Improving the AL network (using the A55G/CT120)

[jwgorman]

After speaking with a certain knowledgeable person on this board, part of the reason for the AL's reputation as a tweeter killer is the way it was constructed. Originally I was going to retain my AL networks in their unmolested state and build new networks, playing with AA and some of ALK's ideas on improving parts of the classic networks, but I really saw no reason to keep it.

I didn't think the sound of the lascalas was horrible with the AL in place, in fact, I sort of liked the recessed mids at first.  As others have mentioned, with -6db shaved off the squawkers you can really crank it up and the speaker sounds fairly balanced.

After installing the CT120 tweets and the A55G mids, and Crites' 3636 autoformers, I decided to experiment with the circuit. I reused the 2.5mH chokes and the terminal strips and played with different values of caps/coils/mid attenuation. The 2.5mH in parallel with the mid is really not a deal breaker either way. It was amazing that the less attenuation I had on the A55Gs (up to -3db) the better! The woofer circuit is, well, I think you could play with the values quite a bit on that. The tweeter filter was very surprising. I could hear a significant difference with inductor values and the cap on the other side of the inductor! I'm pretty close to ALK's values if memory serves and I'm still experimenting. 

I'm not a filter designer by any stretch of the keyboard. I'm just a guy playing with lascalas and caps and coils and an autoformer that allows -1db steps in attenuation.

So this is not gospel, but I will say, with those Crites drivers installed and this network (oh and my EV pro-sub blended in at a reasonable volume) I'm hearing resolution I'd never heard before and the most solid imaging I've heard a pair of klipsch (of mine) make. 

 

[wvu80]

Bob Crites recommends replacing, not rebuilding the AL.

 

Quote

 

Here is a picture of our Type AL-3 crossover for the LaScala. We recommend that the Type AL used in LaScala speakers be replaced with our Type AL-3 or Type AA or Type A/4500

 

http://www.critesspeakers.com/crossovers.html

 

 

 

I know you're having fun messing with the AL XO, and I'm not saying Crites is a better crossover designer than you are, but you have a ton of money in that A55G and the CT-120's.  I'm not sure you are maximizing their potential.  

 

Despite all the work you've done, I wonder if you could do better still with the sound by following Crite's recommendations.

[jwgorman]

Ok, I'll say it, Bob has forgotten more about these networks than I know! I built the AL-3 network, and the A/AA. Actually Bob suggested I take the squawker up to -3db with the A55G and I was skeptical, but he was very right about that. (AL-3 is -4db on the squawker)  and I consider this network a AA-sort-of, with ALK's tweeter filer thrown in, sort-of. I'm not done yet with the experimentation. I'm sure there is some tweeking to be done. 

[Wirrunna]

jwgorman, why have you put the 3636 autotransformer before the inductor, and no swamping resistor? As I understand it, the swamping resistor (while generating pages of arguments on the forum) allows you to change attenuation levels without having to re-calculate other component values.

I suggest that you put the 3636 after the 2.5mH inductor and (unless you are driving this with a quarter flea power single ended amp on a 5 volt power supply) add a 10 ohm 10 watt swamping resistor across the autotransformer. That way you can swap attenuation levels without having to swap caps to keep the same frequency of the cross-over

 

I have been building and experimenting with a few cross-overs for my K Horns lately and  tought myself how to use LT Spice so I could model the expected output before touching the soldering iron. If you have the autotransformer and swamping resistor in circuit you can easily model the cross-over in LT Spice leaving the autotransformer out as it does not affect the frequency response, only the output level.

 

LT Spice is a free download and there are plenty of Utube instructions on how to use it e.g https://www.youtube.com/watch?v=GmzfJa2GS7c  . I had this one bookmarked as I think it was where I started. Attached is a quick model of your handwritten cross-over.

 

For interest, also is the LT-Spice predicted output of my most recent build and the measured graph at the squaker and tweeter speaker terminals using REW.

 

AL3 mod.asc

AL3 mod.plt

ES400 + ES4500 as built with AK3 parts.plt

Clipboard01.pdf

[jwgorman]

"jwgorman, why have you put the 3636 autotransformer before the inductor, and no swamping resistor?" The AL network is set up like that, as are the AL2 and AL3.   

 

BTW all bets are off if you are using drivers other than the crites replacements as I don't know what the K77/K55s sound like with this xover. 

 

Thanks for the link on the LT spice. I'll have to play with that. Looks very interesting!

[Wirrunna]

Oops, forgot the link to the LT Spice download - http://www.linear.com/designtools/software/

At my early stage of fiddling with LT Spice I can't draw auto transformers, so after reading ALK's treatise on swamping resistors  http://www.alkeng.com/klipsch I just leave the autotransformer and swamping resistor out of the model.

                      

[Schu]

You cant just add a resistor on the posiive of the squaker leg of the AA circuit?

[Wirrunna]

1 hour ago, Schu said:

You cant just add a resistor on the posiive of the squaker leg of the AA circuit?

You are absolutely right. The AA uses a fixed turns ratio auto transformer and impedance and capacitance have been calculated for that fixed ratio to achieve the wanted cross-over frequency.

If you replace that fixed turns ratio auto transformer with a multi tap auto transformer, depending on which tap you use, you will alter the turns ratio and therefore should re-calculate the capacitance to maintain the same cross-over frequency.

An alternative is to move the auto-transformer and add a swamping resistor across the auto-transformer which will add some attenuation and also virtually eliminate the effect of using different turns ratios (taps). ALK explained it in his paper on swamping resistors.

As jwgorman is having a play around with an AA I was suggesting that he try using  a swamping resistor and model the effects of different capacitors using LT Spice.