Greg Oshiro

Folks-- It's been a while since I've visited the forum, and even longer since I've posted. This thread is quite spectacular. Regarding the idea of a curved flare for the side walls, I think there may be an easier idea from a woodworking standpoint. In the 1970's the professional sound reinforcement folks started to veer away from mathematically-defined horn flares (e.g., exponential, hypex, tractrix) to horn flares that strove for consistency of the coverage pattern vs. frequency. The earliest of these was the Electro-Voice HR horns, closely followed by the Altec Manta Ray horns. Don Keele was the brainchild behind the HR horns. Cliff Henriksen and Mark Ureda did the Manta Ray work. These "constant coverage" horns had flat surfaces composing the horns flares. The HR horns had radiused transitions between the flat surfaces and the Manta Ray horns had sharp transitions between the flat surfaces. It seems that many followers of the constant coverage trend followed the sharp transitions with no apparent problems. Many years ago I built a spreadsheet for designing horns following the ideas set forth by Keele, Henricksen and Ureda. I offer it here for those folks with more woodworking capabilities than me. There is also a sketch illustrating the dimensions with which the spreadsheet deals. Note that all three designs in the spreadsheet are essentially the same. We get to pick any two of: Design Beamwidth; Mouth Dimension; Mouth Frequency (F-Mouth). Given the AMT-1 design, the Design Beamwidth will be fixed at 90 degrees. Later on, Don Keele went to JBL and developed a set of curved wall constant coverage horns. The patent describes the math behind them, but does not provide a few critical numbers that would allow us to calculate the curves. MantaRayHorn_KlipschForum.xls MantaRayHornSketch.pdf

Roosie-- How is the gain measured? is this AC gain or DC gain? If you have access to an oscilloscope, does the output signal look like the input signal? A drawing of the measurement setup would be helpful.

Ol_mcdonald-- This is also a good resource: I would suggest using a Yamaha SP2060 or the Electro-voice DSP (model number escapes me at the moment). I've heard rumors that the mini-dsp is good also. Chris A has been the most prolific author of bi/tri-amping here and it is worth reading his posts. IIRC there are Klipsch factory settings for the Electro-voice DSP that will get you off on the correct foot. --Greg

I was very surprised to see recent activity on this thread. babadono, please keep in mind that my settings were based on measuring the frequency response and tweaking the crossover settings. Your settings will be different because of individual driver variation and the B&C drivers instead of stock Klipsch. I used a very old-fashioned system called TEF-25. There is free software called Room Eq Wizard (REW) that IIRC Chris A is familiar with that appears to be able to measure what needs to be measured. The difficult part is interpreting the measurements and deciding what crossover settings to use. Strangely enough, I spent the day in the woodshop making temporary baffles to try some BMS 4540 drivers on Electro-Voice HPT60 horns in place of my K77's. --Greg

Chris-- You've been busy. If you are in an experimental mood, try this: Measurement mic at the listening position. A large pile of fuzzy stuff (technical term) on the floor to kill the floor reflections from all 3 speakers to the measurement mic. Use REW impulse response to verify that the floor reflection is sufficiently attenuated. Use REW upsweeps to plot frequency response (magnitude *and* phase) with the longest time window you can to exclude significant room reflections. Measure one Jub first and set the window parameters. Use the same settings to measure all 3 speakers. Make the magnitude and phase responses from all three speakers match as best you can. This will be difficult at high frequencies. One inch of distance difference at ~13.5Khz is 360 degrees I expect additional input delay will be needed on the Belle to compensate for the (I assume) shorter distance to the Belle relative to the Jubs. Allpass filters may have to be added to the Jubs to make the phase response match that of the Belle around the Belle MF/HF crossover frequency. Adjust level of the Belle relative to Jubs to taste by ear. I've always wanted to try this, but I have no center channel. --Greg

I'd use Speakon NL4 series. 4 conductors so the HF and LF can be pre-wired. I would not use identical connectors for LF and HF. I've been known to mis-patch things when working too quickly. The consequences of the LF feed going to the HF drivers can be mighty bad.

