• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

DIY speakers with Waveguide

dc655321

Addicted to Fun and Learning
Joined
Mar 4, 2018
Messages
585
Likes
553
#41
https://www.diyaudio.com/forums/multi-way/76977-arta-60.html#post5492173
" I've tested latency variations with Umik-1 and external sound card (but not with ARTA). Conclusion was that severe timing/phase differences exist between different measurements. Timing may stay quite stable few measurements but then delay might jump ~0.5...>1 ms back and forth. Rebooting or starting another program between measurements was not needed to generate delay jumps, and latency adjustment of sound card did not help.
So, it is not valid gear for advanced speaker measurement needed to simulate multi-way to off-axis, power, DI responses etc. because that requires very stable timing i.e. semi-dual (or dual) channel connection and measurement mode selected in ARTA (or REW). "


https://www.diyaudio.com/forums/multi-way/76977-arta-60.html#post5492272
"Single channel gear such as simple USB mics are very common today. They have spread from room acoustics measurements to speaker design which is unfortunate because more and more new users end up to bad, inaccurate/random and very slow design methods. This is not my fight, so I simply recommend to ignore messages and authors supporting or advertising such gear from now on. "

REW has given possibility for 2-ch measurements only for some time, after kimmosto got John to understand this. REW is primarily a room measurement/analysis program, impedance meas and 2-ch are extra add-on features .
Please don't take this the wrong way, but can you show or point at some data/evidence indicating a problem measuring timing/phase with current versions of REW+UMIK-1?
 

dc655321

Addicted to Fun and Learning
Joined
Mar 4, 2018
Messages
585
Likes
553
#42
Not sure if I have it right in my mind.
Me neither! In my mind, that is...

If I'm wrong then moving the reference speaker shouldn't impact the relative delays of the two drivers.
It obviously will have an impact if the reference is moved between measurements.
But, I don't really think that is what you meant.

At the end of the day, I just want to know if I can reliably use the equipment I have on hand to build a DSP-based active speaker, or if I need to shop/research better instrumentation.
 

Juhazi

Addicted to Fun and Learning
Joined
Sep 15, 2018
Messages
859
Likes
932
Location
Finland
#43
Using a reference speaker has nothing to do with measurements for loudspeaker xo design!

Setup for semidual: ( gives exact "time of flight" acoustic delay to calculate distance z between source and microphone.)



"Please don't take this the wrong way, but can you show or point at some data/evidence indicating a problem measuring timing/phase with current versions of REW+UMIK-1?"

No, I am not kimmosto nor am I capable of performing such measurements. Please contact the vituixman directly yourself!

Dear dc, if you are making a dsp-based active speaker, doing acoustic measurements for simulation is not necessary. I have set up several dsp-active multiway speakers using UMIK-1 and REW using no simulation at all. First I measure on- and off-axis of each driver in the proto or final box, do eq and gain adjustments. Then I make an educated guess of which slopes are set at which freq. and set xo parameters. The way I set relative delays is by looking at step response of two or more drivers playing in a sweep and by checking response summation/nulling difference. But this is the "wrong and winding road to hell" method, if you ask kimmosto!

I followed these instructions
https://www.hifizine.com/2010/12/prototyping-4-way-open-baffle-speaker-with-the-minidsp-2x4/
https://www.hifizine.com/2011/03/refining-a-4-way-open-baffle-speaker-minidsp-2x4/

And this was my first dsp project
https://www.diyaudio.com/forums/multi-way/231353-aino-gradient-collaborative-speaker-project.html
 
Last edited:

dc655321

Addicted to Fun and Learning
Joined
Mar 4, 2018
Messages
585
Likes
553
#44
Using a reference speaker has nothing to do with measurements for loudspeaker xo design.
I thought the conversation was largely about measuring acoustic centers.
I'm not concerned about xo design.

But, time alignment would be cool.
And it's essential for the design I have in mind to build.
EDIT: seems to me like a characteristic any properly engineered transducer ought to have.

