Attenuator build help

[voskarp]

Hello everyone, my first post here so I hope I put it in the right section.

I have built an attenuator into my amplifier cabs (a Vox AC4C1 and a 5e3 Deluxe clone I made). I've made a simple attenuator out of a bypass switch and a L-Pad (16 Ohms, 50W for the Vox and 8 Ohms 100W for the Deluxe).

I also plan to ad a 4.7uF cap (in series with a pot, 500kOhms???) for bypassing some needed treble.

I'm thinking of adding inductor coils as an additional reactive load, after the bypass switch and before the L-Pad. As far as I understand it there should be one with half the impedance in series and one with the same as the speaker in parallel. Am I on the right track with this, and if so what values should I be looking at(?) or have I got this all wrong?

[voskarp]

OK, did some more research and found this:

http://www.marshallforum.com/media/atte ... 1566625439

I guess I could use the first stage, before the L-Pad, right?

[martin manning]

That attenuator was just discussed here: https://ampgarage.com/forum/viewtopic.php?f=6&t=32853 You can lop off stages depending on how much level reduction you need. The switches do exactly that.

[voskarp]

Thanks!

Instead of the R stages (R1-R8) I should be able to use a variable L-pad, as I already have it installed.

Edit: Maybe replace R3-R8 with variable L-Pad is better. Leaving R1 and R2 as they are.

[voskarp]

Something like this: EDIT: Wrong, correction further down.

[martin manning]

R2 is part of the first stage so yes, keep that. What is the resistance of the pot in your variable L-pad?

[voskarp]

L-Pad for the Deluxe is 8Ohms, same as the speaker.

[martin manning]

Sorry, missed that in your table. Basically you are doing as shown in the image below, with the -10dB stage switched out (from http://www.marshallforum.com/media/atte ... 219.11164/ ).

[voskarp]

Yes, it's a similar circuit.

The problem I've been having with using only the rheostat L-Pad is that not much happens for the first 75% of the sweep, and then at "bedroom" level a slight turn attenuates too much. I think that adding a stage before the rheostat L-Pad will help with that.

It is also a bit of a tone killer, but not alarmingly much. That's why I wanted to ad the treble bypass, and the reactive stage; loosing low end is actually a good thing (as lower frequencies go through the walls easier).

[martin manning]

Notice in the circuit above the variable L-pad goes from 55Ω shunt followed by 6Ω series to 5Ω shunt followed by 16Ω series, which keeps the impedance [edit] seen by the upstream stage high. Your 8Ω pot will not do that as it is turned down. I've skimmed through that thread on the Marshall forum, and replicated some of the simulations. I think I would build the simpler M2 version, with the single air core inductor, as it does seem to function as well as the more complex M. You could wire up a couple of stages, say the reactive stage and the -10 dB stage above, and see if it meets your needs before doing a lot of work.

[voskarp]

I'm not sure if there's something I don't get, but the "pot" is in itself a variable 8Ohm L-Pad.

This is how I use it now: EDIT: Wrong, correction further down.

[martin manning]

Look at a simple DC circuit, where the amp's output is fed into the 8Ω pot's wiper, and the speaker is nominally an 8Ω load across the pot's element (your schematic).

Case 1: The pot is turned all the way up (100% rotation). The amp sees the pot as an 8Ω shunt, followed by zero Ω in series, followed by the 8Ω speaker.
The amp sees 8//8 = 4Ω load, and the speaker is absorbing 50% of the power output, a 3 dB reduction.

Case 2: At 50% rotation you have 4Ω shunt, 4Ω series, and 8Ω speaker.
The amp sees 4//(4 + 8) = 3Ω load, and the speaker is absorbing 17% of the power delivered, for a 7.7 dB reduction.

Case 3: At 25% rotation you have 2Ω shunt, 6Ω series, and 8Ω speaker.
The amp sees 2//(6 + 8) = 1.75Ω load, and the speaker is absorbing 4% of the power delivered, for a 14 dB reduction.

Case 4: At 0% rotation the Amp sees a 0Ω load, and the speaker power goes to zero.

A correctly configured L-pad should reduce the power absorbed by the speaker, while presenting nearly the same load impedance to the amp. You could turn the pot around and run the wiper to the speaker (like a normal volume control), but the amp will still see varying load impedance (4Ω at 100% and 8Ω at 0%), and the output impedance as seen by the speaker will approach zero as the volume is turned down.

Edit: corrected dB calcs

[voskarp]

EDIT: Schematics corrected.

[voskarp]

The variable L-Pad looks like this:

The last value before open was 40Ohms.

Measurement about half way was 4Ohms serial and 20Ohms parallel.

Measuring on the disconnected speaker plug I get 7,3 to about 10 in the middle and 7,5 full rotation.

[martin manning]

That makes a lot more sense. At 50% rotation the amp sees 4 + 8//20 = 9.7Ω, and the speaker is getting 42% of the power, for -3.8 dB. I'd have guessed you would see 4 + 20//4 = 7.3Ω at the unconnected output at 50%, so that agrees with what you measured. The speaker is looking at an output impedance that approaches zero as you turn down, though.