Multi-tapped input/output autoformer

[Erik Mandaville]

Although many of you know this, I would like to mention that Bob C. shared some information with me concerning the autoformer used in his network projects.

It looks to be much more flexible than the more commonly found examples, and should do much of what a good variable L-pad can do. Subjectively, depending on one's preferences, tastes, etc., even better. The L-pad was a quick solution to taming what seemed to be an overly bright midrange horn, and it turned out to work quite well -- IMO.

In other words, the L-pad is not being described in any way as THE solution to midrange attenuation. I have used them numerous times in other loudspeakers and crossovers I have built, and they can work well. These are probably the best I have used, though, and they will withstand much higher RMS power than they will ever have to deal with.

Just like the PP/SET issue, L-pads might not be for everyone, but it might be worth a try. Some don't like to move a way from tradition simply because of the fact that it wouldn't be, well....traditional. And of course there is something to be said for the power of tradition. It's why I absolutely adore a pair of speakers that in a very real sense are antiques!

I think a multitapped autoformer could be interesting with flexibility and convenience of make-before-break rotary switching, which would provide the ability to audition different settings 'on they fly,' and that, in my opinion, is a more direct and effective way of making comparisons. One wouldn't need to rely so much on memory, but could hear the differences immediately, and just use the one that sounded best overall -- and solder the thing in place!

Erik

[3dzapper]

Erik, There is no reason that that couldn't be done on one of AL K's constant impedance networks. It would be much easier than switching the leads around. I have a couple of good NOS eight pole rotary switches laying around somewhere if you want to play.

Rick

[Tom Mobley]

Erik,

Are you talking about the UT 3619? That's the one Al K has been using all along. I think he had it designed and built by Universal Transformer for his ALK crossovers. It's very high quality compared to other stuff. Bob might be using something else entirely, but IIRC (gotcha again? If I Recall Correctly) Bob was stocking these.

I need one.

[BEC]

Tom,

The one I am using is built to replace all the Klipsch autotransformers. Not only the T2A but those that came along after that one. I had UT build one that is essentially the same core and coil as the 3619 but tapped differently to provide attenuation of -1 to -12 db in 1 db steps. AL has also worked out the numbers to use the new one in his networks.

Bob

[Deang]

http://forums.klipsch.com/idealbb/view.asp?topicID=57942

I still wonder about what John Warren said back when I was playing with this. In effect -- that dropping the output of the squawker 3db by going to the #3 tap doubles the reflected impedance the 13uF cap sees by a factor of two. This supposedly reduces the crossover point from 380Hz to 160Hz. John said if you double the impedance the cap sees, you need to "half" the 13uF cap to keep the crossover point the same.

Bob, I know you ran a plot on this that brought the math into question, but I did do more digging and it seemed to me that John was right. I don't think you can just arbitrarily move from tap to tap.

[Erik Mandaville]

This is why the simplicity of an L-pad is appealing to me. However, I think the mild changes in crossover point are going to be very subtle in terms of their influence on the what is actually heard.

With Bob's autoformer, one would probably need two switches -- one for the input taps, and one for the outputs. It shouldn't be too difficult to work out something a little more flexible than individual connections. From what I can tell -- all that is needed is a voltage divider -- consisting of either a wire wound element in the form of a variable L-pad, or transformer -- as a means of reducing the energy going to the midrange. What's nice is that the type of L-pad I used maintains an even impedance load for the amp regardless of the position of the wiper. I just can't tell a difference between the autoformer and L-pad, but the L-pad really is much easier to work with -- both physically and in terms of fine tuning to personal taste.

Rick: Thanks for that offer! I think I am good to go with this L-pad approach. It wasn't even something I intended to do. Craig was complaining of a very strong midrange, and I wanted to do try try the L-pad to see if what was wrong was in some way related to the settings on the autoformer in the AK-3 crossover (which did not have multiple taps for adjusting the output). It has worked so well, that I think I'm going to just stick with it. The reason I mentioned the switching arrangement for the autoformer had to do with thinking of something that would bring a little more flexibility or ease of use for those who prefer an the autoformer to variable or fixed resistance. If you've got switches that will work, why not give it a shot!?

Erik

[3dzapper]

I may put that on the "future projects list" Erik. Maybe with Bob's new Xformers if Al has the spec's worked out. That would make tweaking the sound a snap.

Rick

[ben.]

I sure wouldn't want to hear the K55/K400/401 try to operate much lower than it does now, other than out of morbid curiousity. Of course, one could wire in capacitors to deal with the various impedance changes from different AF taps, but that's getting pretty complex and costly.

[3dzapper]

Ben, That's the beauty of AL K's design, changing the taps does not have that effect on the xover frequency or impedence "seen" by the amp. It allows the decible output of the squaker to be matched to it's envrionment simply by changing a couple of wires and returning to the original settings all without soldering in addition to providing crystal clear sonics. A tweakers dream.