Dear dc, if you are making a dsp-based active speaker, doing acoustic measurements for simulation is not necessary. I have set up several dsp-active multiway speakers using UMIK-1 and REW and no simulation at all. First I measure on- and off-axis of each driver in theproto or final box, do eq and gain adjustments.
Similarly to xo design, I have no interest in simulation either. Not sure where that notion comes from (same place as xo design?)?

The way I set delays (relative to each other "way") is by looking at step response of two or more drivers playing in a sweep and by checking response summation/nulling difference.
Sure. Using white/pink noise, spectrum averages, and polarity flipping, delays can be introduced to produce similar null/summing results between level-matched adjacent drivers. Hopefully without venturing down "a winding road to hell" :)
 

Juhazi

Addicted to Fun and Learning
Joined
Sep 15, 2018
Messages
859
Likes
932
Location
Finland
#45
In post #1 617 said that he is planning/simulating a multiway passive loudspeaker with vituixcad
In post #2 you said that you have a similar plan

So I have all the time thought that this thread is about passive xo design!

Active dsp-controlled multiway speakers do have crossovers as well, but done with multichannel dsp instead of electric passive circuits. The acoustic responses of drivers and their summation is always what we are trying to control and match with a xo.

When we are setting up a multiway active speaker or PA system, we are interested in relative delay differences between drivers or units, to get good phase match. Using flight time delay values is one way, summation/step response analysis is the other. With dsp it is easy to change delays and measure again in few seconds, but for simulation we need solid and reliable "startup" measurements.
 
Last edited:

HammerSandwich

Addicted to Fun and Learning
Forum Donor
Joined
Nov 22, 2018
Messages
578
Likes
614
#48
Sorry, not from me. I bought a Behringer USB interface to play with this stuff on the cheap. Their $25 power-only box might be worth a look, if you're happy with the Asus.
 

617

Senior Member
Forum Donor
Joined
Mar 18, 2019
Messages
414
Likes
607
Location
Somerville, MA
#49
There's been some interest in the cabinets I had made - see attached pictures

Cabinet with drivers. Driver rebates are so snug that the units are press-fit right now. I asked for a small rabbet on the back of the front baffle to create a shadow line. Tends to look nicer than hoping for a perfect glue joint. You can also see one of the patches on the lower left corner - not really noticable but if I stained these I bet it would pop out. Frankly I wouldn't stain these cabinets - either a simple clear finish or veneer.

For finishing I used a General Finishes water base clear. Dries very hard with no yellowing but application is trickier than the wipe-on poly I normally like to use. Wipe on poly will have more amber color and bring the grain out more, which is not the look I wanted for these pale birch cabinets. I used 4" foam rollers, 2 or 3 coats, then finished with wax. The wax evens out the sheen and gives a silky finish if you rub it with scotch brite pads.


DSCF6752.JPG

I put a little chamfer on the back panel so that there would be a nice little line all the way around. Binding posts are the knock in dayton ones. The satin nickel is by far the nicest finish - the gold doesn't look great in my opinion. Port is a knock in port from PE, ideally it would be 1/4-1/2" longer but I don't think bass tuning needs to be that exact. I will probably end up stuffing the ports anyway.

DSCF6753.JPG


I got nice stainless steel pan head screws to hold the back on. Printed a template and drilled them all by hand with a little drill guide block. The guy who makes these isn't making fine furniture - I'd say the quality of the cuts is fine but I did use a little wood filler in some parts. The plywood is 'void free' but there are tiny irregularities and pits on the cut edges if you look close. If I was making a big speaker I wouldn't care but for a small speaker you tend to notice stuff like that more. Again, these would be a dream to veneer, and the plywood is much nicer to deal with than mdf. Textured PA paint like duratex would make a very professional looking product with next to no effort.

DSCF6754.JPG

I've spent hours veneering and finishing speakers which look nicer than this, but I think I prefer this low-effort cabinet. Paint takes days of work to get looking right and the seam lines always (always) show through no matter how much bondo and glazing compound you use.

I would order from this business again - the cabinets were made precisely to my measurements and thoughtfully crafted.
 

