the RC values to stay with ceramic XR7 or better and R with a tolerable input offset.

actually, I am not that stuck, but the fact that I will ask the students to make this circuit makes me biased towards lower cost. I am thinking now to make the same circuit with another opamp OP07, I see that the offset input bias current has typical value of 1.2 nA , so I think this can be fine in my application.
https://www.ram-e-shop.com/ds/ic/OP07.pdf

also there is another supposedly better (and more expensive ) CA3031, which has a term called input offset current ( I am not sure if it is the same as input bias current ) in range of 0.5 pA, if I understand, this is a good choice to solve the origin of such problem, is n't it?

https://www.ram-e-shop.com/ds/ic/CA3130.pdf

 

[D.A.(Tony)Stewart]

Re: Deviation from expected behaviour of sallen key op amp

It's a good choice to use and for teaching the details of all aspects of a CMOS Op AMP
1) read the datasheet 1st and understand at least the curves and Do not exceed page.. Each student gets 1 question min.
2) Teach how 1e12 Ohms input impedance picks up E-fields easily and demonstrate with a 1e7 scope probe without touching ground. THese e-fields using a finger can damage the inputs. ESD can damage any pin.
3) Teach how internally compensated Op Amps are different and why they are designed as integrators for unity gain compensation. ( to make them a first order open loop response to get 60 deg phase margin) You will need 56pF.
4) understand that even buffered CMOS drivers need to be current limited complementary drivers to prevent shoot thru,and thus are designed to be short circuited on output to ground, but this means the RdsOn varies from 1k to a few hundred Ohms for Vcc=5 to 15V as computed from slope at 1mA in Fig 22,23. Using R divider ratios, then they can understand why it doesn't go rail to rail with a 5k or 1k load.
6) for integrator designs or S&H, this chip has null offset pins unlike most these days, so with care and understanding of balanced input impedances , even the Iio and Vio can be tuned with care.
7) Noise can be a problem with high impedance antenna effects, so short jumpers, are preferred, because even putting finger near inputs with 10M will induce a voltage unless you are grounded. Finger tip is around 1pF minimum with smallest contact area and can increase to xxx pF so you can estimate the coupling X(f)/Zin

 

[Audioguru]

Re: Deviation from expected behaviour of sallen key op amp

The LM358 is very old. It is one of the first low power opamps.Old design and low power produce a noisy opamp.
high value resistors produce thermal noise. Why are your resistor values so high and why is your frequency so low?

The CA3130 opamp is also VERY old. It is the first Cmos opamp ands was designed by RCA about 50 (?) years ago and of course it is noisy.

The OP07 is also very old but it is better than the other two.

 

[D.A.(Tony)Stewart]

Re: Deviation from expected behaviour of sallen key op amp

I remember correcting an example in the datasheet for a staircase generator in EE labs in the early 70's using the excellent CA3130.

Intersil's datasheet CA3130 Aug 1 2005
Equivalent Input Noise Voltage eN BW = 0.2MHz, RS = 1M~10MΩ 23 µV
the use of the CA3130 is most advantageous
in applications where in the source resistance of the input
signal is on the order of 1MΩ or more. In this case, the total
input-referred noise voltage is typically only 23µV when the
test-circuit amplifier of Figure 7 is operated at a total supply
voltage of 15V
So for your LPF, noise is negligible. since BW is much lower and gain is unity.
Price at D-K $1.70 (100)

TI's datasheet for OP07CDR
IN(PP) Peak-to-peak equivalent input noise current f = 0.1 Hz to 10 Hz 15 pA
This means for 10MOhm x15pA= 150uV Input noise voltage p-p
Price at D-K $0.56 (100) in DIP-8

CA3130 is better for noise, input current offset, Iio (0.1pA vs 800pA typ@25'C) and bandwidth (4MHz vs 0.4MHz)

OP07C is better for price.

 

[mamech]

Re: Deviation from expected behaviour of sallen key op amp

The LM358 is very old. It is one of the first low power opamps.Old design and low power produce a noisy opamp.
high value resistors produce thermal noise. Why are your resistor values so high and why is your frequency so low?

actually the beginning of the story in this thread illustrates why I reached these values of resistors. I had to have such low frequency so I can capture the signal using arduino with scilab (max 200 sample/ second) , as a kind of low cost portable oscilloscope for students during the automatic control course, and due to the limitation of arduino with scilab, I had to have such low frequency, which obligated me to choose large resistances values. also it was intentional that the circuit oscillates so the student see a real example for a (slow) second order system using opamps.

 

[Audioguru]

Re: Deviation from expected behaviour of sallen key op amp

Don Lancaster's Cmos Cookbook says that the old CA3130 opamp "is somewhat noisy".

 

[D.A.(Tony)Stewart]

Re: Deviation from expected behaviour of sallen key op amp

that's a pretty old book.

Intersil wouldn't be still making it without improvements in the process and obviously still in demand.

 

[Audioguru]

Re: Deviation from expected behaviour of sallen key op amp

Lots of old opamps with horrible spec's (741?) are still manufactured for school kids to play with and for replacements.
RCA invented the CA3130 and Harris took over their semiconductor division then Intersil was formed from Harris workers.

 

[mamech]

Re: Deviation from expected behaviour of sallen key op amp

I have tried CA3130. it gave me the wanted result and no offset exist now, thanks God.

thanks for everyone who helped.

- - - Updated - - -

Lots of old opamps with horrible spec's (741?) are still manufactured for school kids to play with and for replacements.
RCA invented the CA3130 and Harris took over their semiconductor division then Intersil was formed from Harris workers.

what kind of reliable opamp families that you recommend generally? actually, I got used to use these lousy families of lm324, lm358, etc... so I hope that you tell me what do you use personally for low cost signal conditioning circuits.

 

[Audioguru]

Re: Deviation from expected behaviour of sallen key op amp

I have tried CA3130. it gave me the wanted result and no offset exist now.
What kind of reliable opamp families that you recommend generally? actually, I got used to use these lousy families of lm324, lm358, etc... so I hope that you tell me what do you use personally for low cost signal conditioning circuits.

I deal mainly with audio circuits so I use low noise, wideband and low distortion audio opamps like TL07x, LM833 and OPAx134.
I have replace LM358 ICs with MC33172 ICs because they do not have crossover distortion, have the same inputs that work as low as 0V with a single supply voltage, an output that can also go down to almost 0V and have the same very low supply current.

Yesterday I went through my assortment of circuit boards I built but are not used anymore and I found one with a CA3130 opamp. I cannot remember what it was for or if it worked well. I am glad to hear that yours works well.

 

[D.A.(Tony)Stewart]

Re: Deviation from expected behaviour of sallen key op amp

As expected CA3130 is still a great part.

CA3130 is better for noise, input current offset, Iio (0.1pA typ@25'C) and bandwidth (4MHz )