Jump to content
The Klipsch Audio Community

Al Klappenberger

Regulars
  • Content Count

    3918
  • Joined

  • Last visited

Everything posted by Al Klappenberger

  1. I am getting tired of this "just another way to do it" crap! Putting the transformer at the source end of the filter is just as much "another way" as the method Henry Ford used to start the engine on his model "T". It had a crank on the front. You would slowly turn the engine around with the crank until you felt the compression stroke begin. You would then yank the handle quickly to start the engine. My father showed me how it was done on the old model "A" he had when I was a kid. The engine would start every time. He also told me about his "Model T wrist" that was still hurting him from the day years before when the engine on his "T" kicked back and nearly broke it! Now days, we turn the key and an electric motor starts the engine for us. I suppose the electric starter is just another way to do it, just as good as the crank! BUNK! Putting the transformer at the wrong end is just an OUTMODED way to do it to a few people have remained invested in! Then there is Ahoooga horns. Showld we keep using those. After all, it's just another way to make a big noise! Quit listening to people who are invested in old outmoded ideas and keep promoting them to make a buck! NOTE: I have received a "warning" by the moderators... I am out of here. I DON'T NEED THIS FORM TO SEEL MY STUFF! The blind can continue leanding the blind!
  2. Mark, Actually you are correct. I usually don't talk about this becasue it's just more of the same. Remember that the K33 is actually 6 Ohms in sereis with 1 mHy voice coil inductance? Well.. The low crossover (woofer to squawker) is at 6 Ohms. The other networks I offer are 8 Ohms. By the parallel resistance equation Rt = 1 / (1/r1 + 1/r2)) so, 1/(1/24 + 1/8) = 6 Ohms. This matches 8 Ohms to 6 Ohms. The other reisitor is 24 Ohms! AND, I don't want to hear more crap from anybody about reduced efficency! It is NOT going to run your electric bill up! The 24 Ohms is a 5W resistor when a 1W would do. Same with the 10 Ohm across the transformer. I use 10W where a 2 W would do nicely. Have you ever heard of any of my networks getting hot! NO, and you won't! Al K.
  3. I thought this stupidity was done with! This is an example of somebody talking about things he clearly doen't understand! I thought I tought him better, I suppose not! The resistor is to absorb all the possible impedances created by moving the transfomrer taps to whatever filter it terminates! Infinity Ohms in parallel with 10 Ohms is 10 Ohms. This means even an 8 Ohm filter will see no more than 10 Ohms even with the driver completely disconnected. The is only a 1.25:1 missmatch! Putting the transformer at the source end only continues to be done becasue PWK did that! He did it becasue the 13 uF of the AA (for example) would be aout 50 uF if it was in a 50 Ohm filter! In his day a 50 uF cap would be unreasonable. THE ONLY REASON TO KEEP PUTTING IT AT THE SOURCE END THESE DAYS IS TO JUSTIFY BUILDING OUTMODED DESIGNS! Dean, it's time for you to stick to your day job! Al K.
  4. For anyone who hasn't followed this very closely, this is Belle. It's the little figure climbing the huge mountain formed by the impedance cure of John's transformer in one of his plots. Al K.
  5. I said it wasn't worth the trouble to calculate the time domain response of your "thing" but I changed my mind. It was fun! Here's what a 2KHz square wave would look like out of your midrange channel. Al K.