617

Senior Member
Forum Donor
Joined
Mar 18, 2019
Messages
414
Likes
607
Location
Somerville, MA
#50
Well I managed to take measurements today. I followed the new vituixCAD 'measurement preparations' document to the letter. It helps that I use ARTA, which is what VituixCAD is really designed for.
https://kimmosaunisto.net/Software/VituixCAD/VituixCAD Measurement Preparations.pdf

I merged my woofer far field measurements with a near field measure but did not include the port. With the port, I would be seeing a boost centered at 66hz or so, extending bass to an f3 of 45.
LR4_2400_SPL.png

LR4 at 2400hz

The way I like to design is to use the polars to find the best crossover frequency and slope, then I take dedicated on-axis measurements and design in 2D. I do this because I lack confidence that I am aligning everything perfectly with the merged and manipulated measurements - I have more trust that all my phase relationships and SPL relationships are perfect if I can measure both drivers without moving the mic or the speaker or any measurement settings. Perhaps I'm old school but I do aim for a very flat and smooth axial response.

To find the best crossover frequency and slope, I use VCAD's active components to quickly hammer the response into shape which allows quick visualization of different slopes.

I'm not used to designing with paper midwoofers and waveguides, so I was pleasantly surprised to see that the only headache in this design is the breakup and diffraction from the woofer. The tweeter's response is pretty easy to deal with apart from the huge gain in the boosted region.

Due to the directivity matching of the waveguide and woofer, I am afforded a wide bandwidth where the crossover point could be - I tested from 1800-2400hz, a very typical range, and found that the DI was smoothest with a higher crossover. This is to be expected. What surprised me is that the slope (second or fourth order typically) is not really that important - normally a shallow slope is used to 'blend' the directivity of two different sized drivers, but here no blending is needed. As a result, a fourth order, relatively high crossover looks best.

1574565747449.png

With a 2400hz LR4 crossover, I have nice smooth and tight (for a small speaker) directivity from 600hz on. I regret not measuring out to 180 degrees - when I test the crossover I will be sure to do that. None of the irregularities you see at e.g. 4k are due to the crossover - the directivity appears damn smooth through the crossover region. I love waveguides!

So what is my target crossover? The issue is the paper cone breakup. Using a steep crossover helps a great deal, but this driver breaks up relatively low, so I'm going to try to do LR4 as high as I can, ideally 2400 hz but perhaps lower, or perhaps I will do a crossover which is fourth order for a few hundred hertz, and then gets steeper to kill the breakup of the woofer.
 

Arnandsway

Senior Member
Joined
Sep 13, 2018
Messages
325
Likes
218
#51
Could you show the individual polars of the woofer and tweeter? To me it's still a little vague how one decides the best crossover point based on those.
 

617

Senior Member
Forum Donor
Joined
Mar 18, 2019
Messages
414
Likes
607
Location
Somerville, MA
#52
Could you show the individual polars of the woofer and tweeter? To me it's still a little vague how one decides the best crossover point based on those.
Woofer directivity - I don't know what's going on at 10-20K, looks like an error from normalization (this diagram shows the directivity if 0 degrees was perfectly linear)
1574610691153.png


Tweeter - similar errors from normalization:
1574610792268.png


I guess one simple way to choose a crossover frequency would be to juxtapose the images and find a point where they sort of match up, I made a little wiper diagram:
https://cdn.knightlab.com/libs/juxt...html?uid=fbb57e58-0ed2-11ea-b9b8-0edaf8f81e27
 

Juhazi

Addicted to Fun and Learning
Joined
Sep 15, 2018
Messages
859
Likes
932
Location
Finland
#53
There are many ways and preferences of how to select xo and optimize crossover type. Vituixcad has optimizing too, where predicted room response and DI smoothness are priority. This might lead to distortion problems with the tweeter. Prioritizing shallow slopes might lead to distortion and poor directivity consistency etc.

Your plan looks good to me! Typically WG tweeters are crossed at 1.5 - 2kHz, but this smallish wg obviously works better this way.

Off-axis responses of single drivers without smoothing or normalization shown as overlay is what I want to see. Graphical spectrogram gives the big picture only.
 
