A "Better" idea for an attenuator

[KGW]

Watch your individual power dissipations, though. 1/10 the length of a 100 watt resistor is only a "10 watt resistor."

--mark

That's a good point. So the series resistor in the Airbrake would be under rated? Another issue is that lower frequencies draw more power. On top of that, the impedence and power all change with frequency. An 8 ohm speaker has an impedence of 8 ohms at say 4K, but the impedence can be all over for other frequencies. Note, my spreadsheet shows the input and output impedences at the fundamental frequency.

On my Hotplate, I can hear the sound degradation at any attenuation level. Sustained single notes start getting thinner and lose low end. The effect is less noticable on chords and shorter single notes. However, the Hotplate is better than nothing.

[mhuss]

So the series resistor in the Airbrake would be under rated?

I don't know, I haven't done the math. Is the AB rated for 100 watts?

Another issue is that lower frequencies draw more power. On top of that, the impedence and power all change with frequency.

Only on a real speaker, not a resistor. So, this may affect the sound, but it will only marginally affect the power dissipation in the resistive attenuators we're talking about, as we're typically sucking up 2/3 or more of the power with resistors, even at mild attenuation settings.

On my Hotplate, I can hear the sound degradation at any attenuation level.

A lot of people have said this (about many attenuators). There's so many physical and psycho-acoustic things going on, I wonder if it's not just the attenuation itself. In other words, I'll bet even a theoretically "perfect" attenuator would appear to color the sound, just because "quieter sounds different," and a change in volume affects the physics (speaker/air/guitar) differently.

--mark h

[unklmickey]

On my Hotplate, I can hear the sound degradation at any attenuation level.

A lot of people have said this (about many attenuators). There's so many physical and psycho-acoustic things going on, I wonder if it's not just the attenuation itself. In other words, I'll bet even a theoretically "perfect" attenuator would appear to color the sound, just because "quieter sounds different," and a change in volume affects the physics (speaker/air/guitar) differently.
--mark h

mark,

i won't even address the physics of sustain/acoustic feedback.
i'm not remotely qualified to do so in a comprehensive manner.
suffice it to say, this can be a part of the equation that should not be dismissed.
is my opinion that a purely resistive attenuator's affect on the loading of the amp, and the damping of the speaker, is the largest single factor affecting tone.
IMHO, the psycho-acoustic phenomenon, and physics run a close second and third.


a friend and i, put together a small multi-track recording studio in the early '80s.
when i made recordings of his band, he was able to eventually adjust to hearing his guitar through headphones,
with his amp mic'ed in an isolation booth.
many others were not so quick to adapt, and their performances suffered.
we often did "band" recordings, with all instruments and vocal recorded simultanteously, but on separate tracks.
since the drums were only allotted 2 tracks "for stereo", i needed to focus on getting their mix perfect.
i only needed to hear (in the control room) the other instruments well enough to get a clean recording.

a guitar player that was in the control room with me, thought the quiet guitar sounded like dung.
however, i played back the guitar track alone, without additional effects, but at a louder level.
the guitarist was amazed at my skill in extracting such a wonderful tone from such a lackluster recording.
needless to say, when i judiciously panned the guitar and suBtle temporal effects for the final mix, i was now seen as some sort of Genious.
while i'll admit i could make a guitar sound as big a house, without being rediculously loud, part of it had to do with listening level.
during the playback for the final mix, the volume level in the control room was always more modest. you want to hear things the way a person would in an "average" listening environment.

the bottom line is: YES, volume does affect the perception of "tone".

...

Another issue is that lower frequencies draw more power. On top of that, the impedence and power all change with frequency.

Only on a real speaker, not a resistor. So, this may affect the sound, but it will only marginally affect the power dissipation in the resistive attenuators we're talking about, as we're typically sucking up 2/3 or more of the power with resistors, even at mild attenuation settings. ...

i view this a bit differently, mainly having to do with how much energy is required to achieve "equal loudness" at different frequencies.
but that's a conversation for another time.

my fingers are getting tired.


cheers,

unk

[UR12]

mark,

is my opinion that a purely resistive attenuator's affect on the loading of the amp, and the damping of the speaker, is the largest single factor affecting tone.
IMHO, the psycho-acoustic phenomenon, and physics run a close second and third.
unk

Interesting perceptions! So what if a person built a speaker cab that sounded good at attenuated levels (less dampened)?

[unklmickey]

...So what if a person built a speaker cab that sounded good at attenuated levels (less dampened)?

i'm about to go on vacation, and i smell a good discussion evolving into something larger than life.

Dana,

that is a simple idea on the surface, but gets more complicated the more you look.

think of a speaker as both a motor and a generator.

if you put an electrical impulse into a speaker without damping, you don't get a click.

you get a thump.

the mass, spring of the suspension and spider and air in the enclosure all act as a system.

at any resonances in the system, the electrical impedance of the speaker is high.

when working within the limitations of the amplifier, one main focus is on having the appropriate damping on the speaker to control cone overshoot.

also we need to be concerned with maintaining a somewhat tame frequency response in the acoustic output.

but BEYOND the limitations of the amplifier (where guitarists LIVE), there in another huge concern.

the loading (or lack thereof) that the speaker presents to the transformer radically affects what is happening on the plates of the output tubes.

it is way past my abilities to accurately understand and express all that is occuring. but i am certain it will be excruciatingly difficult, if not impossible to get the same distortion products in the output stage when you have a purely or mostly resistive load.

so you could create a speaker system that would sound "normal" on the other side of an attenuator, when played clean.

but it wouldn't impress the same characteristics on the plates when you get into the region of power amp distortion.

some attempts have been made to simulate the load a speaker presents with RC networks and even voice coils.

they are a first step, but they probably have a long way to go.


one possible strategy, would be to first mimic electrically, as close as possible, the load a speaker presents.

then rather than driving a speaker directly, the electrical signal would be sent to a second amplifier.

the second amp would be nothing like a guitar amplifier.

it's task would be to tailor the response curve (this changes according to volume), to "normalise" the output.

and of course it would be much more powerful than necessary, and have a high damping factor.


is that good strategy? i don't know, i'm just imaginating and thinking out loud.

most times when i do that, i end up off on a tangent.

but, once in a while some good comes of it.




cheers,

unk

[d95err]

Watch your individual power dissipations, though. 1/10 the length of a 100 watt resistor is only a "10 watt resistor."