  6. WOW.. This plot shows very cleary that my Universal beats the tar out you thing! MAN! This is the ultimate insult to the Klipsch community. Do you really think the people interested in this thread can't read a plot! Speaking of FFT.. I could always use the foruier analysis section of PCFILT to show the impulse response of both networks (in the time domain you mentioned earlier), but it's not worth the trouble! You just keep embarrassing yourself to anyone who knows anyting! GO AWAY! Al K
  7. Your next plot: Here you have two plots on top each other. The difference is simply that you have a different value inductor in series with the K33 woofer than I do. That's why the circles move apart in the vertical reactance plane! BIG DEAL! The reason the "Universal" is universal is because it has a 1st order filter in the woofer channel. Just like yours. 1st order filters are so blasted sloppy that it doesn't matter what inductor you use! It will work equally poorly with any woofer. All a 1st order filter does is make two drivers play nice together! Take the inductor out completely like PWK did with the AB network for the Belle Klipsch and you will be left with the K33 alone. If you do that the circle will shift the other direction. This is yet more of your techo double-talk, or more accurately, double vision! Al K
  8. The next plot. The polar one: Now this is very clever.. Just stick up a plot showing the same thing as before again. The travels of Belle around in a circle. It's just REPITITION! We know what the complex impedance of the K33 looks like. I WENT THROUGH THAT EARLIER! Another trick associated with techno double-talk. Say the same thing over and over using different fancy words, or plots! We'll dazzle them with a round one this time! Al K.
  9. @@John Warren, @@John Warren, 1. The NRE-net has a very large reactance associated with the autotransformer, horn reflections, etc. The reactance is highlighted by our little mountain climber (let’s call her Belle). * This is because your transformer is in the wrong place! 2. Below about 500Hz, the two plots meld together as the K-horn Z dominates at lower frequencies. * It dominates because the crossover directs the energy to the woofer. This is why you see the resonance of the K33 at about 36 Hz. The impedance plots diverge because the amplifier sees you poor midrange filter above the 400 Hz crossover. There is NO GOOD REASON FOR THAT! 3. The BULK of amplifier power is delivered to the Klipschorn bass unit so amp loading, if altered in anyway why the speaker loading, is going to be the result of this reactance. * Double-talk! It's the inductive reactance of the K33 woofer. Same with the phase plots. It's the K33 your looking at! None of this is relevant to the network. If you had bothered to look at my earlier posts illustrating the effects of the K33 on the lowpass filter of the ES300 network you MIGHT understand that I explained this in detail! 4. Now, the midrange on the ALK is sort-of constant Z but it's over a region where the amp really doesn't care. * And just what range is it that the amp doesn't care about? A good amp doesn't care over any frequency range. This is the simply an excuse to do a crappy design. A little SET amp DOES care though. It can't handle extreme peaks like yours shows, at ANY frequency. It actually louses up the frequency response as the little amp unloads! There is no good reason to subject any amp to a 23 Ohm peak in the middle of the midrange! ------------------- * This entire set of plots simply illustrates a poor design. YOURS! DUMB!
  10. I have noted that in may cases JW will resort to Techno double-talk to cover his butt when he runs out of serious points. Now that the obscene items have been taken down by the moderators he has nothing else left! These latest plots about my Universal are such a total fantasy that I don't know where to begin! It's going to take me a while to travel to never-never land to try to understand how some of that crap could possibly have been fabricated, but I will! WOW! Al K.
  11. Warren, How about you posting the schematic and parts values on your network so I can do a little "simulation" on it. Everybody has had the design and values on my Universal for years. Including "group delay", as if it mattered! You might start by telling us what it actually is! All the "simulations" I did on yours was by assuming you got the parts values correct for an ordinary 3rd order Butterworth diplexer. That might not have been a valid assumption. BTW: Your measurements are all screwed up! Al K.
  12. It was only funny to me! To the general public is was juvenile and obscne! AL K.
  13. Yep, that's why I got banned! I will be happy to be banned again. Then I will be needing another sock puppet. Are you up for the job again Dean? You did a real good job on the Eliptrac HF thread a while ago! There was always a need on this forum for somebody to identify the snake-oil. It always seems to get interpreted as a personal attack, no matter how carefully the critique is crafted. I might as well go for it! An another thing, I don't need this forum to sell my stuff! I got all the work I can handle and then some. That's why I foolishly let you build my Universal! NEVER AGAIN! Al K. Edit: I forgot to mention: Dean, you shouldn't feal slighted. There's lots if insults scattered around this thread for you too!
