That would work, but unwinding an inductor isn't the easiest thing to do (the wire twists and goes all over the place), much less rewinding it tightly. Depending on the target inductance you may need a lot of wire. For example, if your target attenuation is 6db and your target filter inductance is 5mh, you will need to unwind and rewind a 20mh inductor, tapping into half the windings, without losing count or dropping the spool 
You probably refer to 25Hy it was my typo mistake, It should be 25mHy. But that is so obvious? Especially with the picture of used wire?..
But it is so obvious that in the speaker circuits we have almost exclusively [mHy] values?.
Hi I used 0.35mm "M"laminations from 2 Siemens chokes. To end up with 2 idetical as it possible cores I mixed, one by one, laminations from M type has fixed gap, this was 1mm. Besause it was too much of the gap for this purposes I put the lamination in opposite position to ech another to "shorten" the Gap.
Next I made test coil with few known values of windings.
AND measure the Inductances with 3 values of windings. I also measured with laminations in 1mm Gap value. (It was a bit slow because I have to tear apart 1 by one lamination from the core and put it back together later...)
With all these datas I calculated permeability of the core material and with that I was able to predict how it will be with any number of windings that can be packed in the given layer carrier.
The Permeability wes very high. That is why the value in [mHy] a bit high... And with when laminations stacked 1 oposite to next and so, the calculated gap value is about 1/3 of fixed 1mm so that is factor that brings additional increase of ind.
So I made some program in matcad to find the number of windings by the decibel order. And finally mount the coils.
I measured finished L with a friend with his HP "monster" 40Kg LCR meter
, (don't know the model number...).After that I measured each L with sweep signal and each tap to verify the att. levels...
This wss pretty much all...
.
Point is that You cant make good choke with common laminations of 250 to 450 core permeability. You need much higher permeability of the core laminations. AND mode important this imortant data has to be verified FIRST. For these purposes permeability of 800 is a abs min. and with a very small gap. You can use even a "no gap" because in this L att. (You don't have DC component to eventualay saturate the core...)
.
On the forum almost nobody measure the permeability, even dont talk about it. Members did not give the inductance of the primary windings and so... But everybody talking about the sound, impedance and other irrelevant things for good bandwidth. Using totaly inadequate laminations from very low permaebility... (That is why most of the tube designs have very low bass
And no "F" from physics...).
It this cae of Inductance attenuator, crucial things are
- high permeability of core (to use less windings BUT bigger diameter wire.)
- Low as possible Rdc of windings.
- sufficient number of windings for the desired level of induction with some average to max. Voltage level of AC signal (induction not inductance).
- And some special approach for taking out TAPs, because of the big dia wire. I used copper in thin lines to extract taps without disturbing order of windings to save space and make the layer tighter...
- My recommendation is to determine first what level of att You need and what step You need? I choose 2db step, and -8 or -10db, only 4-6 step taps. I used -2db step with 6 tsaps...
Last edited:
My advice is firs try to find appropriate lamination for cores. In the terms of permeability first, and second it shoud be a bit bigger.In such cases, phrase can be quoted.
Thank you, yes, it was useful, I'm a little slow in such things, but I began to understand better. Now I need to think more
Do I understand correctly... I can buy a transformer inductor of the value I need for the filter. Then disassemble it, count the number of turns, and wind two serial coils with the same wire?
I understand that it is easier to buy a ready attenuator. But first, they are quite expensive. Secondly, it is always interesting to DIY
To find have optimum space for larger diameter of wire AND sufficanr induction. It is very low probability for success if You randomly pick or buy. The permeability data is crucial for start of the design. And that has to be tested in praxixs...
The target permeability is classic 800 as for my opp some minimum, and preferably more. But that is not the common for the power trnasformers. The use 250-400 of lamination relative permeability... Some "C" cores has 800.
.
space is other issue and core should be bigger. Eventual "free" space in the coil is not disadvantage but opposite - all windings are closer to the center of the core... But not go to the extremes in that because of increasing the core You increase also the lenght of windings, and Rdc of the windings...
.
I foung very good M cores from old devices, that Siemens used for L. But it is hard to dismantle and put back...
OR some ready made "C" cores that are more easy to mount and exchange Gap.
.
But in all cases You should first check and be sure about the permeability data.
.
