The 5532 is not suitable for the servo chip due to high (massive) input bias currents. I suspect that could be the issue here as the configuration should work.
Use a TL072 (or similar) which is a FET device and 100% the best part for job here.
Are you aware that R5 and R6 alter the AC gain of the circuit as they appear in parallel with R1 and R2 ?
Ideally R5 and R6 should be much higher so that they can exert a minor trim to correct offset.
Also the 5532 is about the worst chip to attempt direct coupling the wiper of the volume control to, again because of the offset currents flowing. Not only will this make the control noisy in operation, it will also give a variable correction quantity for the servo to work with as different settings of the control will change the offset. To eliminate this you should AC couple the input (with a suitable bias resistor of course).
The output of the servo should ideally be centred around zero volts in normal operation but that will never happen with 5532's used for the other two devices. You can tell how close to the limit the servo is by measuring the DC voltage at its output. On 12 volt rails you do not want to see more than around - or + 8 volts here. Anything greater means the servo is near the limit of what is correctable. You can use this feature to set a value for R5 and R6. Making them larger will raise the servo output voltage but that is fine as long as enough headroom is left for correction.
Use a TL072 (or similar) which is a FET device and 100% the best part for job here.
Are you aware that R5 and R6 alter the AC gain of the circuit as they appear in parallel with R1 and R2 ?
Ideally R5 and R6 should be much higher so that they can exert a minor trim to correct offset.
Also the 5532 is about the worst chip to attempt direct coupling the wiper of the volume control to, again because of the offset currents flowing. Not only will this make the control noisy in operation, it will also give a variable correction quantity for the servo to work with as different settings of the control will change the offset. To eliminate this you should AC couple the input (with a suitable bias resistor of course).
The output of the servo should ideally be centred around zero volts in normal operation but that will never happen with 5532's used for the other two devices. You can tell how close to the limit the servo is by measuring the DC voltage at its output. On 12 volt rails you do not want to see more than around - or + 8 volts here. Anything greater means the servo is near the limit of what is correctable. You can use this feature to set a value for R5 and R6. Making them larger will raise the servo output voltage but that is fine as long as enough headroom is left for correction.
thank you mooly. another issue was that 5532 in the input duty was already oscillating the circuit with 8v offset where it should be >0.01v to begin with.
i got the problem sorted with using tl072 instead. but i do like the sound of 5532 more.
after ac coupling the circuit, omitting the dc servo, what can i do to eliminate the oscillation with 5532 in the input duty?
i got the problem sorted with using tl072 instead. but i do like the sound of 5532 more.
after ac coupling the circuit, omitting the dc servo, what can i do to eliminate the oscillation with 5532 in the input duty?
The servo chip should not alter the sound at all because it isn't directly in the audio path. All it does is provide a steady state DC correction voltage at its output.
If you have AC coupled pin 3 of the NE5532 from the wiper of the pot then you must also tie pin 3 to ground via a suitable bias resistor of say 220k.
If the stage is still oscillating then you need to try adding a cap across R3 and R4. If they are 1k then try a 100 or 150pF cap.
If you have AC coupled pin 3 of the NE5532 from the wiper of the pot then you must also tie pin 3 to ground via a suitable bias resistor of say 220k.
If the stage is still oscillating then you need to try adding a cap across R3 and R4. If they are 1k then try a 100 or 150pF cap.
You need a IC with low input bias current and low offset voltage.
The slew rate must not be high.
Both fet and bipolar input IC can be used.
I have used OP27EP for offset correction.
47 kOhm and 3,3 or 4,7 uF for the integrator.
The sound does alter a little bit.
The NE5532 is not the right IC for the offset correction.
The slew rate must not be high.
Both fet and bipolar input IC can be used.
I have used OP27EP for offset correction.
47 kOhm and 3,3 or 4,7 uF for the integrator.
The sound does alter a little bit.
The NE5532 is not the right IC for the offset correction.
In what way did the sound alter using the OP27 compared to what before?
Cheers George
U1A is the audio amplifier.
U1B is the DC servo.
U1A should be chosen for good audio performance. U1B should be chosen for good servo performance.
The chosen layout uses a dual for both the audio and servo.
You may find that an opa2134 works well in both duties.
The layout would have been better if one dual did the two audio channels and another dual did the two DC servo channels.
Back to Mooly's R5 & R6. I'd try 10k and measure the opamp output voltage.
If it's less than 3Vdc, then change to 22k and measure again. If it's still less than 3Vdc, then change to 47k.
With the three 1k0 in the feedback routes the gain is 2times.
The output is two times the input signal PLUS two times the remaining signal from the DCservo.
Changing to R5 to 10k reduces the DCservo to 10% and using 47k reduces it to 2%.
All the distortion and filtered audio from the DC servo is reduced before being injected into the -IN node by using a high value R5.
U1B is the DC servo.
U1A should be chosen for good audio performance. U1B should be chosen for good servo performance.
The chosen layout uses a dual for both the audio and servo.
You may find that an opa2134 works well in both duties.
The layout would have been better if one dual did the two audio channels and another dual did the two DC servo channels.
Back to Mooly's R5 & R6. I'd try 10k and measure the opamp output voltage.
If it's less than 3Vdc, then change to 22k and measure again. If it's still less than 3Vdc, then change to 47k.
With the three 1k0 in the feedback routes the gain is 2times.
The output is two times the input signal PLUS two times the remaining signal from the DCservo.
Changing to R5 to 10k reduces the DCservo to 10% and using 47k reduces it to 2%.
All the distortion and filtered audio from the DC servo is reduced before being injected into the -IN node by using a high value R5.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.