Mr. Burwen is something of a legend in the analog electronics field. IIRC he had something to do with the founding of Analog Devices, a major player in the integrated circuit market.

Does it hum with the HDMI connected to the TV when the RF connector (F connector) is not connected and all other "normal" connections are present?

I remember being *taught* almost everything one needs to know to design crossover filters in my last 2 years of college (Electrical Engineering). I got out of school in 1977 and still haven't *learned* it all. With the modern measurement tools now available, most of the math isn't neccessary. All that's required is the qualitative knowledge of what the desired transfer function magnitude and phase should be and what various filter types do so we can select the right filter and adjust its parameters to get the desired result. Reading anything by Sigfried Linkwitz also helps...

Mustang-- Excellent! I think we all can appreciate a DIY kinda guy. If you're still interested in finding a metal can capacitor, I found this: http://www.technicalaudio.com/pdf/Electronics_Catalog_Extracts/Mallory_FP-WP_Series_capacitors_1985.pdf I think you want a Mallory FP288. If you can find one NOS, read up on "re-forming" electrolytics before installing. I also heard a rumor that CE Manufacturing is making brand spanking new metal can caps, primarily for restoring vintage guitar amplifiers.

Ryan-- I own an Adcom GTP-500. It has a high output level capability that can make it seem noisy when that output capability is not used. I suspect your pre-amp is similar. If your volume control is lower than 12 0'clock at your normal listening level, I suspect you are in this situation. You might try some line-level attenuators between the pre-amp and the amplifier. Maybe something like these: http://www.parts-express.com/pe/showdetl.cfm?partnumber=266-244 I have never used these particular things, but I believe others here have. If one pair helps your situation, a pair in cascade might be even better. If you're handy with a soldering iron and would like to build your own custom attenuator, let me know and I'll scrawl something for you. --Greg

Those measurements look very promising. Where is the mic located? How about a 20Hz to 2KHz sweep of just the LF including the crossover filter? dB-SPL and phase. How much amplifier power is available for the LF?

