e-Lips

[richidoo]

Hi Bill, sorry I missed your post. I haven't worked on it for a few weeks.

I don't know what the final crossover will be. I am anticipating active now, I have some more experiments lined up for that. But I would like to try speaker level filters too if the time and funds permit. It's a simple design with lots of linear overlap 100-3000 so it should be fairly easy, maybe series 1st order with. But in all the digital active XOs I have tried the steeper filters usually sound cleaner. Which means neither of these is an ultra low distortion driver, but they both sound really good alone and together with steep filters. Maybe with the right Fc and good parts it will sound better than the digital simulations.

Today I finished the active analog crossover. It is the design of Rod Elliott. Phase coherent throughout the filter band. LR4 at 112Hz, using LM833 opamps, Dale resistors and Vishay 1813 caps. Twisted Pear bipolar series supply, and built right into the Buffalo32 DAC enclosure.

It was a little bright and congested on the feastrex' at first, but the caps and opamps are brand new, and I still have zipcord to the Feastrex. I will replace that with some decent SC tonite. Just in the time I started writing this the center image has improved drastically, dynamics starting to wake up too. Here's hoping.  The active analog filter is a lot less hassle than a PC xo.  The only price is ease of configurability, and minimum phase filters. The best passive speakers in the world are all minphase, it's not that big a sacrifice.  You only notice when compared to minphase in lower freqs. High freqs should be minphase anyway.

I listened to a single pair of woofers for a while this month. They are really incredible drivers. I was easily able to enjoy music listening on them, they play clean and detailed to 5k. They sound so good in the midrange I was thinking maybe the Feastrex can cut in at 2k instead of 100Hz. Expensive tweeters    There is a little sharpness at the very top  of the woofers as expected from paper cone, but the bass performance is amazing. No color, great texture and extension, even in OB. With baffle step corrected +3dB <380Hz it should be quite impressive. The drivers improved a lot with more time on them.

[Response Audio]

As I mentioned, I really like the visuals of these speakers. Maybe they need some satin black paint

As far as the crossover frequency, I would stay below the midrange frequencies so you have a single point source for the critical midrange frequencies. I wouldn't go higher than say 500Hz-600Hz. Preferably below 250Hz as you are planning. Nothing better than a single driver handling the midrange frequencies in an OB design IMHO. I am wondering if the narrower baffle is limiting the low end extension.

I really wish I had more time to play around with the boxes of drivers and other speaker "parts" that have collected at the shop over the past 15 years

[richidoo]

My wife suggested black, but she was thinking of how pretty my Gemme Katanas looked, which is buffed lacquer and far beyond my skill. She also suggested hammertone bronze     I keep picturing candy apple red. The intention is to make them from steel with veneered wood baffle top layer. But these will probably get the Glidden treatment before the steel comes.

Thanks for mentioning the baffle step, I think you're right about that. I had only considered the bass drivers for baffle step correction. But the Feastrex also needs it. That explains why it sounds thin in the mid bass now. In previous experiments I cured that by raising xo freq so that the bigger drivers handled the mid bass, but they're not as clean. Maybe now with better wire and break in, they can play higher cleaner. The step should be about 400Hz for the feastrex where baffle is thinner, and about 375 for the woofers.  I don't have any step correction yet. The woofers have plenty of extension with their high Q of 0.7 and low resonance. I'm only using one per side now and it keeps up with the Feastrex easily. Two will allow very low with some boost. All in good time.  Again, thanks for the tip about the baffle. I think the next crossover circuit will be a fair bit more complicated.

[Face]

You could cross at the baffle step to keep things simple.

[richidoo]

Thanks Face. I was just reading Martin King's paper on baffle step compensation
http://www.quarter-wave.com/General/BSC_Sizing.pdf
and that was one of his recommended solutions.

Using MK's formula for determining the step freq, I get 285Hz, which should work OK as a xo frequency too, as long as woofers can keep up with the feastrex that high.

John Murphy's formula concurs with 285Hz.
http://www.trueaudio.com/st_diff1.htm
I don't know why I was thinking 380Hz.

The only thing bad about these speakers is how they sound listening from the side. We listen to music while cooking and eating dinner, and the kitchen and DR are in the bass null of the dipole. So it sounds pretty threadbare. But sitting in front of them is so different.

[richidoo]

Today I upgraded the zipcord SCs going to the feastrex drivers to JPS, an awesome improvement needless to say. I still have a couple feet of zip in the dual woofer hookups, which will be replaced soon then all speaker wire will be JPS.

I padded down the input of the current source amp by 10dB so it is the right gain for my sources.  Thanks to Sol for the pad design. Input impedance dropped from 56kOhms down to 14.5kOhms, no biggie.

