Bass cut - will this work?

[justinis]

I am making a passive preamp to connect my DAC and tube pre to my power amp. I live in an apartment building, so I want to include a bass cut switch for playing tunes late at night. Below is the schematic I've come up with (1 channel shown). Anything wrong with this approach? Any recommendations?

I realize I will lose some gain when the bass cut is activated, but I don't think that should be a problem; I will only be using it at low volume levels anyway.

The power amp is a custom design. The first stage is a 12AX7 with a 475K from grid to ground. I'm adding a volume pot as well. Schematic and frequency response plot below.

 

[gadget73]

Basically this is how tube gear in the old days did it, though usually they did not have a resistor bypassing the cap. As long as your source can drive into <50K input you should have no problems.

 

[ConradH]

Looks perfectly reasonable to me. The only question I'd be thinking about is finding the best filter frequency that gives you the best sound with minimal disturbance to the neighbors. If loss doesn't matter, a two pole filter might have less sonic effect with only slight increase in complexity.

 

[BinaryMike]

Something doesn't add up here. 500R is only 1% of 50K, so the plotted frequency response isn't possible for the circuit as shown.

 

[tubeactive]

Depending on the source and which lowest frequencies you want to limit, .01 might actually be too high a value. Simple R-C networks like this one, "single order," cut at a 6 db/octave slope.

So, here is the simple math needed to figure your "single order" network. R x C = Time Constant (TC) in uSeconds. Then, the (audio related) mathematical constant of 159,155 divided by the TC, in uS, yields the actual, designed in, -3 db point re: frequency, in Hz (cps).

.01 x 500,000 = 5000 uS 159,155 / 5000 = 31.8 Hz Your high pass filter(low freq. cut) is -3 db at around 32 Hz. This presumes your source will not "load down" the network.

The higher the resistor, the lower in frequency of cutoff...The lower in capacitance, the higher the frequency of cutoff. I suggest you build your network with a 500K Ohm, linear potentiometer across the .01 cap. That way, you can "dial-in" your low cut needs. When you have satisfied your ears and your neighbor's acceptance, you can then, perhaps decide to measure the R value of the pot and perhaps solder in a fixed resistor. Then, you can reuse the pots elsewhere...

 

[justinis]

Looks perfectly reasonable to me. The only question I'd be thinking about is finding the best filter frequency that gives you the best sound with minimal disturbance to the neighbors. If loss doesn't matter, a two pole filter might have less sonic effect with only slight increase in complexity.

Thanks! How would you recommend adding the second poll? Just another RC after this one in series? Same values?

Something doesn't add up here. 500R is only 1% of 50K, so the plotted frequency response isn't possible for the circuit as shown.

You're right, sorry! Good catch. The 500R should be 500K. I've changed it in the image.

Depending on the source and which lowest frequencies you want to limit, .01 might actually be too high a value. Simple R-C networks like this one, "single order," cut at a 6 db/octave slope.

So, here is the simple math needed to figure your "single order" network. R x C = Time Constant (TC) in uSeconds. Then, the (audio related) mathematical constant of 159,155 divided by the TC, in uS, yields the actual, designed in, -3 db point re: frequency, in Hz (cps).

.01 x 500,000 = 5000 uS 159,155 / 5000 = 31.8 Hz Your high pass filter(low freq. cut) is -3 db at around 32 Hz. This presumes your source will not "load down" the network.

The higher the resistor, the lower in frequency of cutoff...The lower in capacitance, the higher the frequency of cutoff. I suggest you build your network with a 500K Ohm, linear potentiometer across the .01 cap. That way, you can "dial-in" your low cut needs. When you have satisfied your ears and your neighbor's acceptance, you can then, perhaps decide to measure the R value of the pot and perhaps solder in a fixed resistor. Then, you can reuse the pots elsewhere...

Thanks. Great idea. I may just build the whole thing with a pot on a little breadboard so I can test it before building the amp with a fixed resistor.