Last edited:

Arnandsway

Senior Member
Joined
Sep 13, 2018
Messages
325
Likes
218
#54
Woofer directivity - I don't know what's going on at 10-20K, looks like an error from normalization (this diagram shows the directivity if 0 degrees was perfectly linear)
View attachment 40108

Tweeter - similar errors from normalization:
View attachment 40111

I guess one simple way to choose a crossover frequency would be to juxtapose the images and find a point where they sort of match up, I made a little wiper diagram:
https://cdn.knightlab.com/libs/juxt...html?uid=fbb57e58-0ed2-11ea-b9b8-0edaf8f81e27
Looking at the polars, I guess it could be crossed over around 1,5 khz as well? Or am I missing something?
 

jhaider

Addicted to Fun and Learning
Joined
Jun 5, 2016
Messages
540
Likes
447
#55
Did you see that the newest version of VCAD will generate full Spinoramas if you take the full measurement complement? Simply amazing what this software can do.

I guess one simple way to choose a crossover frequency would be to juxtapose the images and find a point where they sort of match up, I made a little wiper diagram:
https://cdn.knightlab.com/libs/juxt...html?uid=fbb57e58-0ed2-11ea-b9b8-0edaf8f81e27
Another way to do this is to use Keynote - line up the polars and adjust the transparency of each figure.
 

Juhazi

Addicted to Fun and Learning
Joined
Sep 15, 2018
Messages
859
Likes
932
Location
Finland
#56
Those will not show what the xo does, and we must look at vertical lobing too. Lobing because of separated sources cannot be eliminated, but low xo and shallow slopes will make it softer and less attenuating. Vertical polars must be taken to consideration too!
 

Severian

Active Member
Forum Donor
Joined
Jan 11, 2019
Messages
125
Likes
81
#58
Great thread! I'm often surprised there isn't more discussion of DIY speakers here.

I currently use a pair of SEAS Idunn speakers that I built using a kit from Madisound, which at the time included very nice finished cabinets. They use the SEAS waveguide tweeter that I suspect you were referencing in the OP (27TBCD/GB-DXT) along with the U18RNX/P woven polypropylene woofer. I think the drivers, particularly the tweeter, sound exceptional although I've always found the voicing of the speaker to be too bright for my tastes. In my current configuration I have them crossed over to a monster DIY subwoofer setup and use my DSP to drop the high frequencies a few dB. I've always been interested in redesigning the crossover to achieve something closer to this voicing.

At the moment I'm pondering my next move. While I love the overall sound quality I'm currently getting, I feel that the single 7" woofers aren't giving me the tactile impact and dynamics that I want. As a result I'm running my subs higher than I ought to. They also don't play quite as loud as I'd like. I haven't decided whether I should stick with the hi-fi style drivers but in an MTM alignment, or whether I should take the plunge into pro woofers and waveguides, or perhaps the new DIY Sound Group coaxials.
 

Head_Unit

Member
Forum Donor
Joined
Aug 27, 2018
Messages
28
Likes
15
#59
I think the best characteristic is "disappearance of burglary under angles." Seriously though, bless Google translate for yielding English-I could figure out the German but it would be way slow!

The "with" directivity curves remind me of an AES paper by I think Panasonic where they put a kind of half-donut in front of their tweeter. Similar curves, great evenly reducing dispersion, rolloff towards the high frequencies. For this one the sensitivity is kicked up about 9 dB at 1500-2000, if matched to even a fairly sensitive woofer you could apply a quite high frequency crossover and the tweeter would be cruising on reduced power. Now if only we could similar miracles out of phase plugs! I wonder if anyone has ever tried really huge phase plugs, maybe even with segments in the center or something.

Actually, matched with some hypothetical 96 dB tweeter, there would be a top-end rolloff. But wouldn't something like Audyssey EQ that out?
 

dc655321

Addicted to Fun and Learning
Joined
Mar 4, 2018
Messages
585
Likes
553
#60
I think the best characteristic is "disappearance of burglary under angles."
Clear as mud, right? :D

I wonder if anyone has ever tried really huge phase plugs, maybe even with segments in the center or something.
Wouldn't that just be a type of acoustic Fresnel lens? I think that has been done, but can't remember where atm.
 
Top Bottom