I listened to the 112Hz active analog crossover for a few days. It was pretty good, but with no good way to compensate for baffle step, the thin upper bass was starting to annoy me. I was also starting to think about other crossover freqs and slopes, so I thought I would stick the Behringer DCX2496 back in so I can experiment. I programmed it with 285Hz Linkwitz Riley 24dB crossover, like Face suggested, since it's easy to hide baffle step correction at the crossover freq when using a passive crossover. But in the Behringer I  made separate filters for baffle step compensation so I can adjust the crossover freq independent of the correction. I started with +3dB shelf with -6dB slope cornered at 285Hz for the woofers where the baffle is wider, and at 350Hz for the Feastrex to match its narrowest distance to the edge of the baffle. I eventually bumped the woofers up to 4dB and the feastrex to 3.5dB. Good balance now, no more mid bass hole.  Thanks Bill, for getting me thinking about baffle step more rationally.

At 285Hz crossover frequency I was hearing some chestiness in vocals, so I started lowering the xo freq. I reread Jon Valin's blog on stereophile about Lotus Group's Granada at CES 2009, he said they were using 200Hz LR4 crossovers. So I went down to that, nice improvement, still a little chesty. Down to 150, almost there. 100Hz, too thin. 125Hz is sounding pretty good, and much cleaner than 285. Maybe it will change a little when I calibrate the gain levels of the amps. I am tempted to wonder if the Feastrex is faster than the AEs at that low freq, but I still have zip cord paralleling the coils together and I'm still only using one woofer per side. Adding the second woofer will increase accuracy by reducing displacement at the same volume. Maybe that will match Feastrex speed up into the 800Hz range where the woofers are finally becoming inaudible when the crossover is set to 200Hz (but I doubt it.)  The other reason to have two woofers per side is to increase acoustic impedance to be able to play the very lowest notes, which will still require some boost to get down to 25Hz, due to cancellation effect of the open baffle. But the 2 drivers will be able to move enough air at 20Hz.The behringer is already down to only 30% processing power left but two more parametrics should be OK. The voice coils can only take 200W so we'll have to see how much it takes to make the low notes at volume. I don't need too much volume anyway.

One of the downers of the day is the Behringer is having dropouts. Not sure why. It was doing it last month too. It seems to get worse when the volume setting of the digital stream is louder, especially on compressed rock music which is floored all the time. It will work fine for 10 minutes, then start to hiccup again. When I hear it I usually pause the source then restart, sometimes it works, sometimes it hiccups right away again. Sometimes it screws up the filters or memory or something that changes the tone for a few seconds until it heals by itself. I suspect that it is the lousy electrical path that the spdif must travel inside the Behringer. mgalusha sent me a foot of coax cable and 75ohm terminating resistor a loooong time ago, to bypass the convoluted path and thin ribbon cables the signal takes inside the Behringer.  Hopefully that will help. It maybe that it wants AES/EBU 5V signal instead of SPDIF .5V, but I never heard others using Didden mod mention any dropouts, so hopefully it is something simple. I'll drop Jan Didden an email too, always good to have an excuse to email with him.  I also want to try swapping LM833 opamps for the LM4562 that came with output stage mod.

I increased the capacitor of the zobel filters on the Feastrex drivers. That will lower the driver impedance in the highest frequencies. With the current amp the impedance is critical for power output. The treble was still too hot for me before with 1.5uF, but 2uF sounds perfect, just a little laid back in the top.

At one point I felt that the top and bottom were not clicked together as one, so I moved the speakers back away from me by a foot, that seems to have helped a lot. Funny that my brain didn't like it at first, I was adjusted to the sound. After 10 minutes I liked it much better. Could be the driver delay was off, because the baffle angle is intended to align the woofers and mid when the speakers are 10 feet away. I think I had them too close. Hard to believe I can hear 1 mS delay at 125Hz though. More likely rear wave reflections were confusing the sound at the closer location. Whatever...

Altogether a constructive day, and I'm pleased with the way things sound now. But still more to go.
Rich

[richidoo]

I thought the dropouts might be the Sonos, which is fairly high jitter. I swapped in the CDP and it seemed to fix it, but then it started again. It is different behavior, now, longer period of clean, with shorter periods of worse dropouts. So there is a connection between source and dropouts, so it could be jitter, or memory buffer or something. Hopefully the wire will fix it.

I increased the slope to 48dB/octave and increased the crossover freq to 150Hz. A little fuller sounding. 200Hz still sounds too thick.

[Face]

Are you running any delay to compensate for the difference in acoustic centers?

[richidoo]

Not yet. The tilt of the baffle aligns them within .25" if I sit at 10' away. That way I can use a passive crossover later. But I can adjust the delay if I want to with the Behringer. I'll have to test the impulses to fine tune it. I need to see a FR soon, so the impulse will come with it. Thanks Face

[richidoo]

I just nabbed a Rane RPM 26z Multiprocessor on Ebay for a great price.     My kids will give it to me for Christmas and I will be more excited than them! 

http://www.rane.com/rpm26z.html

This has more processing power, higher bit depth processing, and better processing algorithms than the Behringer. I can control it with a PC app called DragNet via Ethernet so I can tweak it from the sweet spot and save 24 presets. It uses the same 8 channel volume control chip as the Didden DCX mod, and can be controlled by external linear pot. I believe this is the same processor engine used in the Granada, if not the same Rane model.