 

[ConradH]

Yeah, that 500 ohms kills it! I'd also probably change the cap to 0.033 uF to improve SQ and hopefully (doubtfully) still keep the neighbors happy.

edit- It's really hard to get good attenuation with a passive circuit, especially without inductors. Either the losses are high, or the slope is shallow. IMHO, for sound quality you'd like to keep the response reasonably flat down to 100 hz or so, but have it rolled off significantly by 60 and below. You can split the cap (double each value) and put a resistor from the junction to ground, say 50 k, but I'm not sure it's worth the effort.

 

[kward]

If you want to study second order filters, google Bessel, Butterworth, Chebyshev, or Sallen-Key. (Named after their inventors.). It's a bit more complicated and as you say, unless implemented with some gain (opamp or high gain triode usually), will attenuate quite a bit.

For a first attempt, I'd probably stick to a first order R/C filter that is flat down to say 150 Hz give or take, and then let it roll off naturally from there at the first order rate (6 dB per every halving of frequency).

 

[BinaryMike]

It's possible to get slopes steeper than 6dB/Oct with simple RC filters by staggering. Harrison's article presents pre-cooked design data for useful examples: https://www.dropbox.com/s/ro9kcjo25tqae7t/RC Filter Design.pdf?dl=0

 

[ConradH]

It's possible to get slopes steeper than 6dB/Oct with simple RC filters by staggering. Harrison's article presents pre-cooked design data for useful examples: https://www.dropbox.com/s/ro9kcjo25tqae7t/RC Filter Design.pdf?dl=0

That's a really nice article, packing about three chapters of a filter design book into a brief coherent form. IMO, one can do better with LC filters but, at the desired high impedances and low frequencies, the inductors become absurdly impractical.

 

[justinis]

If you want to study second order filters, google Bessel, Butterworth, Chebyshev, or Sallen-Key. (Named after their inventors.). It's a bit more complicated and as you say, unless implemented with some gain (opamp or high gain triode usually), will attenuate quite a bit.

For a first attempt, I'd probably stick to a first order R/C filter that is flat down to say 150 Hz give or take, and then let it roll off naturally from there at the first order rate (6 dB per every halving of frequency).

Interesting, thanks. I think I'd have to increase my capacitor to something like 1uF. Sound ok to you?

It's possible to get slopes steeper than 6dB/Oct with simple RC filters by staggering. Harrison's article presents pre-cooked design data for useful examples: https://www.dropbox.com/s/ro9kcjo25tqae7t/RC Filter Design.pdf?dl=0

The thing I worry about with this type of filter is that I can't easily short it out with a switch. The resistor that comes after the first capacitor (and connects to ground) will still be in the circuit. I think I'll need a 4PST to short over the caps and open the path to the resistors on both channels. Thoughts?

 

[kward]

Interesting, thanks. I think I'd have to increase my capacitor to something like 1uF. Sound ok to you?

Assuming first order filter, what frequency do you intend to have the filter response down by -3dB? If you want the -3dB frequency to be 100 Hertz, with a 50K volume pot, you would want a cap of size ≈ 33 nF.

 

[justinis]

Assuming first order filter, what frequency do you intend to have the filter response down by -3dB? If you want the -3dB frequency to be 100 Hertz, with a 50K volume pot, you would want a cap of size ≈ 33 nF.

Right. But it won’t be “flat” at 150 Hz in that case. I guess it’s all relative.

 

[kward]

It is. For a first order filter, you just need to position the -3dB point for best compromise between pass band and stop band.

 

[justinis]

It is. For a first order filter, you just need to position the -3dB point for best compromise between pass band and stop band.

Got it. Thank you!

 

[BinaryMike]

The thing I worry about with this type of filter is that I can't easily short it out with a switch. The resistor that comes after the first capacitor (and connects to ground) will still be in the circuit. I think I'll need a 4PST to short over the caps and open the path to the resistors on both channels. Thoughts?

You could use a single SPDT at the top of the volume pot to choose between filtered and unfiltered inputs, assuming your preamp can drive the filter network as well as the pot simultaneously without distortion.