Rick

[mikebse2a3]

Dean said:

I still wonder about what John Warren said back when I was playing with this. In effect -- that dropping the output of the squawker 3db by going to the #3 tap doubles the reflected impedance the 13uF cap sees by a factor of two. This supposedly reduces the crossover point from 380Hz to 160Hz. John said if you double the impedance the cap sees, you need to "half" the 13uF cap to keep the crossover point the same.

Bob, I know you ran a plot on this that brought the math into question, but I did do more digging and it seemed to me that John was right. I don't think you can just arbitrarily move from tap to tap

-------------------------------------

Hey Eric

This is similar to what I mentioned in another post to you.

It seems to me since the 13mfd cap isn't seeing the autoformer/driver impedance which was higher than the 16 ohm L-pad your using your electrical crossover point has shifted upward.

The calculated impedance of the 13mfd cap at 400Hz = 30.612 ohm

so if my thinking is correct the autoformer/driver impedance at 400Hz is also approx. 30 ohms. thus the approx. 400Hz crossover point.

The calculated impedance of the 13mfd cap at 800Hz = 15.306 ohm

so again if I'm thinking correctly with the 16 ohm impedance created by the L-pad your electrical crossover point has shifted to approx. 800Hz.

One thought I have is if you used the L-pad after the autoformer your crossover point would stay pretty close to the design point but you would be able to reduce the output of the squawker.If you try this I would really be curious how this sounds to you. This might tell us if the sound improvement you've noticed is caused by the reduction in squawker level or as is probably the case also by the crossover shift and would it maybe sound better or worse if the electrical crossover point is maintained in your setup.

I realize that where talking about a 6db per octave crossover in Eric's case so the crossover isn't as quick at dropping the output of the squawker as a higher slope filter would be and thus is more forgiving of this crossover shift but I believe it has to be a part of the differance in the sound as well as the level adjustment of the L-pad.

Please don't take anything I'm saying as criticism of what your doing.I'm just really curious what all is really changing with this Mod. and I think it benefits everyone when we can understand why something has the effect it does which when it comes to things in audio "thats a real challenge".I realize you really like the changes you've made and thats what matters and I appreciate how you have been willing to share and discuss your ideas and experiances.

mike

[Erik Mandaville]

Mike:

No, I don't mind at all! I appreciate your explanation and interest, which is much more helpful and interesting than simply say something like "Bad idea, it won't work," "PWK didn't like L-pads", etc. If the the autoformer was designed to fulfill both the role of 'L' in a first order parallel network, as well as serve as a means to attenuate the output of the midrange, than I agree it would be a good idea -- and just as easy to do -- to try the L-pad AFTER the the coil (autoformer). In fact, I think it could be better. I was originally going to try the L-pad without the x-former in one network and with it (downstream) in another, but it sounded so good in the first one I tried (sans autoformer) that I just went ahead and did the same for the second.

My main interest was in the manner in which the L-pad vs the x-former would influence the overall sound, since changes in crossover frequency are not always easy to detect. The fact that the 'A' network is of a very shallow type certainly would make it a little more forgiving than I think would be the case with higher order designs (as you said!) However, there is also the possibility that one could,in the process, start sending upper midrange information to the high-pass branch, which can be kind of tough on tweeter voice coils.

That changing the output taps on the autoformer would slightly alter the crossover point makes sense, and would therefore mean that its function in the network extends beyond simply a means of dividing the voltage going to the mid-horn. What I have heard is sounding very good, but doesn not, I don't think (don't know for sure, though) have to do with the point at which the midrange passes the HF baton to the tweeter. It has to do with turning down a midrange driver that I was told by the trained ear of a professional musician was too high. I didn't realize how overbearing it really was until I had the opportunity to more easily and directly control it -- specifically, being able to listen to the difference in sound as the L-pad was rotated one way and then the other.

So! Let me go tack solder in the autoformer (factory setting) with the L-pad coming after it. There is actually another way of doing this that might be better still. Once we know the desired crossover point, the correct value inductor can be used in series with the capacitor (I have most often seen 'L' come before 'C' in the series connection between the input and driver. What I see happening, at least in the type 'A' is that there may be some interaction between the tweeter and midrange branches of the network, since they have certain components in common. (another story for another time, but something I have been thinking about...) What I think we will find is that only a very small amount of attenuation will be needed to achieve desired output levels when using the L-pad after the autoformer.

There is no question the L-pad makes adjustment much easier. I think I personally would rather use a high quality air core inductor than the autoformer, which can be done once we know the crossover frequency. From the values of the parts involved in relation to the impedance of the driver, I've got an idea of what that would probably be: if crossover is around 800Hz, we might be looking at 13 uf in series with what looks like about 2.7mH. I'm not using a calculator for this, just a very crude slide rule -- old fashioned! I need to look at this again when I'm not so tired. In any event, we will give your thought I try, Mike, and I sincerely appreciate your input with this! I think the L-pad is a good idea, and constructive and objective input from others helps!

Erik

edit: sheesh -- sorry this is so long.

edit#2 the cap value might actually be a little lower -- but I just can't stay up anymore -- my brain is shutting down!