.
space is other issue and core should be bigger. Eventual "free" space in the coil is not disadvantage but opposite - all windings are closer to the center of the core... But not go to the extremes in that because of increasing the core You increase also the lenght of windings, and Rdc of the windings...
.
I foung very good M cores from old devices, that Siemens used for L. But it is hard to dismantle and put back...
OR some ready made "C" cores that are more easy to mount and exchange Gap.
.
But in all cases You should first check and be sure about the permeability data.
.
Last edited:
The number of turns directly depends on permeability of core. It is not fixed it is different for each type of core material. The permeability is higher the less turns neded for desired value of inductance. And Rdc is smaller too.
For simple -6db, that is half of the input voltage 1/2 of total turns will be at -6db. For the example...
BUT it is not so good to have just one tap even if You measured the right amount of attenuation?
I will recommend one tap around the desired tap. Because the things in praxis will be different and this is reactive componenet that is different from pasive non-reactive component att.
.
I will try to find the MathCad sheet to post. This is universal for any permeability, and custom step in -db.
But again, the permeability of core laminations is crucial.
You're right, it is more logical to have several taps. I just thought there was an easy way to make an attenuator myself to test how it sounds, but after all the comments, things don't look that simple.
My interest in the autoformer started because I accidentally discovered that using L-pad (Mundorf M-Resist Supreme resistors) on a compression driver changes the sound compared to a connection without L-pad (in this case I used EQ to lower the driver level).
I also read a lot of reviews that the autoformer sounds more lively and dynamic than l-pads, which is similar to my experience.
Here's an article from Paul Klipsch on the subject. Hopefully you can download the PDF.
https://community.klipsch.com/applications/core/interface/file/attachment.php?id=12394
https://community.klipsch.com/applications/core/interface/file/attachment.php?id=12394
Last edited by a moderator:
Before You start to search fr proper transformer core for att application, maybe to try Bridged T network. This is also resistor network atenuation, but for my oppinion giving much better results in sound against L-pad. It has 4 transistors in the network, and keeping the constant impedance...
https://www.everythingrf.com/rf-calculators/bridged-tee-attenuator-calculator
or google
bridged t network online calculator
(Qulaity and Type of the restors are factor for the quality of sound. Does not have to be super expensive - opposite try with MOX power res.)
Thanks for posting
In this document L attenuator is used without typical R that bypassing highest tap to gnd?
I measured in my example With / without this R, and without R it was not good at all
Impedance curve was disturbedand sonically was not good too...
Perhaps this crude LTSpice model will help you understand the differences between an autoformer and an lpad as far as the resulting voltage transfer function. I'm using a model of a Klipsch T2A autoformer and using it's -3.35db tap.
The middle circuit is a typical Klipsch high pass circuit to the squawker. It's the green plot. It's a little hard to see, but it starts around -44db.
The top circuit is an lpad circuit set to -3.35db. Because the T2A increases the load on the 13uf capacitor (14.6*2.162, or 31.57), I need to use a 28.114uf capacitor in this circuit. It's the top red plot. However, the 21.1mh inductance between tap00 and tap04 is missing. In order to get the same plot (the lower red plot really close to the green plot), I needed to add a 21.1mh inductor across the driver.
The bottom circuit incorporates a swamping resistor. I chose a value that, in parallel with the reflected impedance of the autoformer (31.57 ohms), brings the load on C3 back to 14.6 ohms. Notice the blue plot, as it now appears as if a parallel 44.8mh inductor is across the driver. 44.8mh is the inductance between Tap00 and Tap05.
Mike
The middle circuit is a typical Klipsch high pass circuit to the squawker. It's the green plot. It's a little hard to see, but it starts around -44db.
The top circuit is an lpad circuit set to -3.35db. Because the T2A increases the load on the 13uf capacitor (14.6*2.162, or 31.57), I need to use a 28.114uf capacitor in this circuit. It's the top red plot. However, the 21.1mh inductance between tap00 and tap04 is missing. In order to get the same plot (the lower red plot really close to the green plot), I needed to add a 21.1mh inductor across the driver.
The bottom circuit incorporates a swamping resistor. I chose a value that, in parallel with the reflected impedance of the autoformer (31.57 ohms), brings the load on C3 back to 14.6 ohms. Notice the blue plot, as it now appears as if a parallel 44.8mh inductor is across the driver. 44.8mh is the inductance between Tap00 and Tap05.
Mike