I have the schematic so we'll check it out for upgrade opportunities. It is full of JRC 4580 opamps in the signal path, easy enough to upgrade those, or bypass with transformers or caps.

[richidoo]

The Behringer DCX-2496 digital crossover has a known problem with the digital input receiver/SRC chip, Cirrus Logic CS8420. That's what's causing the dropouts. There are half dozen upgrade kits to fix it, but it'll cost about $120, giving a huge performance improvement along with the fix. My unit also needs a linear power supply upgrade, regulators and RCA output jacks to be ready for hifi prime time. It's money that I would never get back. I only have $500 in it so far, but at current prices it's $700. I think it will make a nice discounted launching point for a modder.  Car audio guys are using them too.

I'll check out the Rane first before I list the DCX for $350. PM if interested.

The Rane is shipped, arriving Saturday. Not sure I can wait til Christmas to start using it. 

[richidoo]

The DCX crossover worked flawlessly for 8 hours today, not one dropout. I had some fun tuning the crossover, and trying a bunch of different configurations. I even listened to Beethoven 7th all the way through.

2nd order at 300Hz is sounding good. A little extra body from the big woofers feels good. A mild presence notch at 3800, Q=2, -1.2dB takes away the wizzer zing from trumpets and violins. I'll probably see more once I measure it.

I also tried a 1st order cross at various freqs. I was surprised that it sounded pretty good. With the wiring upgrade and passive parts it could be cool.

After getting used to the 2nd order for a while I went back to the 4th order and even 8th order, they were just too tight. Even in digital realm I don't prefer steep min phase filters. Steep linear filters don't bother me.

A couple hours later it started dropping out again. Rebooting it didn't help. So I shut it off. Maybe it is a morning person crossover.

[Carlman]

Glad things are working (in the morning)... I also like the presentation of 2nd order crossovers.. Just more presence and coherence.  

Sounds like you're getting your sound to develop into something you like.. congrats.

[richidoo]

Thanks man!  You're always such a positive and encouraging influence, I appreciate it.

The Rane has arrived, looks to be in good shape. Mrs Claus says I can't play with it until Christmas, but she goes back to work tomorrow, so I'll test it to make sure it is works...     It's used merchandise, after all.

But from reading the manual it appears to be infinitely more powerful and flexible than the Behringer.

The nice thing about the 2nd order crossover is that it wouldn't be too expensive to make a nice passive crossover with it. Usually the passives will sound better than the electronic anyway. Even 1st order sounds OK, at freqs from 200-1kHz, but I haven't been able to fine tune that yet.

I've also been thinking that a first order passive cross, at very low freq would be able to let the Feastrex sound its best with minimal influence from a filter, but still protect against overexcursion in the deep bass - while the shallow falling 1st filter on the bass would still be in phase with Feastrex, but the bass inductor size padding of feastrex can create baffle compensation and bass boost at lowest freqs. I'd need a steep rumble filter on the line level too, but a preamp can drive that so no added opamps needed, and small cap size of line level allows using good stuff. Sensitivity would suffer by the amount of bass boost desired. Biamping would solve that, but you have the option to use single amp, and the load would be extremely easy, and tube amps can make plenty of voltage.

Anyway, can't wait for you guys to hear it. But not until I am 100% satisfied.   

[richidoo]

I did fire up the Rane yesterday to make sure it works, being used and all. I couldn't listen because it has euroblock connectors so I have to rig it for RCA before I can hear it. If it sounds good I'll modify it for RCA outputs.  The software is really cool. I made a simple configuration for the E-Lips with a software switch between digital and analog inputs, a 15 band parametric EQ on each channel, the crossover, the output volume control, and output connections. The basic configuration can't be changed live, but the settings of each part can be changed live and stored under 24 presets. I think crossover slope and freq is just a setting, which will allow quick A/B comparisons. But no X! 

Meantime, before Christmas when the Rane officially arrives I am still tinkering. The past few days I have been running the Feastrex driver solo, with no bass support. This has always been the most open and natural sound so far, especially with the 300Bs. The drivers' bass does roll off, but remains extremely clear. Even in OB, it is musical to listen to it by itself, especially with the current source amp which fills it out. So I had an idea to try my diy analog line filter on the bass drivers only, and let the Feastrex run free. It worked, and sounds surprisingly good. This is the high water, so far. I didn't try this before, because there is a lot of phase error, which I thought would be more audible. I can perceive a slight disconnect, but the bass is nicely supported now which is better.    The phase error is no different than what you'd have by putting a stand alone sub with any separate speaker without a high pass filter on the bottom.

Anyway, based on how nice it sounds now, I have big hopes for the Rane.