1
\$\begingroup\$

First asked this over at sound stackexchange but they pointed me over here as a better place for it.

A general question but inspired by a specific example so framed as such.

I have powered studio monitors attached to a computer's unbalanced headphone port, and there is some amount of underlying noise, either from the computer's DAC or induced along the unbalanced cable or some combination. Overall not particularly surprising. With headphones plugged in instead, I don't hear the same noise at all.

My thought is that the likely cause of this difference is that the headphones, being a relatively low impedance load (64 ohm in this case) act to short out the noise so it's not heard. Basically the noise, wherever it's coming from, isn't strong enough to drive the headphones' drivers enough to be audible.

When the powered speakers are plugged in instead, they have a very high input impedance that then goes into their amplifiers (10k ohm). Now the same noise can drive the input easily, and is then amplified and played through the speaker.

So, the question is, if I were to put a resistor in parallel with the speaker input, or even plug the headphones in in parallel for that matter, would this decrease or eliminate the noise? Am I thinking about some part of this wrong?

thanks!

\$\endgroup\$
  • \$\begingroup\$ Another significant difference is that the speakers are plugged into AC power, while the headphones are not, yes? Your idea is not crazy. Any RF in the air will couple voltages into your wires (i.e. "induce" like you say) - If the load impedance is lower, that noise can't create as much voltage. However, that resistor will also reduce the desired signal level and may mess with the frequency response a little. Maybe try a 10k Potentiometer, see if you can dial-in something that works. I'd try a battery-operated amplifier first, see if the noise source is power-line or ground-loop. \$\endgroup\$ – Kyle B Apr 23 at 4:30
  • 3
    \$\begingroup\$ Easy way to test: get a headphone splitter and plug in the headphones at the same time as the speakers. Does the noise go away? \$\endgroup\$ – user1850479 Apr 23 at 5:31
  • \$\begingroup\$ The splitter idea is a practical solution if it works. If not, an inline attenuator in the signal cables is a rock solid fix. \$\endgroup\$ – Graham Stevenson Oct 8 at 19:32
  • \$\begingroup\$ When you turn the volume pot on the powered speakers, does the noise loudness change? Is it a hiss, woosh or other kind of white/pinkish noise, or hum, buzz? \$\endgroup\$ – bobflux Oct 8 at 22:24
2
\$\begingroup\$

Three possibilities here:

  1. Random electron-movement noise: a 10,000 ohm resistor in 10,000 Hertz bandwidth, ignoring any A_weighted responses, will provide 12 nanoVoltRMS * sqrt(10,000 Hz) == 1.2 microVolts RMS noise; whereas a 62 ohm resistor will provide 100 nanoVolts RMS.
  2. your wiring may pick up RFI of AM, FM TV, cellphone energy and the powered-speakers have opamp inputs that non-linearly respond to that RFI/EMI and rectify it and downconvert it to within audio range and you hear that; cure is better cables, or RFI filters added at inputs to the powered-speaker.
  3. As mentioned, powered systems almost always have switch-mode power supplies for efficiency and low heat generation, but the switch-mode transformers still have primary and secondary windings with some capacitance between the two windings. The energy coupled between those windings will, per the assistance of mother nature, explore ALL possible paths back to the high dV/dT switching transistors in the supply; one of those paths will be your cable into the speaker.
| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ 1: Thermal noise is absolutely not the issue here. A: it's far too small in magnitude to be at all significant. Even the noise of a 10k resistor (20kHz bandwidth) is 115dB below 1V signal level (1.8uV). \$\endgroup\$ – Graham Stevenson Oct 8 at 18:47
0
\$\begingroup\$

I am inclined to go with your analysis that the high input impedance and the gain of the powered studio monitors is the root cause of your problem.

The solution would be to keep the volume controls of the monitors constant at around 25% and increase the system setting in the PC.

| improve this answer | |
\$\endgroup\$
0
\$\begingroup\$

You don't mention what volume settings you are using but a common problem is that of driving a high level output at low volume causes a large deterioration in signal to noise ratio.

schematic

simulate this circuit – Schematic created using CircuitLab

Figure 1. (a) Existing setup. (b) Attenuated setup.

Let's say that you are connected as shown in Figure 1a. The PC's output can be a couple of volts to drive the headphones well but the monitor amplifier is more sensitive - say 100 mV. As a result you turn down the volume on the PC via software control. Note that this reduces the volume into the PC amplifier reducing the signal level but the amplifier's noise level remains the same.

The solution in this case is to add a potentiometer or "pad" consisting of a potential divider. Now the PC audio can be cranked up to a decent level to drive its internal amplifier while R4 attenuates both the signal and the noise driving the monitor amplifier.

Try it with a 10:1 reduction as a starting point if using fixed resistors - 10k and 1k would do - or an audio taper pot.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Studio monitors (the type used by the OP) typically have an input sensitivity of ~ 1V rms (usually 0.775V 0dBu or 1.23V +4dBu for full output) so it's unlikely to be an oversensitive input. Adding a fixed attenuator is still a good idea though. \$\endgroup\$ – Graham Stevenson Oct 8 at 19:09
0
\$\begingroup\$

Your computer's soundcard is unlikely to have an output impedance of greater than 100 ohms. More likely 30 ohms or less. Loading it with 64 ohms will only introduce a few dB of attenuation.

Unless you have some really rubbish cables, I doubt you have any problem there either.

It's likely that the high acoustic sensitivity of the amplifier is responsible. About one volt of input signal will produce full output. That will easily create an SPL of around 110dB (or more). 120dB is not uncommon for some powered studio speakers.

The noise level of an average soundcard may be somewhere around 85dB below full output. Worse if mediocre. They don't exactly operate in a 'friendly' EMC envirobnent. So, the acoustic level at your speakers will easily be around 30dB which is entirely audible. It's roughly equivalent to quiet street noise.

The inline attenuator is a good idea whether using 2 fixed resistors on each signal line or a potentiometer.

| improve this answer | |
\$\endgroup\$

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.