Autoformer for Klipsch crossover

[Fast996]

Hello Dave Slagle,

I plan to build a crossover for my pair of AK5 Klipschorns. These crossovers use a autoformer. The autoformers are available from Klipsch but I would like to optimize a autoformer design for the crossover. I would like to know how this could be done and what
sound benifits could be realized.

Paul Klipsch always said he preferred the DCR in the basement.

Here is a schematic of the crossover

Regards,
RogerD

[Fast996]

BTW I found this on this forum. It looks similar to a klipsch crossover.

and a custom autoformer?

[Fast996]

Would these speakerformers fit my application?

[dave slagle]

I am not sure exactly what is going on in that crossover. I would need to know the inductance values of the autoformer to get a better idea since it is unclear if the autoformer is part of the filter or a simple attenuator.

dave

[Fast996]

I am not sure exactly what is going on in that crossover. I would need to know the inductance values of the autoformer to get a better idea since it is unclear if the autoformer is part of the filter or a simple attenuator.

dave

Hello Dave,

I believe the T4A is -4db and it is not a filter,but let me verify that.

Roger

[dave slagle]

OK

If the autoformer is not part of the filter network and is purely intended as a level match device then it is serving another purpose. a 4dB cut is a 1.6x voltage ratio and a 2.56 impedance ratio. This will make the 5mHy choke that follows it appear as a 12,8mHy choke and the impedance of the driver will also appear at that value. It is actually a clever way to level match and make the crossover inductor a more reasonable part.

If you try to make the level adjustable you run into a problem so the speakerformers are not a good choice and a single value should be selected and dialed in. Adding adjustability to the autoformer makes the speaker and crossover impedance values change as you adjust them As you attenuate more the crossover frequency gets pushed lower due to the increasing reflected value of the inductor and the higher reflected load starts to make the filter underdamped. The attached picture shows the response of the filter with the autoformer swept from 0 to -8dB in 1dB steps.

This is not to say that the speaker formers will not work, the will work perfectly. It is just that the adjustability function will lead to odd results. 1 click either way from the 4dB will work mostly as expected but going further than that will materially change the crossover frequency and slope.

[Fast996]

Dave I asked over at the Klipsch forum and so far I have this.

"
If the inductance of the T4A, taps 0 - 5, is the same as the T2A, taps 0 - 5, then the inductance of the T4A, taps 0 - 3, should be 18.1mH (45.6mH / 2.512"

and it is 4db attenuation.

[Fast996]

OK

If the autoformer is not part of the filter network and is purely intended as a level match device then it is serving another purpose. a 4dB cut is a 1.6x voltage ratio and a 2.56 impedance ratio. This will make the 5mHy choke that follows it appear as a 12,8mHy choke and the impedance of the driver will also appear at that value. It is actually a clever way to level match and make the crossover inductor a more reasonable part.

If you try to make the level adjustable you run into a problem so the speakerformers are not a good choice and a single value should be selected and dialed in. Adding adjustability to the autoformer makes the speaker and crossover impedance values change as you adjust them As you attenuate more the crossover frequency gets pushed lower due to the increasing reflected value of the inductor and the higher reflected load starts to make the filter underdamped. The attached picture shows the response of the filter with the autoformer swept from 0 to -8dB in 1dB steps.

This is not to say that the speaker formers will not work, the will work perfectly. It is just that the adjustability function will lead to odd results. 1 click either way from the 4dB will work mostly as expected but going further than that will materially change the crossover frequency and slope.

Ok....let's move away from the speakerformer. Is there any reason or way to improve from the stock auto former? If I have you build autoformers?

[dave slagle]

well a nickel core goes a long way to getting better sound

dave

[Fast996]

well a nickel core goes a long way to getting better sound

dave

Ok Dave...I will email you when I'm ready.

Thanks,
Roger

[deafbykhorns]

Did this ever get built? I have the same autoformer and looking for an improvement
Daves AVC did wonders for my preamp

[famish99]

I've been working with Dave via email to adjust the airgap on the normal speakerformer to make it work for this application. I wonder if HyMu 80 would work for a run of these or would they saturate too quickly?

[dave slagle]

This comes down to the applied voltage and crossover frequency. 80% nickel has a saturation of 1/2 that of the supplied 49% which is the difference between 500Hz and 1000Hz or 2W and 8W.

dave

[famish99]

Ah ok, once I get the crossover put together, I’ll hook up a scope to the autoformer input and check for how much power I’m seeing across it.

[deafbykhorns]

Heres a table for a autoformer used on KLIPSCH type xover with more taps
Autoformer is at output so just a volume control, i think Dave could build something special!

[deafbykhorns]

This matches the 3636 transformer taps above, i believe those might be inductance values in first table.
I just rememberd that I did this for the 3636. It's computer generated data. The columbs will only line up if you use a fixed pitch font. Cut and past the the table to some text editor.
Al K
..................
5 Y X 4 3 2 1 0
5 0.00000 0.10880 0.20567 0.29205 0.50000 0.64519 0.74881 1.00000
Y 0.00000 0.09687 0.18325 0.39120 0.53639 0.64001 0.89120
X 0.00000 0.08638 0.29433 0.43952 0.54314 0.79433
4 0.00000 0.20795 0.35314 0.45676 0.70795
3 0.00000 0.14519 0.24881 0.50000
2 0.00000 0.10362 0.35481
1 0.00000 0.25119
0 0.00000
The actual turns ratios of the 3636 transformer for each tap are listed
in the far right column of the computer generated table. The turns ratio
(and therefore the voltage ratios) between any two taps with input between
0 and 5 are given in the table. These would be the ratios possible for use
with my networks.
Attenuation is of course: dB = 20 * log10(ratio)
Here's the same table expressed in dB:
5 Y X 4 3 2 1 0
5 -19.27 -13.74 -10.69 -6.02 -3.81 -2.51 0.00
Y -20.28 -14.74 -8.15 -5.41 -3.88 -1.00
X -21.27 -10.62 -7.14 -5.30 -2.00
4 -13.64 -9.04 -6.81 -3.00
3 -16.76 -12.08 -6.02
2 -19.69 -9.00
1 -12.00
0
Setting sorted by attenuation:
atten atten between
ratio dB taps
0.08638 -21.27 X 4
0.09687 -20.28 Y X
0.10362 -19.69 2 1
0.10880 -19.27 5 Y
0.14519 -16.76 3 2
0.18325 -14.74 Y 4
0.20567 -13.74 5 X
0.20795 -13.64 4 3
0.24881 -12.08 3 1 dB
0.25119 -12.00 1 0 -- 12
0.29205 -10.69 5 4
0.29433 -10.62 X 3 -- 10
0.35314 -9.04 4 2
0.35481 -9.00 2 0 -- 9
0.39120 -8.15 Y 3 -- 8
0.43952 -7.14 X 2 -- 7
0.45676 -6.81 4 1
0.50000 -6.02 3 0 -- 6
0.50000 -6.02 5 3
0.53639 -5.41 Y 2
0.54314 -5.30 X 1 -- 5
0.64001 -3.88 Y 1 -- 4
0.64519 -3.81 5 2
0.70795 -3.00 4 0
0.74881 -2.51 5 1
0.79433 -2.00 X 0
0.89120 -1.00 Y 0
1.00000 0.00 5 0