3
\$\begingroup\$

I am using the line input on my PC's soundcard (Realtek High Definition Audio) in order to use a software oscilloscope program (Visual Analyzer - www.sillanumsoft.org). I want to be sure not to damage the soundcard with too large an input voltage. I can't find any specs for this soundcard, but since it's designed to be used for audio signals, I'm thinking that I should certainly be safe to at least 1 - 2 volts p-p. Can someone with more knowledge of analog to digital conversion to a soundcard input please help me with this? Since I may want to examine a slightly higher voltage (maybe a non-audio signal), this limit is necessary for me to know. Thanks in advance.


I'm a stack newbie and should have researched a bit more before asking this question- I found understanding at https://en.wikipedia.org/wiki/Line_level. I'm using the line in, never the mic input. I'm an electronics technician, but I have no idea how to check my card's input impedance properly as Marcus suggested. I'm guessing that I'd need to choose some nominal frequency to approach measuring impedance (maybe 400hz) - but I'm used to analog mics, amps and speakers - no idea how to approach analyzing the impedance of an analog to digital input.

\$\endgroup\$
3
\$\begingroup\$

I'm thinking that I should certainly be safe to at least 1 - 2 volts p-p

For a microphone input: certainly far too high (think more in the range of 10-20 mV pp); you risk damage.

For a line-in input, this might be OK, if still much too high.

In any case: you'll need some knowledge of what the input impedance of your sound card is, and match your voltage source's impedance to that, otherwise you'll get bad measurements.

\$\endgroup\$
2
  • \$\begingroup\$ Microphone inputs on PCs are not passive inputs, they actually have a resistor of a few kOhm pulled up to 5V to power an electret mic (or to be pedantic, the jFet in the mic capsule). It's fairly hard to blow them up, but sticking voltages in is generally going to give questionable results. They usually use a Realtek chip, datasheets are online. They often also offer a "preamp" setting which can be used to add lots of noise if so desired. \$\endgroup\$ – Ian Bland Nov 25 '17 at 0:03
  • 1
    \$\begingroup\$ The line inputs generally can handle well over 2V p-p, though they aren't set to any specific audio standard of 0db so far as I've been able to ascertain. Consumer 0db for instance is about 2.9V p-p. The input impedance is generally precisely some number of kOhms. Which doesn't matter too much so long as you're not shoving a high output Z into it. My experience is that Realtek's "high definition" means you've got quite a few least significant bits to waste on the system noise before the audio signal emerges above them. \$\endgroup\$ – Ian Bland Nov 25 '17 at 0:07
1
\$\begingroup\$

There are several things to keep in mind:

  • Clipping level: this voltage will clip the soundcard input, but do no damage. This can depend on the software gain setting. For best signal to noise ratio, you have to do a balancing act, stay below clipping but not too much, like a few dB below clipping. The level which gives best THD will usually be a bit lower.

  • Damage level

This depends on the input circuitry and the associated protection. For example if the input opamps are powered by +/-15V then voltage in this range should be safe. Whereas if they are powered by +3V3 the safe voltage range will of course be a lot lower.

Since I have no idea what input circuit you have, I'll elaborate a bit.

What blows the opamp is excessive current flowing into the inputs. This happens when the input voltage exceeds the supply range. However the current is limited by a resistor which will be placed in series with the input, and also, usually, a coupling capacitor (which also has a voltage rating).

You will also usually get some ESD protection diodes which will shunt the excess current to ground or the power supply.

Increasing the resistor value will reduce the fault current in case the input voltage is too high. For example with a 100k resistor, you can zap the input with 50V AC, only 0.5mA will flow, and nothing will burn. If it is 50VDC, and the input capacitor is rated for 6.3V, then it may not be a good idea though...

Problem is increasing the resistor value increases noise, especially current noise if the opamp is a BJT input type.

Anyway, rule of thumb:

If we allow 1mA current to flow into the inputs, then we should use a protection resistor proportional to the maximum expected voltage. For example if you want to sample a signal from a device powered by +/-15V, these will be your max values, so a 15k or 22k resistor would be a good choice.

However, the high resistance value may result in high-frequency rolloff when it interacts with the input capacitance of the soundcard. You can measure this with a loopback test: output - resistor - input and make a frequency response sweep in your software, check if the resistor affects frequency response.

If the signal amplitude is too high, then you can use a simple voltage divider. Remember the output impedance of a voltage divider is the value of both resistors in parallel, so it is much lower than the upper resistor, thus it has better HF frequency response...

\$\endgroup\$

Your Answer

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