I have a 75W amplifier with a 8ohm speaker connected.

I need to record the audio from that output, so i was thinking in using a voltage-divider to condition the signal and record it through the Line-In input of my sound card.

What I have to take into account?, what happen if the signal don't have a GND but both lines are driven by the amplifier?.


3 Answers 3


Tom Carpenter is pretty close to what I would do (and have done many times in the past). The difference is that I always put the attenuation pad in front of the isolation transformer.

Here's why.

It takes a transformer with a large core to handle low audio frequencies at the voltage levels that a 75W power amp puts out. The transformer becomes correspondingly larger as the power levels rise.

It makes far more sense to first attenuate the signal, then run it through an isolation transformer. This allows the isolation transformer to be appropriately sized for the output signal.

There are two distinct attenuation schemes but both provide similar results: single-ended or balanced. Single-ended is more appropriate for power amplifiers that have single-ended outputs (negative speaker connection is circuit ground) whereas balanced (differential) isolation may be better suited for power amplifiers having bridged outputs. In actual cases, either works well.

75 Watts RMS at 8 Ohms is about 24.5 Vrms (full output just before clipping). This is about 30 dBu.

Typical professional gear runs at +4 dBu for full output (call it 1.2 Vrms). Consumer gear runs at -10 dBv - call it -12 dBu.

If you are recording the speaker signal on a piece of consumer gear, you need about (30 dBu - (-12 dBu)) or about 42dB attenuation. Let's call it 40dB for ease of calculation.

40dB attenuation is a ratio of 100 to 1. In other words, reduce the speaker signal by a factor of 100 in order to feed the input of your consumer-grade gear.


simulate this circuit – Schematic created using CircuitLab

Schematic credit goes to Tom Carpenter - I simply copied and modified his schematic.


You can attenuate the signal using potential dividers. That will work fine as long as you account for loading from the input impedance of whatever you are connecting it to.

Given there is no obvious ground for your signal (you say both sides are driven), the simplest way to solve this is to AC couple to both sides of the speaker. This basically involves simply adding a series capacitor with sufficient size so as to form a high-pass filter with your potential divider. You can place one capacitor from one side of the speaker to your own circuits ground, and a capacitor from the other side of the speaker to the potential divider.

In cases where there could be a common ground between either of the speaker pins and your circuit, capacitors alone would not be sufficient. It would be wise then to add an audio isolation transformer (these are 1:1 transformers rated for audio frequencies) to ensure that there is no way for a common ground to short out the potential divider.

There's at quick glance a good resource here about circuits with audio transformers.

In any case something like this circuit should work:


simulate this circuit – Schematic created using CircuitLab

The capacitors and transformer isolate your circuit from the speaker circuit by removing any DC currents. This means you are able to then somewhat arbitrarily pick either node to be your ground. You could also place the potential divider before the transformer if you want.

In terms of picking the component values, you would have to measure the peak to peak voltage of the speaker (or guess that \$V \approx \sqrt{P\times R}=\sqrt{75 \times 8}\$ = 25Vpp). You can set the potential divider such that it divides that voltage down to the ~2.5Vpp rated voltage for a line-in port (so about a 10:1 divider). Pick a resistance that is low enough not to be loaded by your line-in port, but high enough not to overload the speaker. Maybe something on the order of 1kΩ total (910Ω for Rt and 100Ω for Rb perhaps).

You would pick the capacitance such that you form a high pass filter with a cut-off frequency below the audible frequencies - say a 3dB point of about 10Hz. You can use the following formula for an R-C filter:

$$f_c=\frac{1}{2\pi R C} \rightarrow C=\frac{1}{2\pi R f_c}$$

In your case the R value would be the sum of the two resistors in your potential divider, while C would be the total series capacitance of both AC coupling capacitors. That means the value of the capacitors in the circuit would be double the value you calculate from the formula (capacitors in series don't sum!). Based on the 910Ω/100Ω combination I guessed earlier, that would give a capacitance of ~33μF for each capacitor.

  • \$\begingroup\$ it's not wise to connect circuit ground to a speaker output line from an amplifier. Many modern amplifiers now use bridged output stages, where both speaker lines are live with respect to ground. \$\endgroup\$ Commented Dec 12, 2017 at 20:04
  • \$\begingroup\$ Add an isolation transformer after the attenuation pad and all is good. \$\endgroup\$ Commented Dec 12, 2017 at 20:05
  • \$\begingroup\$ Just wondering, Tom, if the speaker+ is, in fact, grounded this isn't going to work to well. Would it be better to add a second Rt in the lower line just in case? That would limit the current while still allowing a return path. The lines could then be swapped to obtain the signal in this condition without risk of high currents anywhere. \$\endgroup\$
    – Transistor
    Commented Dec 12, 2017 at 20:06
  • \$\begingroup\$ @DwayneReid / Transistor, I see your points - took a quick simulation to notice that would be a problem. You are correct, an isolation transformer would very wise. \$\endgroup\$ Commented Dec 12, 2017 at 20:18
  • \$\begingroup\$ But why are the capacitors still there, now that "we" have an isolation transformer? Only a very broken amplifier can have a DC output (otherwise the speaker will burn out), and the transformer won't pass DC anyway. \$\endgroup\$
    – pipe
    Commented Dec 12, 2017 at 20:34

While Tom's answer is the most generic solution, I'd like to offer a simpler way. Circuitlab couldn't really draw this so I present to you this masterpiece:

Floating laptop connected to speaker with voltage divider

Use a laptop and disconnect every external power supply. This includes all USB devices with external power, etc. Run it on battery while recording.

This could be an option if:

  • Your battery lasts for as long as you need to record
  • Your recording hardware can be powered from the laptop or you have a laptop with integrated line-in
  • You only want to record one channel
  • \$\begingroup\$ The sound source it's powered from a vehicle battery, so technically the amplifier's ground and the desktop PC's ground are isolated. It could be possible to leave out the isolator transformer and use just the power-divider as you propose? \$\endgroup\$ Commented Dec 15, 2017 at 17:16
  • \$\begingroup\$ @GasparSantamarina Yes, that should work as well. As long as one of the devices are floating. \$\endgroup\$
    – pipe
    Commented Dec 15, 2017 at 17:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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