Half Power switch - best way?

[Dave H]

You have to be careful thinking about the damping of a speaker by a low source impedance. The audio industry has traditionally over sold the size of the effect.

Thanks, that is exactly what I was thinking about.

[Mike Sulzer]

So why not a really simple "speaker simulator" to soak up half the power? If 8 ohms, then move the tap to 4 ohms, and put in parallel with the speaker a resistor in series with an inductor. The resistor has the same resistance as the dc resistance of the speaker, the inductor the same as its inductance. Sure, it is not perfect, but none of the reasonable options are. And this is really simple and easy to do as long as you can measure the speaker and buy the right components, or at least reasonably close. For example, I just took a brand new Red Fang out of its box, and it reads 6.8 ohms, and 1 mH, measured at 120 Hz with an Extech 380193. You can find two resistors that add up very close, and the inductor is a standard value. Of course, you could engage in a little experimental sound modification by shifting the components off a bit if you like to experiment.

Edit: To continue the example, I have 8 ohm and 50 ohm power resistors in my junk box. Parallel gives 6.9 ohms. An air core 1 mH core (designed for crossovers) costs a bit over $10 from Parts Express. It is .5 ohm, so you could knock down the resistor value by that amount.

[jmaf]

3) Other? (no triode switch stuff please).

Derive a 2nd +B voltage that is 0.707 times the main +B and switch between the two. When you use the 2nd voltage you'll have half the power. E.g. If you got 500 VDC, something around 350VDC will give you half the power.

Easier if the amp is self biased.

[Chuck H]

So why not a really simple "speaker simulator" to soak up half the power? If 8 ohms, then move the tap to 4 ohms, and put in parallel with the speaker a resistor in series with an inductor. The resistor has the same resistance as the dc resistance of the speaker, the inductor the same as its inductance. Sure, it is not perfect, but none of the reasonable options are. And this is really simple and easy to do as long as you can measure the speaker and buy the right components, or at least reasonably close. For example, I just took a brand new Red Fang out of its box, and it reads 6.8 ohms, and 1 mH, measured at 120 Hz with an Extech 380193. You can find two resistors that add up very close, and the inductor is a standard value. Of course, you could engage in a little experimental sound modification by shifting the components off a bit if you like to experiment.

Edit: To continue the example, I have 8 ohm and 50 ohm power resistors in my junk box. Parallel gives 6.9 ohms. An air core 1 mH core (designed for crossovers) costs a bit over $10 from Parts Express. It is .5 ohm, so you could knock down the resistor value by that amount.

That would be a fixed attenuator. "Attenuator" seems to be a dirty word. I use one similar to your description, except that the resistive load is a rheostat wired to maintain the load resistance/impedance (somewhat) with variable adjustment between the speaker and dummy load. I like mine.

A fixed value circuit could be adjusted for, say, 100/35 watts too. Rather than the usually less gratifying 100/50 watt circuits.

[Mike Sulzer]

That would be a fixed attenuator. "Attenuator" seems to be a dirty word. I use one similar to your description, except that the resistive load is a rheostat wired to maintain the load resistance/impedance (somewhat) with variable adjustment between the speaker and dummy load. I like mine.

A fixed value circuit could be adjusted for, say, 100/35 watts too. Rather than the usually less gratifying 100/50 watt circuits.

I agree that more lost than half is better, but we gotta give the wizard what he wants! I think it is important to maintain the correct inductance in the "speaker simulator", which is why I prefer the fixed soaker. If you want to do more than half, you need two resistors and two inductors, but that really is not too hard to do.

[Chuck H]

+1
To that end the circuit should be simple and efficient. Also, IMHE, big inductors dissipating 50+W can be problematic with high gain circuits. It might be necessary to locate the "load" in a second box away from the main chassis. As to a circuit involving the second inductor, that can be problematic too. The available parts in the smaller values for high power are air core. It works fine with humbuckers. If you have SC pickups you can't get within ten feet or it sounds like a whistling pete going off. I imagine it would be impossible to put such an inductor inside the amp chassis without instability.

I solved for that by using two inductors of half value wired in series/out of phase and stacked tight with a copper plate between them to minimize mutual inductance. It worked pretty well for the circuit function and eliminated the pickup feedback. This isn't shown in the schem below.

I really enjoy the broad variable effect with the rheostat more than I would absolute load accuracy. But in keeping with the OP then, my vote goes back to the cathode lift or fixed power scaling. But here's the circuit anyway Clearly not good to 100W but that can be adjusted for.

[GainFreak]

Speaking of reactive attenuators this one is from Victor Kempf from AMT:



I guess their Power Eater PE-120 is based on this one.

[Mike Sulzer]

+1
To that end the circuit should be simple and efficient. Also, IMHE, big inductors dissipating 50+W can be problematic with high gain circuits. It might be necessary to locate the "load" in a second box away from the main chassis. As to a circuit involving the second inductor, that can be problematic too. The available parts in the smaller values for high power are air core. It works fine with humbuckers. If you have SC pickups you can't get within ten feet or it sounds like a whistling pete going off. I imagine it would be impossible to put such an inductor inside the amp chassis without instability.