  14. I agree, I think Lee and I have accomplished what we intended to do, debunk some snake-oil. Once this thread has run out of gass I plan to retire back to oblivion where I came from. It's entirely up to how much more techno double-talk gets put up. There is simply nobody else around here to do it. BTW: I sure wish I knew how to spell! Al K.
  15. Just for kicks, here's the ES300 lowpass as measured using the other simpler method. It's real easy to do, you just ...... On second thought, I better let a certain Piled High and Deep engineer reverse-engineer the procedure! Al K.
  16. Now I can get to the point. The microphone mounted under the K33 woofer box is used to measure the actual performance of a network woofer filter by subtracting the response of the woofer itself from the response with the network filter in the signal path. I have found an easier way to measure the performance of the lowpass directly without the microphone though. I do that with every network I build. This audio difference method proves the other simpler method is valid. Raw data from the analyzer can be downloaded through the HPIB buss just as its plots. It simply requires different software. The plots are in HP graphics language (HPGL). The raw response data is in real and imaginary pairs for each for 800 frequency points. Using this raw data allowed me to write software that allows me to calculate anything I want from the data and convert it to HPGL format. The plots attached show the response plots for the woofer box with and withOUT the ES300 lowpass in the signal path. The ES300 is simply the latest network I built. It's a one-off I don't expect to ever build again. It was available to illustrate the process. The bottom plot is the difference between the two plots from the microphone. It is the response of the lowpass filter using a real K33 as a load. The top plot of the attachment is the PCFILT simulation of the ES300 lowpass using impedance data downloaded from the K33 box via the HPIB buss. It's close enough! I might add here that I have done a lot of experimenting trying to find a way to smooth out this rough response. I can't, not even with the classic R-C Zobel network! It's only +-3 dB or so which is why it is really not significant. Every network lowpass I have analysed, both mine and those by Klipsch, have the very same problem. We are STUCK WITH IT!
  17. I don't know what book he found this information in since he had no equipment, other then a soldering iron and an LCR meter, when he was picking my brain. Anyhow, here is how you might measure it and IT is what casues the woofer filter to be rough, NOT the filter itself. The woofer under test is a REAL Klipsch K33 driver from the 1970s era. It's one like in my Belle Klipsch. It's mounted in a particle board box having roughly the volume of a Klipschorn back chamber. The slot and the blocks were adjusted to nearly match the complex impedance measured on one of my Belles earlier. The microphone is a Bruel & Kjaer 4134 pressure microphone designated for 90 degree incidence. It's mounted under the box near the slot for measuring the actual performance of a crossover network by a mathematical subtraction method I'll explain later. Somebody here likes polar plots, so I will use the HP 3563 analyser Nyquist plot in addition to the rectangular R +-jX plots I usually use to display the K33 impedance. The Smith chart is not so good for showing the impedance of a stand alone entity as it relates its impedance to a system impedance, that is, the impedance of a filter designed for that system impedance. The plots were downloaded from the analyser [PLOT] function using my own software over the HPIB bus to a DOS computer. It was plotted with an HP 7470A plotter emulator to get it into graphic form. All the plots from the analyser need to be multiplied by 10 as my impedance bridge uses a 10 Ohm internal standard resistor. The results show how bad a load a real woofer driver is to a filter. It's impedance is all over the place! In order to correctly model this weird load, the PCFILT analysis section was expanded to read impedance data form a file and use it as a parallel element at the termination. More about that later.
  18. "same effect as tone controls" :pwk_bs: :emotion-41: :emotion-45: You have GOT TO BE KIDING! Al K
  19. As somebody once said "who cares" ! And Lee: I like to put it this way: "education is no substituet for brains" BTW: I would hate to see this thread end. I haven't been in such a fun urinating contest since I was banned! Al K.
  20. Flatness 9 dB? Nope.. it's worse, it's more like 9.8 dB p-p including the crossover glitch. That's better than 13 dB with no network! I wonder what's happening below 1 KHz? Al K.
