OPA1664: Output voltage is significantly reduced from pre-amplifier when capacitor is used in the microphone output.

Hi Thomas,

Thanks for your valuable comment. I made circuit using MFB topology as suggested, I simulated the  circuit with the capacitor. (schematic & output shown below). and it is working.

I have a fundamental question to ask, even in the circuit without capacitor in sallen key topology as show in fig.2(a) above, how non inverting input of the op-amp is receiving dc input, as branch 4 is connected to the signal source. How just by putting capacitor it will not work and dc will not make through non inverting terminal (capacitor does not allow dc to pass but where is dc to begin in the first place), but just by removing capacitor non inverting input will get dc path, where is this dc coming from as just before capacitor only AC signal source is present.

(Is there any reading or document that I can go through as to understand the concept)

Regards,

Tapas 

Fig.4(a) [MFB Circuit] 

Fig. 4(b) [oscilloscope output]

Hi Thomas, 

Thanks for your previous replay. I have made PCB incorporating OPA1664 audio op-amp. In the output of every stage I am getting dc voltage along with amplified signal. Op-amp output is getting saturated in positive part very early before negative part because of that. Below I have posted pics with description. I also want to know about distortion of the output waveform (THD), is there any rule of thumb to calculate it (I have a feeling that fourth stage has more distortion than other stages, by looking at oscilloscope), or any equipment is there to measure it. For comparison I have put square wave and ramp signal, below.

        Fig. 1 Sallen key schematic, except that dual supply is being used, which is not shown.

                                                             Fig.2 OPA 1664 used in sallen key configuration, dual supply (+/- 1.5V)

                                                                    Fig.3 Output from function generator ( +/-30mV, 100 Hz freq)

                                                          Fig.4 Input to op-amp 1664

                                            Fig.5 Output from op-amp first stage , dc shifted by 56 mV 

                                      Fig.6 Output from op-amp second stage, dc shifted by 140 mV 

                                 Fig.7 Output from op-amp third stage , dc shifted by 344 mV 

                     Fig.8 Output from op-amp fourth stage, dc shifted by 700 mV 

 Fig.9 Input [cyan] &  Below output [yellow] side by side, Output is AC coupled (& at different  scale) so that both input and output can be seen at same time. 

 Fig. 10 Input [cyan] and output [yellow] (AC coupled); 

  Fig. 11 Input [cyan] and output [yellow] (AC coupled); 

Regards,

Tapas

Hi Thomas,

Thanks for your reply.

V+ is 1.5V

V- is -1.5V

Vcm is connected to ground. (Pic below) 

I am unable to understand why there is dc shift in every stage, if we use AC coupling capacitor, then lower frequencies will get attenuated. I am just wondering does input bias current will cause voltage offset. 

I need your help in reducing offset voltage, so that there should not be early clipping in the positive input cycle.   

                   Fig. 12. Dual supply configuration for OPA 1664

                Fig. 13. Schematic values are seen more clearly, op-amp used is OPA 1664 

Regards,

Tapas

Hello Tapas,

The issue you are seeing with the high output referred voltage offset at each stage is due to the OPA1664 input bias current flowing through the filter's various resistor paths. The input bias current for the OPA1664 is typically about 600 nA, with a 1.2 uA maximum (TA = 25 C). That is substantial enough to cause a 6 mV drop across a 10 kOhm resistor, and the filter sections have several resistors that are multiples of tens of kilohms. Then there is the gain of each stage that amplifies the offset associated with these drops.

Below, in the upper TINA circuit you can see how the bias current flows through the resistors and the resulting output referred voltage offset for the first stage of the filter. The bias currents in conjunction with the natural input voltage offset of the OPA1664 produce an output offset of about 58.4 mV, which is in line with what you are seeing.

One way to lessen the effect is to reduce the resistances. In the lower portion of the TINA schematic the resistors have been reduced by a factor of 10x. The capacitors have been increased by the same factor so that the frequency response of the filter section remains the same. You can see this in the amplitude vs. frequency plots on the right hand side of the circuits. Observe that the output offset of the stage is reduced from 58.4 mV, to 5.4 mV. This scaling could be applied throughout the filter.

I searched for another high-performance audio op amp that has lower input bias current than the OPA1664. There are other audio op amps with lower bias current, but they do not operate with supplies as low as +/-1.5 V - like the OPA1664.

Regards, Thomas

Precision Amplifiers Applications Engineering

Hi Thomas,

Thanks for your comprehensive reply. By putting AC coupling capacitor after fourth stage in the sallen key configuration seems one way to reduce dc offset voltage, rather than putting AC coupling capacitor after every stage, which will impact low frequency performance. What is your take on it?.

Meanwhile I have ordered OPA1604 and OPA1654 having input bias current of 20nA and 10pA respectively, both from soundplus series. Also I will be ordering resistor (10x reduced value) and capacitor (10x increased value) to reduce voltage offset as shown by you in the previous post.

Regards,
Tapas

Hello Tapas,

Adding a coupling capacitor in the signal path before the 4th stage will remove the offset from the previous 3 stages. Then, only the offset of the 4th stage will be present at the output of the signal chain. The coupling capacitor at the input of the 4th stage will modify the filter response adding a 2nd-order high-pass to it. That probably won't be of much consequence if the filter is only used for audio frequencies and the high-pass cutoff frequency is made low; something below the lowest anticipated audio frequency. If you wish to avoid adding the series capacitor, then the RC scaling I suggested should result in much lower total offset.

If you can increase the battery voltages to a minimum of +/-2.25 V, then the OPA1604 and OPA1654 with their lower input bias currents becomes an option. The offset contribution from their input bias current will drop significantly compared to the OPA1664. Make sure their common-mode input voltage (Vcm) is sufficient for the applied input signal range; especially, as the signal progresses to the later stages. The Vcm range for the OPA1604 is (V-) + 2 V to (V+) - 2V, and for the OPA1654 (V-) + 0.5 V to (V+) - 2 V.

Regards, Thomas

Precision Amplifiers Applications Engineering