7
\$\begingroup\$

I'm using the following input stage for an audio application:

audio amp input

The audio signal is analog and single ended. The connector side is exposed, so a person could touch it. The TVS should protect the circuit from ESD. The C10 is a DC coupling capacitor and the other components are used to filter out GSM noise. The amplifier is a 2W class D audio amplifier.

The problem is that in certain situations, when a person touches the connector, a DC voltage is induced on the connector side of the C10 capacitor, which causes a whistling sound, although the only thing I can measure on the audio line is a DC voltage (no sinus-like wave).

The blue line in the following image shows the induced voltage on the connector side of C10. The sound fades out after about 60 seconds.

voltage at cap

My question is: is there a way to prevent this or to discharge the capacitor quickly? Alternatively, why is the whistling sound there and can I prevent it?


EDIT: For the sake of completion: I use the following parts now and touching the connector pins doesn't generate noise anymore.

new schematics

\$\endgroup\$
  • 4
    \$\begingroup\$ Try adding a resistor to ground on the connector side (or both sides) of the coupling (not decoupling) capacitor C10. The value(s) should be large enough to not significantly alter the input impedance but small enough to allow C10 to discharge relatively quickly. \$\endgroup\$ – Alfred Centauri Dec 19 '13 at 13:03
  • \$\begingroup\$ That's the idea I had too, but I thought maybe there is another trick.. Thanks for the heads-up, I corrected the word (coupling). \$\endgroup\$ – user1884015 Dec 19 '13 at 13:59
  • 1
    \$\begingroup\$ For the curious and lazy, here is the datasheet for IP4256CZ3. \$\endgroup\$ – abdullah kahraman Jan 30 '14 at 11:40
6
\$\begingroup\$

This is most likely because you haven't provided any impedance for C1 to work against. If the signal at right is going into a high impedance input of a chip, which it sounds like could be the case, then any DC level injected at left will transfer to the right and stay there for quite a while.

For example, if whatever is to the right has 10 MΩ input impedance, then the time constant with C1 is 10 seconds. Audio is generally considered to go down to 20 Hz at the lowest, so you can and should filter frequencies below that. Let's say you want the high pass filter rolloff frequency to be 10 Hz to have little effect at 20 Hz. That means you want a 16 kΩ resistor to ground immediately to the right of C1. As long as the input impedance of the amplifier to the right is significantly higher than 16 kΩ, this will work fine and provide a nice and predictable DC rejection filter.

\$\endgroup\$
  • \$\begingroup\$ What is a significant higher impedance? The amplifier is the TI TPA2000D1, and has an input impedance of 24 kΩ, 44 kΩ, 74 kΩ and 104 kΩ respectively, depending on the selected gain. \$\endgroup\$ – user1884015 Dec 19 '13 at 17:02
  • \$\begingroup\$ If the input impedance of the amp is really that low, then something else is going on. Even the 104 kOhm, the system should have recovered from a DC step in well under 60 seconds. Make sure you are observing all the input signal coupling requirements of the amp. Some amps have their own internal DC bias and you must couple the signal thru a cap, for example. \$\endgroup\$ – Olin Lathrop Dec 19 '13 at 18:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.