I solved for that by using two inductors of half value wired in series/out of phase and stacked tight with a copper plate between them to minimize mutual inductance. It worked pretty well for the circuit function and eliminated the pickup feedback. This isn't shown in the schem below.

I really enjoy the broad variable effect with the rheostat more than I would absolute load accuracy. But in keeping with the OP then, my vote goes back to the cathode lift or fixed power scaling. But here's the circuit anyway Clearly not good to 100W but that can be adjusted for.

Separate box for sure! I look at one of the advantages of your circuit as this: you do not have to modify the amp. I like your feedback canceling technique, a sort of reverse humbucker. Maybe mu meal sheet between them would be better than copper, and you use some mu meal tape to increase the shielding effect to the outside world.

[mhuss]

Back to the two active and two idling tubes idea...wouldn't the active tubes see a different load looking into the output impedance of the 'passive' tubes plates in parallel with the OT? The open cathode would be an open circuit.

[wizard333]

So why not a really simple "speaker simulator" to soak up half the power? If 8 ohms, then move the tap to 4 ohms, and put in parallel with the speaker a resistor in series with an inductor. The resistor has the same resistance as the dc resistance of the speaker, the inductor the same as its inductance. Sure, it is not perfect, but none of the reasonable options are. And this is really simple and easy to do as long as you can measure the speaker and buy the right components, or at least reasonably close. For example, I just took a brand new Red Fang out of its box, and it reads 6.8 ohms, and 1 mH, measured at 120 Hz with an Extech 380193. You can find two resistors that add up very close, and the inductor is a standard value. Of course, you could engage in a little experimental sound modification by shifting the components off a bit if you like to experiment.

Interesting idea. But it's been tried and causes instability?

Having heard the Koch setup with just a resistor in parallel with the speaker, it does have a very audible volume drop effect. It also peels off a lot of high end, making the amp warmer sounding. What benefit would you get adding the inductor? Trying to preserve the 'full on' tone by making the load reactive?

Derive a 2nd +B voltage that is 0.707 times the main +B and switch between the two. When you use the 2nd voltage you'll have half the power. E.g. If you got 500 VDC, something around 350VDC will give you half the power.

Easier if the amp is self biased.

Also interesting. Won't we get a significant tonal change with a big change in B+ though? Seems like to implement that and maintain a similar tone, you'd have to have your OT running on a separate PT secondary tap from the rest of the amp, so when you switch to lower B+ the preamp tubes would still get about the same voltage?

[g1]

Yes, you can take a separate line just for the OT and screens of the power tubes. The preamp voltages will be unaffected. That's how the circuit in post #14 does it.

[Dave H]

Back to the two active and two idling tubes idea...wouldn't the active tubes see a different load looking into the output impedance of the 'passive' tubes plates in parallel with the OT?

I guess the active tubes will see the impedance given by the slope of the plate VI curve at the idle current (in parallel with the OT). I've done a sim in LTSpice and I calculated an impedance of 67k for a 'passive' 6V6.

[Mike Sulzer]

Interesting idea. But it's been tried and causes instability?

Having heard the Koch setup with just a resistor in parallel with the speaker, it does have a very audible volume drop effect. It also peels off a lot of high end, making the amp warmer sounding. What benefit would you get adding the inductor? Trying to preserve the 'full on' tone by making the load reactive?

If you simulate the speaker in the top end by giving the "speaker simulator" the same inductance as the speaker then you will not lose the top. The resistor alone steals high frequencies since its impedance is lower than the speaker at high frequencies. With the inductor, the high frequencies split properly between the speaker and the simulator. Of course, this simple simulator lacks the low frequency resonance (which is in Chuck H's attenuator), but that might not be as important to you depending on what you like.

Why would it cause instability? The speaker does not.

[Leo_Gnardo]

If you simulate the speaker in the top end by giving the "speaker simulator" the same inductance as the speaker then you will not lose the top. The resistor alone steals high frequencies since its impedance is lower than the speaker at high frequencies. With the inductor, the high frequencies split properly between the speaker and the simulator. Of course, this simple simulator lacks the low frequency resonance (which is in Chuck H's attenuator), but that might not be as important to you depending on what you like.

Good observations, and there have been speaker attenuators that address the inductance issue.

BUT one thing that's neglected so far, our ear's Fletcher-Munson curve. People naturally disregard frequency extremes at lower volumes. What to do? Even if you design an attenuator that perfectly tracks inductance to keep the high frequency curves matched, and resonance for the lows, it's still not enough to satisfy the human ear. You have to exaggerate hi & low boosts in the attenuator in an attempt to keep the sonic "flavor" satisfying at lower listening levels. "Loud switch" anyone?

[mhuss]

"Loud switch" anyone?

The THD Hot Plate actually has "Bright" and "Deep" switches, exactly for this reason.

I have one of the original Weber "Mass" units - basically a speaker motor without a cone (it has two spiders to make up for the lost compliance). It looks from the back like a 4" speaker with a big magnet. I like that a lot for a realistic load, but the one's they're making now aren't made that way anymore.