Different Posted November 22, 2003 Share Posted November 22, 2003 Many members have commented that the ALK crossover has more extended highs than the AA crossover network. One reason this is true is because the series resistance to the tweeter is much lower with high quality polypropylene capacitors than with the stock motor run capacitors. An easy way to have similar results is to add a single .01uf bypass capacitor from the main + input to the + input of the tweeter. (Between barrier strip #2 and Barrier strip #8 of the crossover. This bypasses (only at the highest frequencies) all three capacitors (13uf, 2uf, and 2uf) with a single low resistance capacitor. You can even try this mod with an alligator clipped .01uf capacitor and it does not increase the chance of a blown tweeter Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 22, 2003 Share Posted November 22, 2003 Different, What you have done here is more interesting that you might think. In filter jargon, it's called at "triplet". What it actually is, is a way to generate a stopband "notch" like the "elliptic filter" of later Klipsch networks withOUT an additonal inductor. I did some computer analysis of it and found that a .01 uF cap generates a notch at about 650 Hz but has neglegable effect on the tweeter filter passband. If you increse the value to 1.0 uF. the notch moves up to a very usefull 3600 Hz and actually smooths out the rough passband of the AA and sharpens the filter skirt. Without the .01 uFd cap or with it, you have a 2.8 dB passband dip. With 1.0 uF the dip drops to only 1.5 dB! I suggest you try changing the .01 uF to 1.0 uF. Al K. Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 23, 2003 Share Posted November 23, 2003 Different, I'm impressed! Here's a computer simulation of what happens to the AA tweeter filter with a 0.68 uF cap connected from one of the 2 uFd caps to the other, bridging the two. .68 uF looks like the optimum value. Connecting it in or out with clip leads will let you hear the difference between an extreme slope tweeter filter and a 3rd order all-pole filter. THIS IS WORTH SERIOUS CONSIDERATION! If anybody get interested in this I will pull out my analyzer and do actual measurement on a real AA network. Al K. Quote Link to comment Share on other sites More sharing options...
Marvel Posted November 23, 2003 Share Posted November 23, 2003 Al, I know a lot of folks here seem to scoff at Wayne Parham's ideas (Pi Speakers), but I have some text from his forum where he talks of putting a .47uF cap in parallel with a 25ohm resisitor, feeding the horn driver in a two way. The resistor pads the signal down and the cap allows the top octave to come back up, flattening out the normal response of many compression drivers. He's got the formulae for the value of the resistor, and the cap can be varied as well. This is pretty much the same thing by adding this cap in parallel with the rest of the tweeter circuit isn't it? Marvel Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 23, 2003 Share Posted November 23, 2003 Marvel, NO! What "Different" proposed he thought was similar, but it's true effect is different and quite interesting. The RC network you speak of is a simple and quite conventional equalizer. It's the sort of thing used in constant directivity horns. The single cap "Different" proposes creates a notch in the stopband that sharpens the normal 18 dB / Octave (more actually in the AA) to a response the comes to a peak (notch), then back to an "arc top", then down again. It also flattens out the dip in the passband. The real reason is that the 3 caps are in a "delta" configuration has a precise "star" configuration that puts a seris cap into the inductor lead forming a shunt trap. The .68 uF cap I propose is like adding an 8 uF cap in sereis with the inductor! If your interested, I can provide equations to convert for "delta" to "star" equivalent. It's a tool used extensively in filter design work. Al K. Quote Link to comment Share on other sites More sharing options...
djk Posted November 23, 2003 Share Posted November 23, 2003 KEF uses this in several of their speakers. There is an article in Speaker Builder magazine a few years back that covers this. Formulas were presented to calculate the value of the cap, and a series resistor for same, based on the resonant frequency and Q at resonance of the tweeter. You can also model it in CircuitMaker (done), and probably Electronics Workbench too (done something similar). Quote Link to comment Share on other sites More sharing options...
3dzapper Posted November 23, 2003 Share Posted November 23, 2003 Al, I think everyone is interested in this. For me reading your posts is part of my continuing education. I just got another pair of AAs so that I can keep my originals as built for posterity and have a pair to experiment with. Plus the ALK parts I got from Shawn which I'll guild as a winter project. Rick Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 23, 2003 Share Posted November 23, 2003 Rick, Ok.. I can't do it until late tomorrow, but I'll work up the simple delta to star formula to predict where the notch will fall and post some curves made on a real AA with various value caps. All I have done so far is computer simulations. I expect to find that the passband flatening is just a coincidence in this particular filter, but I know the stopband "zero" is real. Al K. Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 24, 2003 Share Posted November 24, 2003 Guys, Here's the math on the delta to star (or pi to Tee as I like to call it) equivalent conversion. The example is using a 0.68 uF cap as the bridgeing capacitor (Cm). This configuration is sometimes referred to as a "coupled triplet" because it can be used to form a stopband notch in a microwave bandpass filter using coupled resonators such as 1/4 wavelength transmission lines. The capacitor (Ck) would be connect between alternate resonators. You can find an example of that on the PCFILT software side of my web site. It is virtually useless for new designs in lowpass and bandpass filters because it offers no advantage. It simply turns one inductor and 3 capacitors into one inductor and 3 capacitors! It would only be useful if the values of the parts were easier to get after the conversion. This would only be the case here if we wanted a HUGE capacitor for Cm because it will happen with a very small value of Ck. This only happens when the notch frequency is required to be extremely low. Since that is of no use, it isn't done. The .01 uF value suggested by "Different" would generate a Cm of 672 uFd. That would put the notch down at 392 Hz. That isn't good for much! I will do some measurements on a real AA network later today. Al K. NOTE: In the event of confusion about this delta to star conversion, both star and delta configurations exist at the same time. No actual conversion of parts is done. This is simply two ways of looking at the same "circuit". These two ways can simply be considered an analysis method to understand how the "notch" in the rejection slope happens. Quote Link to comment Share on other sites More sharing options...
