DHAxovers for my '75 Khorns are done

[BEC]

Yes, I can.

Bob

[Deang]

Everything done "outside" the speaker effects back EMF. The "curve" shows the resistive force at different frequencies resulting from things attached to the motor. Add a cap -- and the curve changes. Add an inductor with the cap, and it changes again. Everytime you add an element -- the curve changes. Shoot, even adding a horn to the driver changes the behavior. What about the autotransformer? Do you think the curve produced from the change in voltage caused by attachment to an autoformer tap is anywhere near what the driver's raw curve is? What about LCR circuits? Sure, the driver is "merrily swinging" -- but what it's swinging is largely determined by the network.

[BEC]

Deans,

What's watt? It all comes down to watts. The driver needs a certain amount of power to produce a certain amount of output. As the impedance of the driver changes with frequency, the amp needs to be able to compensate. The resistor is like a blindfold to the amp. The amp doesn't know the impedance has changed, but it has changed.

Then there is the power the resistor uses. That one is more complex and varies with frequency. The impedance of the cap and inductor is very low at 1 khz and it is only eating up about 10 percent of the voltage there. At 400 hz and 5000 hz, its impedance is higher and would eat up more voltage, I think. I haven't worked out the formula or measured it.

Bob

[Al Klappenberger]

Bob,

I'm not sure exactly how your thinking, but your comment about the amp
"compensating" strikes me as a point of confusion. All the amp does is
provide a constant voltage. It's doesn't compensate for impedance. All it
does is provide whatever current the load connected to it happens to draw
because of its impedance. Ohms law and all that! The impedance of a K55
driver is pretty constant with frequency too. The big impedance changes
happen when you move the transformer tops. This is where the 10 ohm swamping
resistor comes into play. The impedance can go no higher than the 10 Ohm
swamping resistor. Remember too that the K55 driver has output upwards of 110
dB with the same input VOLTAGE that would generate 104 dB or less into the
woofer. This is why we cut the voltage down with a transformer. In doing
that, the impedance seen by the amp (or actually the squawker filter) goes
way up. If you figure the "efficiency" of the squawker under these conditions
it is WAY up because very little current is drawn. The output level has been
equalized to that of the woofer though. That says the the efficiency curve is
NOT a measure of frequency response flatness. This just becomes a load
impedance to the amp that goes all over the place. The swamping resistor
simply brings the impedance back down by drawing more current from the amp.
The frequency response is still flat but now so is the impedance and the
curve of "efficiency". This allows a much better squawker filter to be
designed. I think this is why efficiency is not a good measure of anything.
The term "sensitivity" is more meaningful since it relates output level with
the constant voltage from the "voltage source" an amp represents. The voltage
chosen is that which yields 1W into an 8 Ohm load (2.83V). It's only 1W if
the load happens to be exactly 8 ohms.

Al K.

[BEC]

Al,

A K-55M that I have measured has an impedance of 14.2 ohms at 400 hz. and 11.5 ohms at 5000 hz. That is not a huge difference, but it is different. The driver is still performing that way even if the amp is not seeing that impedance change.

Bob

[Al Klappenberger]

Bob,

Yep, I always just assume the K55 is about 13 Ohms. That's mainly based on measurements of the K55V. The M is a little different, but not significantly. I don't think that matters to the frequency response of the acoustic output. As long as it sees a flat response input it will do it's best to make a flat response out. After all, a driver can be 8 OHms or 16 Ohms and it still makes a flat response if it's desinged and built right. All impedance is is the ratio of voltage applied to current drawn. It's only important to a filter that drives it. An amp is a voltage source and happily gives it whatever current Ohms law says it deserves!

Al K.

[Marvel]

Then there is the power the resistor uses. That one is more complex and varies with frequency. ...

Bob

Bob,

The voltage drop on a resistor won't vary with frequency. Only caps and inductors have a reactance that varies with frequency.

Bruce

[BEC]

Bruce,

That is what I was saying. The voltage drop across the cap and inductor varies with frequency.

Bob

[Marvel]

I'm sorry Bob. It sounded like you meant just the resistor. I would

have to dig out those formulas too, that calculate capacitive and

inductive reactance. Somewhere in the bookcase ...

[Deang]

"The resistor is like a blindfold to the amp. The amp doesn't know the impedance has changed, but it has changed."

Ground covered several times before. Bob, sometimes I feel like you are a little evasive, so I guess I'll just ask straight out: Why does a swamping resistor "blindfold" the amp from the driver's variations in impedance yet Zobels and notch filters do not? If what you are saying is true, than any parallel resistance added to the driver would have the same effect. I have several loudspeaker design books and have been all over the web, and I have yet to read anything that is even remotely similiar to what you are saying here. The T/S parameters and the understanding of their relationship to each other is so they can be manipulated in order to shape a system's acoustic response, and much of this is done through the network. The whole point of plotting impedance is so one can see what the driver is actually doing while being driven, and you can't properly do this without an amplifier and signal generator. It seems to me that if the amplifier can't "see" what effect parallel resistance has on a driver with the network in place -- then there isn't much point in producing a loudspeaker impedance curve. A swamping resistor is just another form of impedance compensation/correction.