Duder-- Welcome to the wonderful world of tri-amping. Beware the pitfalls... The REW help file is very good. Reading it and googling the 'net will help a lot. Way back in 1978 Siegfried Linkwitz (as in Linkwitz-Riley crossover) triamped his home stereo. He now has a web site: http://linkwitzlab.com/ This describes the 1978 project: http://linkwitzlab.com/Removed%20pages/x-sb80-3wy.htm You have the advantage that the MiniDSP can create filter shapes by tapping numbers into the computer rather than rebuilding analog circuits. I have done both and DSP is easier. I know a bunch of folks who think analog is better, though. Analog is not practical for K-horns. I have measured K-horns and found that placing the mic 36" from the squawker motorboard and 51" above the floor (midway between the squawker and tweeter heights) works well if a large pile of acoustically absorptive stuff (sofa cushions and wool blankets in my case) is placed on the floor between the K-horn and the mic stand. This reduces the floor reflection. You may find that some absorption is also needed to reduce the ceiling reflection. Use the "Impulse Response" (IR) function in REW to measure the reflections. When measuring frequency response, also display the phase trace. I have had good luck with measuring in a manner that attempts to exclude the effects of the room and measure the frequency response of the speaker alone. Using these measurements, select EQ and crossover filters to smooth (and maybe flatten) the frequency response. Then do the rest by ear. Recommended approach: Remove any and all passive crossovers from the signal path. Amp outputs go directly to driver terminals. Use some sort of multipin connector that prevents patching the LF amp channel to anything other than the LF driver. Neutrik Speakon NL8 connectors are a low-cost, readily available method. Verify polarity of all drivers from the input of the miniDSP to the driver outputs using a phase popper. They should all be the same. If they aren't, fix it. Drivers have been known to leave factories with bad polarity. Set up the mic as described with no fuzzy stuff on the floor. Look at impulse response of the **squawker alone** with no filters in place **at very low level**, maybe 0.10 volts at the squawker driver.. Any peaks in the impulse response after the first arrival are reflections. The time difference between the first arrival and each reflection can be converted into the extra distance the sound had to travel (squawker > reflector > mic). Find out what those surfaces are and place absorption on said surfaces. Usually the first two surfaces are the floor and the ceiling. In my case, I just used absorption on the floor Place fuzzy stuff on floor. Look at impulse response. One of the reflection peaks should have changed. Window the IR beginnning just before the first arrival and ending just before the first "significant" reflection. The longer in time that the window is will make the measurement valid to a lower frequency. Try to reduce enough reflections to get a 10 mSec window. Look at the magnitude and *wrapped* phase of the frequency response. Insert an initial high-pass filter in the squawker signal path, say 400 Hz, 12 dB/octave Butterworth. EQ the squawker as flat as you can in the 800-4000 Hz range using all but 2 of the available EQ filters. I prefer to use cut only with filters wider than 1/6-octave. Your experience may lead you in a different direction. Without moving the microphone or changing any REW settings, do a similar thing to the **tweeter alone** with a 5000 Hz high pass filter. The phase trace will be a mess. Ignore it for the time being. Don't attempt to flatten the tweeter to 20 kHz. Maybe a 6 dB boost to extend the response a bit. There will be holes in the slope below the high pass filter. Boosting worked for me in this case. Once the squawker and tweeter are smoothed, increase the delay on the tweeter until the phase trace cleans up. I ended up with 1.6 mS. YMMV. Experiment with a squawker low-pass filter, tweeter level, tweeter delay and the tweeter high-pass filter to get these results: "flat" portion of the magnitude curves at the same level; magnitude curves crossing over at -6 dB; **the phase curves directly overlapping from one octave below the crossover to one octave above the crossover**. The phase curves are much more important than the magnitude curves at this point. One unused EQ from the squawker and one unused EQ from the tweeter may be needed in the crossover region to make this happen. I couldn't get all the way there, but it's a goal. If required, sacrifice the magnitude criteria to get a better phase overlap. This will take a long time. I do this sort of thing professionally and I spent about 10 hours in total on this (3 tries to get where the system is now). This is where you will learn what filters do to magnitude and phase. I had the benefit of an analog electrical engineering education. You get to learn by experience. Turn on both the squawker and tweeter. The magnitude curve should be smooth (but not neccessarily flat) at crossover and the phase trace should be smooth. If the magnitude curve is not flat through the crossover region, EQ it flat **using the EQ that affects the low, mid and high outputs**. I call this the "overall EQ". Boosting may be required. If there's a narrow, deep notch at crossover, something is wrong. Go back and check polarity and delay. Repeat step 9. This has taken longer than I thought. When you get this far, we'll work on the woofer/squawker crossover. It gets weird... This is what I ended up with after my first attempt: http://community.klipsch.com/forums/t/156476.aspx?PageIndex=1 My measurement system is not REW, so your data will look different. Since then, I've had the system in 3 different rooms, the LF has been tweaked countless times and various golden-eared folk have helped me tweak by ear.

moray-- Boy did you re-activate some old brain cells. I do remember that EV array. I'm thinking about a K-400/K-401/SpeakerLab H-350 squawker oriented with the long dimension vertical and a small CD horn tweeter above or below. My (unproven) theory is that the horizontal dispersion pattern will have less beaming at the high end of the squawker and with the horizontal dimensions of the squawker and tweeter similar, this will provide better contnuity of the horizontal polar pattern through the crossover range. I'm also hoping to find a tweeter horn/driver that can cross over lower in frequency than the K-77 and give more flexibility in crossover frequency selection. My woodworking skills and experimental time are limited, so it'll be a while.... Thanx for the interest.

Thank you to those that answered this post, and especially to speakerfritz, who was willing to part with a pair of SpeakerLab HD-350 at a reasonable price. Now it's time for me to design and build an idea that I've had for a while to make the polar behavior of the top hat more consistent throughout its' operating range. I'll take pix and acoustical measurements along the way and post the results. It may take a while, as this is a spare time project.