I'm not so sure about that. Looking at the huge power resistors used in the airbrake pictures I've seen, it looks like the heat will spread through the entire 100W resistor, not just the part being used. The effective heat tolerance may be slightly lower 100W, but probably not as low as 10W.

[KGW]

Watch your individual power dissipations, though. 1/10 the length of a 100 watt resistor is only a "10 watt resistor."

I'm not so sure about that. Looking at the huge power resistors used in the airbrake pictures I've seen, it looks like the heat will spread through the entire 100W resistor, not just the part being used. The effective heat tolerance may be slightly lower 100W, but probably not as low as 10W.

I've been thinking about that one as well. So say we have a 100k/2W pot, then P = I^2R, thus I = sqrt(P/R), so the maximum current at 100k is sqrt(2/100000) = 4.47mA. So say we draw 4mA. Then at 100k, the voltage is 400V. With the pot at 80%, we have 80K*4mA = 320V and P = 1.28W. At 50%, the power goes down to 0.8W. So the math doesn't lie, so mhuss is correct.

We would have to design the attenuator so that at maximum attenuation, we can handle the power. I will need to crack open some books to really understand this. I may put a dummy load on my hotplate and do some measurements before and after at various settings.

If someone else understands the power design issues, an explanation would be most appreciated.

[mhuss]

The way to do the maths is simply consider the tapped resistor as several individual resistors and do the Ohm's law power calcs for each one individually. This will tell you how much power will be dissipated by each section.

A ceramic wire wound resistor is just a length of resistance (think "toaster" ) wire wound on a ceramic form. Ceramic withstands heat but does not conduct heat very well, which is why they use it as a substrate.

To make an, e.g., 20 ohm, 100 watt resistor this way, you select a thickness/composition of resistance wire that will both possess the correct resistance and handle the current ( sqrt(P/R) ~= 2.2 A ) that the wire can carry before getting red hot.

Half of that length (10 ohms) will therefore be able to dissipate half the power (same current, less resistance). (2.2)^2 x 10 = 50 watts.

--mark h

[Normster]

Watch your individual power dissipations, though. 1/10 the length of a 100 watt resistor is only a "10 watt resistor."

I'm not so sure about that. Looking at the huge power resistors used in the airbrake pictures I've seen, it looks like the heat will spread through the entire 100W resistor, not just the part being used. The effective heat tolerance may be slightly lower 100W, but probably not as low as 10W.

The AirBrake is slightly different from a conventional L-pad in that the shunt comes before the series resistor, not after it. In the AB design, about 80% of the power is dissipated through the 100watt fixed resistor so the remaining current passed through the series resistor and rheostat is less of a strain. Also, at the highest attenuation settings, nearly the full length of the adjustable resistor is being used. As with other KF designs, it's simple and elegant.

[slajeune]

Hi Normster,

Not quite sure. Based on Allynmey's layout, it seems as the tip goes to the series resistor and then the shunt. I don't like looking at layouts because they cause some confusion (not always clear from the layout). Looking at the attachement is what it looks like to me.

Cheers,
Stephane.

[Normster]

The schematic is correct, except that J2 is the input.

[slajeune]

Interesting. Now I see that the layout is from the bottom (duh). The input is from the right side (as I'm used to seeing the input on the left side, kinda also missed the BIG arrows....).

Thanks, it clearer now!

Cheers,
Stephane.

[Doug H]

The auto-former idea may work, but I'm guessing that it may not sound very good.

I wouldn't use an autoxformer by itself, as that will vary the load on your amp too much. SS hi-fi amps don't care about load variations that much but as we all know, tube amps are pretty sensitive to that.

the bottom line is: YES, volume does affect the perception of "tone".

I agree. As do Fletcher & Munson... :

http://www.webervst.com/fm.htm


The newer MASS models look interesting because they divide up the spectrum into high and low freq bands and allow you to attenuate each independently. This might allow you to compensate some of the perceived bass loss at low volume levels. The price isn't bad either. I'd like to check one out at some point.

[KGW]

The AirBrake is slightly different from a conventional L-pad in that the shunt comes before the series resistor, not after it. In the AB design, about 80% of the power is dissipated through the 100watt fixed resistor so the remaining current passed through the series resistor and rheostat is less of a strain. Also, at the highest attenuation settings, nearly the full length of the adjustable resistor is being used. As with other KF designs, it's simple and elegant.

It still looks like the series resistor is under rated, though the wattage goes up with the amount of attenuation. For example, the 6 ohms at the "R4" position is 24% of the series resistor, so there is far less that 100W for the first level of attenuation. Also, since the signal will follow the path of least resistance, "R4" sees a lot more current than R1.

NOTE, I have not heard an Airbrake, however, the Hotplate and Weber MASS are purported to sound better. I guess that I could build an Airbrake and see what I think.

[908ssp]

The Alessandro Attenuator is the ladder type. Two huge resistors with multiple taps on both one in series and one parallel so the total impedance stays constant throughout the range of the attenuation. Still doesn't sound nearly as good as a Richter.