[BEC]

Dean,

Hope I don't sound evasive. I have explained this to you and John Warren has explained this to you.

How about explaining what you think the swamping resistor does that is good? I just have trouble seeing any good from it and can see some bad.

I don't know what gave you the idea that I like zobels or notch filters.

Bob

[Al Klappenberger]

Guys,

I guess it's up to me to defend the swamping resistor..

It's actually quite simple. Two resistors (or impedances) in parallel will
have a total value LOWER than the lowest of the two. So.. If you have a 10
Ohm resistor in parallel with ANYTHING the combination can go no higher than
10 Ohms. That is the "swamping" function. Even when the squawker is set to a
very low level, or is totally disconnected, the squawker filter sees no more
than a 10 Ohm load. That is a 1.25:1 error to an 8 Ohms filter and the reason
transformer taps can be moved at will. NOW: Because you are attenuating the
level to the squawker you obviously have power you don't need. Rather than
allowing the impedance to go sky high, the swamping resistor absorbs the
power you don't need instead of the amp not delivering it as would happen if
the impedance rises. Without the resistor the impedance would go way up and
not terminate the squawker filter properly. That will not allow the filter to
have the correct response. With the swamping resistor in the circuit, the
impedance stays below 10 Ohms and the filter (and amp) sees a more constant
impedance. All the power the resistor wastes is power you do not want to use
anyhow. I can see no down side to it's use at all. It's a win-win situation.
It does NOT reduce the efficiency of the speaker because it does NOT bother
the woofer at all and THAT is what determines the efficiency. You are just
bringing the efficiency of the squawker down to that of the woofer. That is
ONE AND THE SAME THING as making the impedance seen by the amp flat over the
entire frequency range.

**** Here's another way to look at it:

Ohms law says that I = E/R. The power in that circuit = E*I or (E*E)/R. Of
course, E = voltage applied, I = current in amps and R = Ohms. For example:
if you apply 2.83V to an 8 ohm load you have 1W of power (2.83*2.83)/8 = 1.0

You will agree that a loudspeaker that has a flat response also has a flat
efficiency curve too, right? WRONG! An amplifier is a voltage source, so it
will deliver a constant 2.83 volts at all frequencies and deliver any amount
of current the load draws according to Ohms law. A good amp will have a
source impedance of about .2 Ohm (that is a damping factor if 40). This means
the applied voltage stays virtually the same no matter what current is drawn
from it. A Klipschorn with an AA network will look like about 6 Ohms at 100
Hz but will look like about 32 Ohms at 2000 Hz. Connect this speaker to a
2.83V source you will have, according to Ohms law: W=(E*E)/R = (2.83*2.83)/6
= 1.33W. At 2000 Hz the amp sees 32 Ohms, W=(E*E)/R = (2.83*2.83)/32 =
0.011W! DB (power) = 10*log10(ratio) = 10*log10(.011/1.33) = -20.82 dB.
Clearly this is NOT the frequency response of the acoustic output of the
speaker. The output does not drop 21 dB at 2000 Hz! It is the efficiency
curve. This is NOT what I would call a flat efficiency curve! With the
addition of the swamping resistor, the amp will see the same impedance at all
frequencies and will have a flat efficiency curve because the same current
will be drawn from the amp at all frequencies. We are simply bringing the
squawker efficiency down into line with the woofer efficiency. With correct
design of the crossover the acoustic frequency response is not changed but
the amp sees a load that is much more like what it was rated to operate into.

NOW: Explain to me what the down side to the swamping resistor could possibly
be other than the pennies it costs to buy it. You can't, OK, I will! Here is
what PWK would have said: At 32 Ohms I can use a 13 uF capacitor to limit the
low frequency range of the squawker. At 8 Ohms I will need 52 uF to do the
same thing. It's the year 1960 and I can't find a paper capacitor that big at
a reasonable cost!

The year is now 2005 and large value polypropylene caps are available from a
long list of manufacturers at good prices. There is no reason to continue
using a designs like the A and AA networks that use a 1960 compromise in
their design!

Al K.

[greg928gts]

See what you started Kudret!

Greg

[Deang]

Nothing becomes smooth without friction!

[BEC]

Al Said: "NOW: Because you are attenuating the level to the squawker you obviously have power you don't need. Rather than allowing the impedance to go sky high, the swamping resistor absorbs the power you don't need instead of the amp not delivering it as would happen if the impedance rises."

That would of course be the choice here. You could leave the power in the amp or send it to be dissipated as heat in the resistor. Sort of like saying "Let's set our car engine to put out a constant 100 horsepower and control our speed with the brakes".

