Questions about a basic atenuator

[Pedro Vecino]

Hi all. I built this pad for a 1 watt amp (ECC82 in pushpull). Originally is to reduce it to 0.1 watt, but results excessive attenuation (too much compression) and sounds pretty dark.
Could someone give me directions to transform it to obtain a lower attenuation and/or less dark?
I dŽont understand the relationship of series resistor/parallel resistor for each impedance.
(SW1A and SW1B are two lines of the same switch).
Thanks!

[kleuck]

Why a 3,3”f to ground ????

[Enzo]

82 ohms in series with the speaker?

[Pedro Vecino]

Hello again. This is the Marshall JCM-1 attenuator. Basically works well but with excessive attenuation. With an 8 ohm speaker connected to 8 Ohms output and attenuation activated, the resistance across the 47 Ohms resistor in series is approximately 6 Ohms.
My interest is in reducing the attenuation (half of the original, for example) while maintaining the ratio of impedances (which I imagine will be correct).
Thanks!

[Steve Conner]

This is an L-pad, so you can use the L-pad design equations to achieve any attenuation you want.

It is actually a backwards L-pad, so you have a dummy load resistor across the amp output, and a resistor in series with the speaker. This drives the speaker from a high-impedance source, so you get a more lively sound with emphasized treble and bass resonance. This is a better match to the sound of a tube amp without NFB, and also gives you some Fletcher-Munson "loudness". (I thought I invented this trick, but I'll share it now that I see Marshall thought of it first. )

For 10dB attenuation and a 16 ohm load, you could use an 18 ohm dummy load and a 47 ohm series resistor. For an 8 ohm load use 8.2 ohms and 22 ohms. Don't use the 3.3uF cap, this is probably the reason for the "dark sound".

Note: If you designed the L-pad forwards, for a 16 ohm load you would have something like 10 ohms in series with the amp output and 4.7 ohms in parallel with the speaker, so the output impedance would be 10 times smaller.

[Pedro Vecino]

Thanks, Steve!
Capacitor (not polarized) effectively limits the harmonics and gives some roundness in the top. I Will adjust its value to do it more discreetly.
Regarding the resistors relations, you propose to set differently for each impedance the dummy resistor. In the original scheme his value is common to both impedances. Would it be possible to adjust the others (which are in series) keeping the same value in the above? A simple division by two in each maintaining the value of 18 Ohms in the dummy resistor? Other value, perhaps?
Thanks again!

[Steve Conner]

The resistor values are not at all critical.

The dummy load resistor should be about equal to the load impedance that the amp wants to see, and the series resistor should be equal to the speaker impedance times (attenuation ratio-1). 10dB is an attenuation ratio of 3.16, so the series resistor should be 2x the speaker impedance.

But the speaker impedance varies so much with frequency that this can only be a rough guide, so you should adjust the series resistor by ear to get the effect you want. This is why I suggested 47 ohms instead of the theoretical 32.

For small attenuation ratios, in theory you should increase the value of the dummy load resistor a little, to take into account that (speaker + series resistor) appears in parallel with it. The L-pad design equations do this, and that's why the circuit uses 18 ohms here instead of the theoretical 16. But in practice, for 10dB attenuation it probably won't matter. Either 16 or 18 would work fine.

In the original schematic, the dummy load resistor is always connected to the 16 ohm tap, even if the load is on the 8 ohm tap. That's why the value is always 18 ohms. (It will reflect to the 8 ohm tap as 9 ohms.)

Edit: I guess I didn't quite answer your question. Yes, you can just change the 82R resistor to 47 or 33, and the 47R to 22 or 15, and leave the 18R as it is.

[Pedro Vecino]

Hello again. At the 8 Ohms output I connected a 47 Ohms resistor paralleled over the same original value (23 Ohms total), and the result is perfect (lighter attenuation than the original although very appreciable). I have also reduced the capacitor value to 1.5 uF and the result is also very good. Reducing very slightly higher harmonics (compared without capacitor) it feels more "solid tone" with the attenuation selected. The two steps produces a very good balance with original sound without attenuation.
I need to adjust the 16 Ohms output but the system is clear as the spring water of Glenlivet
Thanks, Steve!