Kleos - Dipole Line Array

[Johnloudb]

Uh-oh - women and DIY speakers: Like Water for Chocolate. I too am distracted by a lady for some months now. She has agreed that I will be her first date after emerging from her "transition period" after divorce. To say she is fantastic would be an understatement: intelligent, funny, interesting and she is a triathelete (and built like a brick s__thouse). Don't see what she sees in me, frankly. Here's my surprise for our first date on the 23rd: http://greaturbanrace.com/event2013_austin.php
John boy, I DO understand your distraction :W

Best of luck on your date Hank ... great choice for a first date, looks like fun! For what it's worth I think she's made a great choice too!

[Johnloudb]

K, got both speakers done, as in finished, completed, .... So now on to the crossover! First off my dad's birthday is Monday so thinking of buying him LspCAD. Just one question, does it allow for active crossover modeling? I mean just to apply different filter types and slopes to the measured response of the drivers?

[wkhanna]

Got all the drivers installed, just need to wire them up now.

You go, John!

Those are realy looking good!

And all you guys, remember......do not 'Kiss & Tell'.
They (the Girls) hate that! ; )

[JonMarsh]

Well, yes and no- takes a little more work than that. you can work with several digital crossover types, classic sections, etc. Download the manual for a complete overview.

[Saurav]

Back when I had a DCX2496, I did all the modeling and optimization in LspCAD. Works great, but like Jon said, it's not going to figure it all out for you. You'll still have to get it started with something that's close to the topology you want to end up with.

[Johnloudb]

You go, John!

Those are realy looking good!

And all you guys, remember......do not 'Kiss & Tell'.
They (the Girls) hate that! ; )

Thank you Bill! Yep, they are all wired up.

Okay, I won't "Tell"

[Johnloudb]

Well, yes and no- takes a little more work than that. you can work with several digital crossover types, classic sections, etc. Download the manual for a complete overview.

Okay, Thanks! The transfer function module looks like it would do most of what I need to do. Still may just may wait and take some measurements with ARTA which I have right now.

My favorite line in the manual - "This module can not directly drive a passive component."

Back when I had a DCX2496, I did all the modeling and optimization in LspCAD. Works great, but like Jon said, it's not going to figure it all out for you. You'll still have to get it started with something that's close to the topology you want to end up with.

Thanks! I do have a DCX2496, and I saw the DCX2496 module in the manual. Is that what you used? I won't be using the DCX2496 with this though as I'm building my own crossover. But I could the DCX module to try some different filter types. I know pretty much what my crossover filter circuit will look like, from a schematic point of view.

[Saurav]

Yes, I used the DCX module, it knows about the limits of the various controls so it won't exceed them. It can be a good prototyping tool while you figure out what kinds of crossover topologies sound good. The only thing is, with the DCX you can 'cheat' a bit, and dial in arbitrary delays that don't otherwise affect the transfer function. So 'time-aligning' your drivers is relatively easy. Doing the same thing in the analog/passive domain is a black art as far as I can tell.

[Johnloudb]

Yes, I used the DCX module, it knows about the limits of the various controls so it won't exceed them. It can be a good prototyping tool while you figure out what kinds of crossover topologies sound good. The only thing is, with the DCX you can 'cheat' a bit, and dial in arbitrary delays that don't otherwise affect the transfer function. So 'time-aligning' your drivers is relatively easy. Doing the same thing in the analog/passive domain is a black art as far as I can tell.

Okay, thank you.

Well nothing happens without a sound card device for your computer. All the E-MU 0404 drivers I've tried for Windows crash my computer. It worked fine with windows XP, but with windows 7 no dice apparently. Any suggestions? I really don't want to spend more money for a sound card ... but am open to recomendations. Still got my iphone and may just stick with that for measurements.

[Hank]

Keep at it John. The Evil Twin usually has a sound card recommendation...

[Saurav]

I don't really have any good suggestions here. I've only used an M-Audio Transit (external USB sound card) on an old Windows XP laptop. Haven't tried it with Windows 7. E-MU makes good sound cards from what I've read. Do you have a Windows XP install disk lying around, and could you dual boot your computer? Even if that worked, it would be a pain to use.

FWIW, I've always taken measurements on my laptop using ARTA or HOLM Impulse, transferred them to the PC and loaded them into LspCAD for simulation work, and manually entered the settings into the DCX for listening. I've never used LspCAD to take measurements or drive the DCX because the PC is in a different room from the stereo. So your iPhone approach could work if you find a file format that works between them.