Bob

[Al Klappenberger]

Bob,

Your automobile engine is a bad analogy, but trying to relate it to this
case, a gas engine is not a constant impedance source like a power amp. In a
mechanical system, low impedance would be high torque at low RPM. High RPM at
low torque would be high impedance. As you know, a gas engine has a curve
where torque comes to a peak. It will not provide constant torque at all
RMPs. A stereo amp WILL provide a constant voltage at all currents. A gas
engine in a car also requires steps of impedance transformation (it's called
the transmission and a torque converter) to transform that peak torque RPM to
varying road speeds. That's an impedance transformer! Consider what the
"transmission" would be like if your engine would put out constant torque at
all RPMs. You would not need a transmission at all! You would pick a single
gear ratio for all speeds.

Besides, if a power amp can easily deliver 100W into a woofer, it certainly
doesn't mind delivering 2 or 3W instead of .02W at 2000 Hz! The power drawn
by the swamping resistor is peanuts in comparison to what the woofer draws. I
use a 10W resistor and it doesn't even get worm unless you are trying to
annoy the neighbors! The power requirements on your amp at squawker
frequencies is nothing compared to woofer frequencies. Trying to use your gas
engine analogy again, that's like bitching about your cars' air conditioning
drawing power from your engine and messing up your gas mileage when keeping
your windows closed reduces aerodynamic drag and actually increases mileage!

Al K.

[BEC]

Al Said: "Trying to use your gas engine analogy again, that's like bitching about your cars' air conditioning drawing power from your engine and messing up your gas mileage when keeping your windows closed reduces aerodynamic drag and actually increases mileage!".

Mythbusters took that one on recently. Need them to get working on crossovers.

OK, If it is good for the squawker, why is it not good for the woofer. That thing varies impedance all over the place. From 5 ohm or so to around 68 ohms for one made since 1985.

Bob

[Kudret]

See what you started Kudret!

Greg

Greg, Ironically, this is all above my head. [:|]

All I needed to know was which taps to use on 3636. LOL.

Kudret

[Al Klappenberger]

Bob,

"OK, If it is good for the squawker, why is it not good for the woofer. That thing varies impedance all over the place. From 5 ohm or so to around 68 ohms for one made since 1985. "

First off, I doubt seriously that the woofer impedance changes all that much. Any woofer that measure 68 Ohms is in serious need of a Bob Crites voice coil fix! My middle Belle of the three was made in roughly 1985 and it measure just like the normal K33 driver. The direct answer is that the woofer is always the lowest efficiency driver. You need to bring the others down to it. The resistive part of the woofers' complex impedance dictates the impedance of my netwroks across the board. Actually, The 700 Hz extreme-slope network I first designed for my Belles (the ones described in my downloadable paper on extreme slope netwroks) was designed for a 4 Ohm woofer (resistive). The K33 is more like 6 Ohms in sereis with 1 mHy. It had a rough response when it was terminated with my K33 simulator load. I found it worked much better with a 10 Ohm (the maginc 10 Ohms again) connected across the woofer! I tested the sensitivity (efficiency) of the woofer and found that it measured virtually the same as with my ES500T network that I just finished and installed today. It requires no Zobel or any other compensation, so I didn't need the 10 Ohm resistor any more. They both measured 100 dB SPL at 1 meter with 1W input. BTW: I will be posting the design for the ES500T shortly for the DIY croud. The old "X700" netwroks described in the paper were gutted for the parts! If you're interested, I have a picture of both networks posted on my personal web site (the www button below).

BTW: If there has been some serious study done about the air conditioner drain versus window open air drag, I'd like to see it! It does sound too good to be true!

Al K.

[BEC]

Is this OT or what?

Windows down vs. air conditioning

"Urban puzzle": it is more efficient, on a hot day, to run with the A/C on and windows up than to run with windows down (b/c of increasing car's drag).

Computer-based mpg measurements:

11.7/11.8 with A/C on and windows up

11.7/11.8 with A/C off and windows up

11.3 with A/C off and windows down

So, according to the computer, it's better to use A/C with windows up.

This was too quick and easy for TV, so they decided to stage a seven hour marathon, race-til-you're-empty duel, with Jamie driving an SUV with A/C on and Adam driving an SUV with windows down. Though, once the safety inspector intervened, it was no longer a seven-hour marathon, it was a bit slower (45mph instead of 55mph), and a lot shorter (only 5 gallons each).

Jamie's A/C car ran out of gas first -- Adam's windows down SUV ran for another 30 laps -- completely contradicting the computer mpg estimate. Computer estimate based on air flow into the engine, so it would appear that it is unable to properly model the difference between A/C and windows down.

Mythbusted

Al,

The 68 ohms was measured on a new K-33E right out of the box at 35 hz. The older square magnet models measure about 35 ohms for ZMax at around 27 hz.

Bob