3dzapper Posted November 24, 2003 Share Posted November 24, 2003 al, I did not have any .68mF caps to try so I used a pair of .47mFs. There is a noticable improvement in HF clearity unfortunatly that value also seemed to take an half octave off Al Hirts horn. Without the caps there is a blare evident in the horn that is gone with them. I must get some .68s to try. Would it work to parrallel a .47mF with a .2mF? Rick Quote Link to comment Share on other sites More sharing options...
dtximages Posted November 24, 2003 Share Posted November 24, 2003 You guys say this is worth it. Im not very technical with crossovers but i would love to try this. can anyone explain how to do this mod in a plain english way. like pictures or exaclty which cap to use and how much will it cost? apprec. Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 24, 2003 Share Posted November 24, 2003 rick and clspruiell, Yes, you can parallel .47 + .22 uF to make .69 uFd, but I am about to start testing and I may recommend different values. Hang in there a while! I will take a photo of how to connect the new cap and post it when I get done. Al K. Quote Link to comment Share on other sites More sharing options...
Rudy81 Posted November 24, 2003 Share Posted November 24, 2003 Sorry to jump in guys, but I have been following this with some interest. I have the AL-3 on some La Scalas and the only audible difference between the AL-3 and ALK's is the high frequency. If you have time, can you let me know if this mod would also work on an AL-3 network? Quote Link to comment Share on other sites More sharing options...
Frzninvt Posted November 24, 2003 Share Posted November 24, 2003 I have AA's in all 3 of my La Scala's and I am very interested in this. Al, once you get the value sorted out please show us in a photo how and where it needs to be soldered into place. Thanks Different for sharing this with us! Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 24, 2003 Share Posted November 24, 2003 Guys, I have attached a paste-up of 5 plots done with different value caps. The cutoff frequency of the tweeter filter does not move in any significant way, but the sharpness of the filter and passband sag improves dramatically as the value is increased: Cap value, loss at 8200 Hz --------------------- None, 5.3 dB .01 uFd, 5.2 .1 uFd, 4.9 .68 uFd, 3.0 1.0 iFd, 2.3 These test were done using the "corrected transfer function" mode of the analyzer that yields true insertion loss measurements by comparing a plot of input to output with a reference taken with the filter out of the circuit. The difference between the measured frequency of the "notch" and the calculated frequency is almost certainly the fact that the two 2.0 uf caps and .245 mHy inductor are not quite what they should be. This is to be expected especially considering the steel screw through the inductor, intact here, that I believe should be removed. It looks to me that the 1.0 uFd cap is the best value to use. The one tested was one of the Harmony 1.0 uFd by-pass caps from North Creek Music that I used in my "A" series replacement network. I believe this is a GOOD mod. Try it! Al K. Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 24, 2003 Share Posted November 24, 2003 Rudy, The LaScala AL-3 network ALREADY HAS this mod, but in a different and BETTER form. Sorry! I know that isn't what you wanted to hear! Al K. Quote Link to comment Share on other sites More sharing options...
Rudy81 Posted November 24, 2003 Share Posted November 24, 2003 WOW! That just shows how much better your ALK networks are beyond the others. Actually that is good news for you, since I eventually will be ordering the third ALK for that center channel. Thank you AL. Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 24, 2003 Share Posted November 24, 2003 Guys, The attached picture shows where to connect the capacitor. The Harmony 1.0 uFd cap is shown. Al K. Quote Link to comment Share on other sites More sharing options...
3dzapper Posted November 24, 2003 Share Posted November 24, 2003 Here they are ready to go. I'll post how this mod sounds on the Khorns later tonight. These are the AAs I got on ebay last week. The ones in the horns now have 2.2mF Musicaps and 12mF polys jumpered in. I wanted to duplicate your experiment so I am using the oils and left the screw in for now. Rick Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 24, 2003 Share Posted November 24, 2003 Rick, All I can say now is that you have exactly the right part in exactly the right place. I think it will do the same thing for AA networks built with better parts because the passband improvement the tests show is ONLY happening because it is a poorly designed filter. If you did this to the tweeter filter I build into my A replacement networks it would screw up the passband royally. I tried it in the computer to verify that, and it is true. It's just luck that the AA tweeter filters error happens to be in the right direction to get better, not worse, with the addition of the bridging cap. Al K. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.