I am trying to connect an amplifier with 100W RMS per channel output to a subwoofer that only has a single channel line level input. The amplifier does not have a volume controlled line/preamp output. I was able to find different bits and pieces, but not a single comprehensive article. So after piecing together different information to the best of my ability and very limited experience I put together this schematic:


simulate this circuit – Schematic created using CircuitLab

In my research I found following resources:

  1. StackExchange answer describing voltage dividers for different combinations of balanced/unbalanced input and output
  2. StackExchange answer to the same question as above with more detail about source voltage calculation, resistor wattage requirements based on total resistance and how the capacitor affects low frequencies
  3. Wikipedia article on line level transmission
  4. Article on how to calculate resistor values in L pad attenuator circuit
  5. StackExchange question about a stereo to monoconversion

From the wiki article I have learned that the line level signal is 1.736 Vpk. My amplifier specs state the that max power output is 140W per channel at 8 ohms. From that I calculated using Power = V^2/8 that the max voltage is 33.47 V.

So to get the voltage from 33.47 to 1.736, I need to attenuate the signal by ~25.7dB (dbv=20∗log(Vo/Vin))

Based on the article on L-Pads calculation and more specifically using I was able to calculate R1 and R2 values to be 4k7 and 270 ohms.

K factor for voltage = antilog(dB/20) = 10^(25.7/20) = 19.28

Using "Looking from the Load" formulas (I played with the Zl until I got the right resistor wattage and close to values of off the shelve resistors): R1 = Zl (K-1) = 260 (19.28 - 1) = 4752.25 R2 = Zl (K / (K-1)) = 260 (19.28 / (19.28 - 1)) = 274.22

My understanding of the necessity for the capacitor is to remove any DC bias. Based on the "additional details" answer, the Z(cap) has to equal to the total resistance at given frequency to calculate the "cut off"?. The way I understood that answer that if I don't want to attenuate the lower frequencies say above 30Hz, I need to calculate the capacitor value at 30Hz so that the Z(cap) is equal (4k7 + 270). This is my best guess based on the answer provided. Which ended up being 4970 = 1 / (2 * Pi * 30 * C) = ~1uF. I don't know if I got the total resistance correct (since I added the 1K resistors for the stereo to mono convertor part).

Further, I only have 1/4W resistors on hand. So the R1/R2 values were influenced by that. I was trying to get the ~5k ohm resistance to get down the Watts. Again this is my best interpretation of the answer provided and might be wrong on my part.

So here are my questions:

  1. I have read elsewhere and it was also implied the second answer that the source impedance of the Line level signal should be as small as possible - in the 100 to 600 ohm range. Is that the Zl that I got to 260? and if so is that correct and still correct after adding the 1K/1K "averaging" voltage divider?
  2. I understand that getting the polarity wrong on the input could cause problems. Does this circuit still has that problem given the 1K resistors and only using one ground? I know this is just me using it and I can make 100% I get the polarity right, but in theory, what happens if I let say get the right channel connected backwards?
  3. Is the C1/C2 value calculated correctly and/or is that capacitor necessary?
  4. Is there any potential problem with sub being plugged into mains "the wrong way" and somehow sending voltage over the ground to the amplifier? I understand that the typical car audio LOCs have 1:1 transformers to isolate the source from the target, but these transformers are either expensive or attenuate low frequencies, which I am trying to actually use.

As I said, I have limited experience and knowledge of electronic circuits and this is my best shot given the information I found.


2 Answers 2


I think what is missing from your consideration is that the 100W amplifer outputs may be driven from a H-bridge, which means that you cannot connect the negative terminal to ground, as both positive and negative terminals may be "live".

Which also means the positives don't use same reference and so you can't sum them with just resistors.

Just use audio isolation transformers that are specifically meant for this purpose.

Also there is no standard what "Line Level" means and it's arbitrary. Sure, some equipment, under some conditions, might have 1.736Vpk, but you don't know what conditions those are (what music played, on which volume level, and how much there is headroom for peaks vs some longer-term average). Usually, in controlled environments like recording studios or broadcast, you do have a definition for your signals, like standard nominal reference level and defined headroom for peak voltages which commonly is 8x or 10x larger than the nominal.