[Johnloudb]

I don't really have any good suggestions here. I've only used an M-Audio Transit (external USB sound card) on an old Windows XP laptop. Haven't tried it with Windows 7. E-MU makes good sound cards from what I've read. Do you have a Windows XP install disk lying around, and could you dual boot your computer? Even if that worked, it would be a pain to use.

FWIW, I've always taken measurements on my laptop using ARTA or HOLM Impulse, transferred them to the PC and loaded them into LspCAD for simulation work, and manually entered the settings into the DCX for listening. I've never used LspCAD to take measurements or drive the DCX because the PC is in a different room from the stereo. So your iPhone approach could work if you find a file format that works between them.

I recently looked at ARTA and couldn't see how to import files. But seems like I have exported files from it before, been awhile. .FRD rings a bell but not sure and my iphone can export files in .CSV, .MAT, and .TXT format.

[cjd]

I recently looked at ARTA and couldn't see how to import files. But seems like I have exported files from it before, been awhile. .FRD rings a bell but not sure and my iphone can export files in .CSV, .MAT, and .TXT format.

Not sure what you would import into ARTA as it is really a tool for acquiring data. You can export FRD and many other things.

TXT may be FRD, but you'd have to check it.

I also use an M-Audio external USB sound card (with phantom power - they no longer make that one, guess that part stuck with Avid and their new cards are different?) and have just about given up with internal cards for this type of thing.

C

[Saurav]

What Chris said. I would measure with ARTA, export to FRD, then import into LspCAD for running simulations.

[Johnloudb]

Well, all I want to do is apply some 1/6 octave smoothing to some measurements from my iphone. I guess I got shell out $200.00 + for LspCAD and a new sound interface? Hmmm ... might bother Hank about his kind offer earlier, but I would still need the sound interface. Money is tight right now. Well I think I will wing it with my iphone, and see how that works out. It'll be fun

[Saurav]

I believe HOLMImpulse and Speaker Workshop (both free) can load/import FRD files, assuming that's what your iPhone app is exporting. I think the paid version of ARTA has the import/export capabilities.

You only need LspCAD if you want an automatic crossover optimizer. Everything else you can do for free. Speaker Workshop can model a passive crossover.

[cjd]

why smooth so much?

Might be cheaper to drink before looking at the graphs?

[Johnloudb]

Chris, yeah, that could work lol

Everyone else posts smooth response graphs and I just want an idea of how the response of my speaker compares to ones others have posted. I'm all new to this speaker measurement stuff. well, I'm going to have to get a sound card eventually so am looking for a good one I can afford.

[Johnloudb]

Saurav, thanks, I'll see if I can import some files into ARTA.

[Hank]

John, PM me with your address if you want to borrow my PE measurement mic system.
Also, keep cjd's advice in mind :boozer:

[cjd]

Not that it really matters but I always post 1/24 octave smoothing, which is the minimum available. Smoothing is for sissies.

[Johnloudb]

Thanks Hank! Appreciate it.

Chris, I wish I had 1/24 octave smoothing. Glad to know that you don't feel the need for smoothing. Gives me hope.

[Johnloudb]

Okay, I'm still kicking. I took some measurements with my iphone with the speaker positioned where It'll be when I'm done with the crossover. I moving the speaker a bit did affect the response quite a bit. This is about 30 degrees off axis from the woofer, at the listening position.


Images not available


Anyhow, my I will be sticking with my iphone and designing the crossover with those measurements. Just don't money right now for a sound card or anything, girl friend is struggling and I am too now.

[Johnloudb]

Here is the RD50 response graphs:


Images not available


Everything below about 700Hz can be ignored because of a high filter at the amp input cuts off the low frequencies. The top graph is not really flat below 700Hz, it's just a software artifact.

[Saurav]

I get broken image icons for your last 2 posts. I was really interested in seeing those graphs

[Johnloudb]

I get broken image icons for your last 2 posts. I was really interested in seeing those graphs

Hmm, sorry ... my iphone gets those too. I can see them on my computer okay though. I'll reinsert them as attachments this time instead of using dropbox.com

My take is that a boost of about 10 - 12dB below 250Hz might do the trick. And then probably a crossover frequency of 600Hz. If I hear no complaints, I'll start wiring up the crossover.

[Johnloudb]

Response Graphs

Okay, I've attached the response graphs that are missing above. The first two graphs are the woofers, sweep and pink noise response smoothed. The next two are the RD50, same deal.

