4
\$\begingroup\$

I'm working with an audio ADC that supports a max audio input of 500mVRMS before it will start clipping. This is sufficient for lots of equipment that does a 1VPP/.35mVRMS line level, but I'm finding that lots of equipment like computers, and MP3 players and things will use up to a 3.5VPP/1.25VRMS output voltage. I need the input to work with both the high and low voltage signals. I am considering simply cutting the input signal to roughly 1/3 so that any of these signals will fall under the 500mVRMS level, but then my SNR will be worse on signals actually coming in at 1VPP/.35VRMS because I'll have to amplify it much more to get a good volume on the output. I've tried it and it works, but it may not be ideal.

Is this my best option, or is there a more effective way of reducing the higher voltage inputs to prevent clipping before it gets to the ADC?

Note: Changing the ADC would really not be ideal, so solutions other than that would be particularly helpful.

\$\endgroup\$
5
\$\begingroup\$

Old tape recorders (does anyone remember them?) when put into "record" would amplify the sound of almost inaudible stuff until it got a decent signal, then it went into "normal signal amplify mode". It used what's known as AGC - automatic gain control and maybe this would work for you.... but beware - if you are expecting hi-fidelity don't raise your hopes too high. An AGC is a compressor (much used in modern music like trance and dance genres) and it will alter the dynamics of normal music and worst still it'll shift the dynamics of a lot of classical stuff to "pop" levels.

Basically, it doesn't matter whether you are talking digital or analogue - two different music levels (500mVRMS versus 1250mVRMS) will cause the same issue - you want to turn the output level up on the 500mV signals - it's an 8dB loudness change and that, to the ear, is nearly 2:1 on loudness (1 bel or 10dB is a doubling or halving in loudness at 1kHz)

Either live with it or introduce a circuit that can compensate for the 8dB loss but be aware that you will lose signal dynamics on the louder music.

|improve this answer|||||
\$\endgroup\$
  • 1
    \$\begingroup\$ Same mechanism is used in mobile phones. People don't have to shout into them to make themselves heard on the other end, but few people seem to understand that. \$\endgroup\$ – jippie Apr 6 '13 at 8:42
  • \$\begingroup\$ @jippie I still shout in them because it's usually people calling me at annoying times - AGC is also used in AM receivers to accomodate the big changes in amplitude levels encountered in this type of modulation. \$\endgroup\$ – Andy aka Apr 6 '13 at 8:46
5
\$\begingroup\$

One way or the other, you want a resistor divider. That is two resistors arranged like this:

The output level will be R2/(R1+R2) of the input.

Let's say you wanted to divide the input signal by 3. I'd try to keep the input impedance of the divider at least 1 kΩ, which is R1+R2. R1 of 2 kΩ and R2 of 1 kΩ would divide the input by 3 while maintaining a reasonable input impedance.

If you wanted a choice of a few fixed attenuations, you could effectively split R1 into multiple resistors. Feed the input into the tap that gives the desired attenuation. The resistors above the tap would be floating, so do nothing. Only the resistance between the tap and OUT matters for the purpose of finding R1 in the voltage divider formula above. One "tap" could be OUT, which means the signal is fed straight thru. That would be useful when you have a normal line level audio signal as your circuit is intended to work with.

Another option, is to use a potentiometer in place of R1 and R2. A potentiometer (pot) is a single resistor, which in this case is the combination of R1 and R2. But, it has a third terminal, which is a wiper that can be positioned anywhere along this resistor. By adjuting the wiper position, you adjust the divider ratio. This is exactly how volume controls work, which is what you are really asking about.

|improve this answer|||||
\$\endgroup\$
  • \$\begingroup\$ I think 3k is too low an input impedance for a line level signal, but raising the impedance of the divider worsens the SNR. It would probably be best to use a buffered attenuator if a fixed input pad is desired. \$\endgroup\$ – Bitrex Apr 5 '13 at 23:59
  • \$\begingroup\$ @bitrex: 3 kOhms should be no problem for most line level audio outputs. Many are designed to drive a 600 Ohm load. I think you'll have a hard time finding normal audio equipment with a line level output that can't handle a 3 kOhm load. \$\endgroup\$ – Olin Lathrop Apr 6 '13 at 11:10

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.