Modern equipment these days use 2VRMS for full scale, but anyway you might need attenuation and your woofer likely has a gain pot.

Otherwise, your calculations are pretty spot on.

33V divided to 1.7V is a rato of 1/19 and total about 5k resistance is below 1/4 watts, and the divider impedance is below 270.

The 1k summing resistors after it makes the impedance slightly above 600 ohms but like I said above, there are larger problems than that. The 1uF cap with about 1200 ohms impedance has cutoff only 130Hz so maybe 10uF cap would be more suitable. You usually want 10x lower RC cutoff than you need so that phase shift will not be a problem.

  • \$\begingroup\$ “amplifier outputs may be driven from a H-bridge” is there any way to tell? When I measure negative to negative I get about 0.15 ohms and negative to chases about 0.08 ohms. When it comes to isolation transformers, I have read that ones with low attenuation in low frequencies are hard to find/expensive. Amazon has a lot of El-14 options to buy, but as far as I can tell they are rated for 300Hz and above and even then at 3dB max loss. So I guess they would. E consider “telco grade”? I found the LM-NP-1001-B1L recommended by a Reddit post. Not sure if that is suitable. \$\endgroup\$
    – Peter
    Commented Dec 19, 2023 at 20:03
  • 1
    \$\begingroup\$ You could name the brand and model, maybe there is a service manual with schematics. Or you could verify it by reverse-engineering how the output works. Or if there is ever a voltage difference between negative speaker terminals. Using a multimeter to measure resistance does not necessarily mean you get the correct result, especially if what you are measuring is not a resistance, but for example semiconductor. \$\endgroup\$
    – Justme
    Commented Dec 19, 2023 at 20:32
  • \$\begingroup\$ Yamaha R-N303 is the amplifier. \$\endgroup\$
    – Peter
    Commented Dec 19, 2023 at 21:04
  • 1
    \$\begingroup\$ @Peter Service manual schematics show the negative terminals are grounds. However the amp has a line out. Or maybe change the amp or sub to be compatible with each other. \$\endgroup\$
    – Justme
    Commented Dec 19, 2023 at 21:25
  • 1
    \$\begingroup\$ @Peter It's a ballpark simplification of having the 4k7 in parallel with 270 ohms which is actually 255 ohms, and then there is a 1k in series, so the source impedace for the capacitor is 1255 ohms roughly. That's within 5% without actually calculating it. \$\endgroup\$
    – Justme
    Commented Dec 19, 2023 at 22:31

Partial answers.

C1 and C2 are necessary because you never know exactly what is going on in both the main amp output stage and the sub-woofer input stage.

But they are the wrong value. This gets into things like Thevenin-equivalent impedance, but the short form is that R2, C1, and R5 form a high-pass filter with a corner frequency of approx. 130 Hz. That is way too high for driving a sub-woofer.

Step 1, increase the cap to 100 uF. Now the corner freq is down around 1 Hz, letting through all of the rumbles, thumps, and booms.

Step 2, for that value the cap now is an electrolytic type, which usually is polarized. But the audio signal is bipolar about ground, so the cap will be unhappy. The solution is to either get a 100 uF, non-polarized capacitor, or use two, 220 uF polarized caps connected back-to-back in series. I doesn't matter if they are connected + to + or - to -.

Now that that part is fixed, it probably will change. You are correct that the Line output signal usually has a relatively low output impedance. This is easy to achieve by simply scaling all of the resistor values downward. The problem is that the capacitor value will have to increase proportionally. For example, if you scale everything down by 10:1, the two back-to-back electrolytic caps now are 2200 uF each. Not a big problem, but certainly less convenient. Others probably will expand on this area.

  • \$\begingroup\$ So the R1 and R2 don’t play a role in the high pass filler, only the R5 does? The second answer I quoted was suggesting that the filler was related to the Rtot. \$\endgroup\$
    – Peter
    Commented Dec 19, 2023 at 21:59
  • 1
    \$\begingroup\$ R1 is so much larger than R2 that it can be ignored in an approximation. That leaves R2, C1, and R5 as stated. All four components factor into a true calculation of the filter corner frequency. Plus, the two channels each act as a load on each other, further complicating the math. \$\endgroup\$
    – AnalogKid
    Commented Dec 20, 2023 at 0:59

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.