[Hank]

Progress, John - what's the next hurdle?

[Johnloudb]

Progress, John - what's the next hurdle?

Building the crossover! :dancenana: ... either I'm on the right track, or hopelessly lost, or everyone else has given up on me and abandoned this thread?

In any case, I see the light or mirage at the of the tunnel.

[Saurav]

I think your initial XO points make sense based on those graphs. But I think it's going to be hard to design a crossover based on RTA-like measurements. And I'm not even sure if that's what these are, but they don't look gated to me. For instance, if you go for an even-order XO, it'll be hard to tell if you have the phase right because it'll be hard to see a reverse null, and things like that. But this is where I'm far outside my depth, so I should just stop talking

[Hank]

Keep the faith, John - one of the crossover gurus will surely jump in to guide/help you. I wonder if your speaker configuration would benefit from a Cauer filter method. Jon could comment on that.

[Johnloudb]

I think your initial XO points make sense based on those graphs. But I think it's going to be hard to design a crossover based on RTA-like measurements. And I'm not even sure if that's what these are, but they don't look gated to me. For instance, if you go for an even-order XO, it'll be hard to tell if you have the phase right because it'll be hard to see a reverse null, and things like that. But this is where I'm far outside my depth, so I should just stop talking

Okay, I don't know what gated means. But the blue plots are a swept sinusoid variety not RTA. The smooth plots are RTA with pink noise source.

[Saurav]

A gated measurement is where you filter out the reflections from the room so you're seeing just the response of the speaker. If you can export the swept sine measurement as an FRD file, ARTA/HOLM/Speaker Workshop/etc. can turn that into a gated measurement. That will remove a lot of the noise/spikes you're seeing in those charts and let you see the real response of the drivers. Of course, gated measurements have their own problems, most notably reduced resolution as you go down in frequency, so you generally can't use them for woofer work, at least not measurements taken in normal rooms. If you have a large and/or treated room, you may be able to do better. The larger the room, the later the reflections, which means you can include more of the measurement in the analysis, which improves low-frequency accuracy. Don't ask me why the math works that way, I don't remember any more

[JonMarsh]

A gated measurement is where you filter out the reflections from the room so you're seeing just the response of the speaker. If you can export the swept sine measurement as an FRD file, ARTA/HOLM/Speaker Workshop/etc. can turn that into a gated measurement. That will remove a lot of the noise/spikes you're seeing in those charts and let you see the real response of the drivers. Of course, gated measurements have their own problems, most notably reduced resolution as you go down in frequency, so you generally can't use them for woofer work, at least not measurements taken in normal rooms. If you have a large and/or treated room, you may be able to do better. The larger the room, the later the reflections, which means you can include more of the measurement in the analysis, which improves low-frequency accuracy. Don't ask me why the math works that way, I don't remember any more

In agreement; also, the type of window function one uses for the time gated impulse response has an impact on the degree to which the reflections affect the upper range response versus providing low frequency information; a half Hanning or Half Hamming window works very well, such as is used in Praxis or Fuzzmeasure.

Praxis and Fuzzmeaure and REW (Room EQ Wizard; free on Home Theater Shack, runs in Java machine; PC and Mac versions available) use sine chirps (fast swept test waveforms) which are then converted to the equivalent impulse response mathematically, THEN gated and window shaped, then converted back to a frequency response.

Inherently, if you only have 10 msec of data, for example, with a short gate, the LF data is not there because you aren't looking at things long enough to sample a low frequency waveform impulse. Basically, it's gate time divided by 1 sec, to give the frequency resolution, so 10 msec means you only have 100 Hz resolution and limit for low end- and that's just one data point at 100 Hz and other at 200 Hz! With a 50 msec window, you have 20 Hz resolution in your data and a minimum data point at 20 Hz. But that translates to 50 feet (approximately) of path length, which gives the audio a lot of time to bounce around...

OTOH, if your indoor listening setup has too many reflections and issues to get good measurements, well guess what? It's going to sound that way, too...

There are a lot of truths people often try to overlook for the sake of convenience- convenience in speaker setup, room arrangement, system measurement, etc. In the end, that doesn't work or get you to where you need to be. I'm giving you a little "tough love" here, but I think it's called for. I love my iOS devices, but I haven't seen software yet that's suitable for getting the data needed for speaker measurements. If you must go free, go REW. Or perhaps ARTA, though I have no experience with it. Speaker workshop I've tried and didn't personally like, but we all have our subjective preferences.