  21. John, You are actually quite correct. Microwave engineers do use the Smith chart rather then a Bode plot. Filter designers at any frequency actually use a measurement called "Return Loss". This represents the amount of energy bounced back to the source by an impedance mismatch. It relates directly to VSWR (Voltage Standing Waver Ratio) and mismatch error and combines both the resistive and reactive elements of a complex impedance in a single measurement. It is therefore actually displayed by network analysers and used by technicians to tune filters. The industry standard for most filters is 1.5:1 or 14 dB or more return loss. I adhere to this standard. At audio, the wavelength is so long that the Smith chard is not required. Around its outer circle would represent THOUSANDS of feet. Return loss however is still very convenient. The brown trace on the plots should be BELOW 14 dB. Here's a paste up of my ES400 + ES5800 set and your network. You shouldn't have to look very hard to see the same thing as shown on the Smith chart. The little circle in the middle of the Smith chart is the 14 dB return loss or 1.5:1 mismatch limit. I have added a 14 dB reference line to clarify the 14 dB return loss limit. One of these two networks is just a bit over the line! You are almost correct when you say nobody cares. Customers don't understand this stuff, but YOU SHOULD if you are going to try to design filters! Al K.
  22. Shucks folks! I just couldn't wait for you to ask! Here's the Smith chart and response of my ES400 + ES5800 set for the Klipchorn. Do you see anything outside the 1.5:1 VSWR limit? Nope.. neither do I. Al K.
  23. HUH?? Real filter desingers don't use "Bode plots". We use a "Smith chart". Acceptable impedance match must be inside the small dotted circle in the center. That represents 1.5:1 mismatch limit from the 6 Ohm standard forced on us by the K33 woofer impdeance. The plot is of your network. There's not much of it in there! Would you like to see mine? I would live to show it to you! BTW: The smith chart is used by microwave engineers becasue the outer circle represent 360 degrees down a transmission line. Any inner circle represents all the complex impedance possibilities that are a specific stading wave ratio or impedance mismatch. The usual impedance used at microwave is 50 or 75 Ohms. I have it set to 6 Ohms here. Al K.
  24. I thought it looked like an old push-pin K55!
  25. A little transformer history: When my "Universal" was under development, I found I couldn't buy a Klipsch T2A transformer from the original source, Universal Transformer of Texas, because it was proprietary to Klipsch, so: I had Universal Transformer make a similar transformer that had an extra tap that was 15 dB attenuation and bifilar wound. At the same time I figured I would make it a bit larger. I really didn't need to. If I was doing it today, I wouldn't have. The T2 is quite large enough. It was designated the 3619. The tap marking pattern followed the T2A, 0 to 5. The extra tap could not be called 6 since that would imply it was near the top of the winding. It's near the bottom, so it became X. The extra tap allowed me to float the K55 between taps allowing intermediate attenuation settings. I let Universal Transformer to sell it to anybody. Later, Bob Crites had a transformer made with extra taps close to the top that would allow 1 dB steps. That was an improvement over the floating K55 scheme of the 3619. In keeping with the same tap numbering scheme, the new taps became X and Y, It didn't have the 15dB tap I called X on the 3619. As with the 3619, calling it tap 6 wouldn't make sense. This became the 3636. It is proprietary to Bob. Only he can buy it. Not long after, I realized that all the letters and numbers was confusing things, so I had Universal transformer add another tap to yield and 18 dB option and actually mark all of them in dB instead of cryptic numbers and letters. This became the 3619-ET that I use now. Only I can buy it. Dean decided that bigger is always better, so he had Universal Transformer simply make an even bigger transformer and keep the new tap marking scheme of the 3619-ET, Of course there's the wax dip. That procedure was replaced long ago by the sealant normally used by everybody else today. $10 per transformer is no big deal. Just pass it on to your customers. Clever! I might add that ALL of these transformers were actually designed by Mark Cooper of Universal Transformer. Al K.
×
×
  • Create New...