total impedance presented to amplifier

[soundnovice]

hi all
i am trying to understand passive crossover design for 3-way speakers. when all three drivers are connected to signal source (amplifier) then how much impedance the amplifier sees? assume all three have 8 ohms of impedance.

thanks,
pranam

[AllenB]

Here is an example of a fairly typical crossover.

The woofer impedance is in yellow and the tweeter impedance is in blue. This is the way they are seen individually from before the crossover, i.e. from the amps point of view. The combined impedance is shown in red.

In this case the impedance is higher around the crossover point and the individual drivers virtually leave the circuit out of their range so the three will not really be in parallel.

[Inductor]

Originally Posted by soundnovice

hi all
i am trying to understand passive crossover design for 3-way speakers. when all three drivers are connected to signal source (amplifier) then how much impedance the amplifier sees? assume all three have 8 ohms of impedance.

thanks,
pranam

In one word is 8 ohms. But minimum impedance can be lower, look at the impedance curve of drivers. Usually 6 ohm or similar in this case. Final min. impedance dependent of the crossover design and drivers. Plus cables.

[soundnovice]

Originally Posted by AllenB

Here is an example of a fairly typical crossover.

The woofer impedance is in yellow and the tweeter impedance is in blue. This is the way they are seen individually from before the crossover, i.e. from the amps point of view. The combined impedance is shown in red.

In this case the impedance is higher around the crossover point and the individual drivers virtually leave the circuit out of their range so the three will not really be in parallel.

thanks Allen. that solves my doubt. is it always impedance more at xover point? any solution to reduce it? thats why i believe its bad to cross between 300hz and 3000hz as at xover point sound gets attenuated. is my understanding correct?
also which software you are using to generate the impedance curves

[soundnovice]

Originally Posted by Inductor

In one word is 8 ohms. But minimum impedance can be lower, look at the impedance curve of drivers. Usually 6 ohm or similar in this case. Final min. impedance dependent of the crossover design and drivers. Plus cables.

will this theory holds good even when i apply impedance equalization for mids and tweeters? i mean the impedance seen by amplifier will remain 8 ohms after adding a zobel network to mid and a paralle LCR network to tweeter. on adding these impedance correction circuits the equalized impedance will become equal to Re of the driver (correct me if i am wrong here) right?

thanks

[AndrewT]

Originally Posted by soundnovice

I believe its bad to cross between 300hz and 3000hz as at xover point sound gets attenuated. is my understanding correct?

No, you have it back to front.
The power amplifier has a finite output impedance.
The speaker has a finite impedance.
The voltage at the speaker is Vout * (Rload / (Rload+Rout))

let's put some numbers in to that equation.

Vout = 10Vac
Rout = 0r5
Rload = 8r0
The speaker voltage = 10*8/(8+0.5) = 9.41Vac.
The power available at the speaker terminals is V^2 / Rload = 9.41^2/8 = 11.07W

Let the impedance at the crossover rise to 10ohms. Rload becomes 10r0.
The power available at the speaker terminals is 9.07W. Gee whiz we have lost 2W !!

But, the speaker manufacturer designs the speaker to have a flat sounding frequency response, when fed with a flat Voltage response.
The speaker instead of reading 9W vs 11 W reads 9.5Vac instead of 9.4Vac.

This is an increase of 0.1dB. Almost certainly not audible.
As the Power Amplifier output impedance (and all the other impedances of the cables and connectors) gets lower than slight increase of 0.1dB in the example gets even less.
At the limit, if the output impedance is zero the gain/loss as speaker impedance changes is 0dB.

[AllenB]

The impedance is normally higher, but you can cross any way you like so it really depends. If you stick to the normal ways of crossing it should be reasonably predictable.

If you want to flatten this kind of peak you might add a tuned circuit (RCL) across the amp.

The impedance is higher, but then there is energy storage around the crossover region. I'm not sure I understand your question about attenuation. You can set the levels where you want them. Some crossovers have a flat response on axis but a reduced power response. For what it's worth, I use a crossover just below 1kHz.

The software is xoversim. The curves were measured and the crossover simmed onto them.

[soundnovice]

Originally Posted by AndrewT

No, you have it back to front.
The power amplifier has a finite output impedance.
The speaker has a finite impedance.
The voltage at the speaker is Vout * (Rload / (Rload+Rout))

let's put some numbers in to that equation.

Vout = 10Vac
Rout = 0r5
Rload = 8r0
The speaker voltage = 10*8/(8+0.5) = 9.41Vac.
The power available at the speaker terminals is V^2 / Rload = 9.41^2/8 = 11.07W

Let the impedance at the crossover rise to 10ohms. Rload becomes 10r0.
The power available at the speaker terminals is 9.07W. Gee whiz we have lost 2W !!

But, the speaker manufacturer designs the speaker to have a flat sounding frequency response, when fed with a flat Voltage response.
The speaker instead of reading 9W vs 11 W reads 9.5Vac instead of 9.4Vac.

This is an increase of 0.1dB. Almost certainly not audible.
As the Power Amplifier output impedance (and all the other impedances of the cables and connectors) gets lower than slight increase of 0.1dB in the example gets even less.
At the limit, if the output impedance is zero the gain/loss as speaker impedance changes is 0dB.

if an impedance equalization network (such as zobel) is applied then Rload will remain same across the frequency band. in that case power available at speaker will remain same. is this what u meant by flat sounding frequency response?

[soundnovice]

Originally Posted by AllenB

The impedance is normally higher, but you can cross any way you like so it really depends. If you stick to the normal ways of crossing it should be reasonably predictable.

If you want to flatten this kind of peak you might add a tuned circuit (RCL) across the amp.

The impedance is higher, but then there is energy storage around the crossover region. I'm not sure I understand your question about attenuation. You can set the levels where you want them. Some crossovers have a flat response on axis but a reduced power response. For what it's worth, I use a crossover just below 1kHz.

The software is xoversim. The curves were measured and the crossover simmed onto them.

can you point me where i can find calculators or formulae to find out the values of RCL components.

[AndrewT]

Originally Posted by soundnovice

if an impedance equalization network (such as zobel) is applied then Rload will remain same across the frequency band. in that case power available at speaker will remain same. is this what u meant by flat sounding frequency response?

No,
if the speaker designer designed the driver and crossover combination to give an apparent flat sounding frequency response, without using a Zobel across one and/or other driver, when fed a flat electrical signal (i.e. flat with respect to voltage), then adding a Zobel does two things.
It reduces the efficiency of the speaker. Not at all popular nowadays. The Zobel also reduces the voltage seen by the driver at the Zobel's frequency. This creates a dip in the frequency response.