I'll tell you right off the bat that the easiest way to get going with your system for now would be a 4th order LF at 600Hz, flat on the RD driver as long as you're using the notch filter, and dipole single pole boost on the LF from the crossover point down to where you cross to your subs- you may need to tweak that, but it shouldn't be far off. Get off to the subs around 100-125.

[Johnloudb]

Okay, thanks very much Saurav and Jon for your insight. I know my measurements leave a lot to be desired, but just going to go with that for now. My crossover will be easy to adjust and reconfigure so I don't have to have everything figured out right now. What does concern me is that our current speakers (Nelson Reed 804C) do measure quite flat with my iphone, with some ups and downs in the low frequencies. The room is well treated for the speakers plus subs. We have ASC Bass traps in the corners and 6 Skyline diffusers along the back wall. It's a decent sized room 14ft deep and 18-20ft wide. So the side wall reflections are not bad, especially with my dad's dvd library covering both side walls.

But this speaker is a different animal and I'll have to work on placement and stuff. I have many more Skyline diffusers in the garage to put behind the speakers.

Anyway, I'll post some measurements of our current setup later. And post my crossover schematic in just a second.

[Johnloudb]

sometimes I can't add attachments like I used before the forum upgraded. I did a few posts back but I usually see no option (manage attachments) to attach uploaded photos. I don't get it, maybe it's a browser problem? I use google chrome.

[Johnloudb]

If you can view the images, click on them to make them larger.

Here is the 4th order crossover circuit with a 600Hz crossover frequency.

Image not available

Here is the equalization and gain circuit that goes just before the crossover.

Image not available

Here is the response for my crossover.

Image not available

[Johnloudb]

I was getting broken image links on my last post. I reloaded the page a couple times and the pictures loaded. If anyone is having problems viewing the images, let me know.

On the low-pass (top) equalization circuit R34, R35, and C1 make up the shelving filter with a corner frequency of 113Hz. I'll use a potentiometer in place of the resistors to control the LF boost.

On the high-pass (bottom) equalization circuit R41, R43, and C2 make up the shelving filter for the RD50 with a corner of frequency of 1.6kHz. Again, this will be adjustable and I may not use it at all, unless I think the speaker sounds a bit hot.

On the crossover, the top filter is the lowpass filter.

[Johnloudb]

Just measured our room and it's 15ft by 21ft, so I hope that will accommodate this speaker okay. Hope 3.5 ft behind the front baffle to the rear wall is good enough. Can not do much better than that and maintain a minimum listening distance of ten feet from the speaker.

Okay, trying to speed things up hear now and will try to get most of my crossover parts ordered tonight. I'm not going to bread boad this thing cause it's too much of a hassle and I'm using my own generic PCB design that allows use of surface mount components. I'm using the very good bf862 JFET (suface mount) and the LME49720 opamp. And battery power supply with voltage regulation. That's all for now till I get some parts.

[Saurav]

OTOH, if your indoor listening setup has too many reflections and issues to get good measurements, well guess what? It's going to sound that way, too...

I would love it if you could write up an article some day on indoor measurements in compromised rooms. I know there's no silver bullet, but it's hard to find good info on the various tradeoffs and concerns. All the stuff out there says "measure outside" or "treat the room", hardly anyone goes into how to look at the data in a less-than-perfect room. And if you get into dipoles, tall / line sources, etc., then the information gets even more scarce.

[Hank]

Damn, John! I thought nobody could exceed Evil Twin's crossover parts count, but you may be on your way to king of the hill status. Are you considering going active in the future?

[Hank]

duplicate

[Johnloudb]

Damn, John! I thought nobody could exceed Evil Twin's crossover parts count, but you may be on your way to king of the hill status. Are you considering going active in the future?

Hank that is an active crossover. Yep, lots of parts but all necessary achieve the desired goal. All the parts that start with a J are JFET transistors and X1A and X1B are op-amps (triangles), the V's are power supplies. It's a Spice simulation schematic and does look similar to other crossover schematics posted here ... I'm sure you'd have noticed the difference on closer inspection though.

Anyway, I matched maybe 1/2 of the 12 transistor pairs I needed, and probably need to order some more.

[Hank]

Hi John - nope, didn't notice - just got on line for a quickie to see if you'd posted recently. Active it is - should be a fun project.
Later

[JonMarsh]

yep. Looks like we have a Sith apprentice in training here, Hank! :B