critique my DIY speaker build - AVS Forum

AVS Forum > Audio > DIY Speakers and Subs > critique my DIY speaker build

DIY Speakers and Subs

formfactor's Avatar formfactor
04:04 PM Liked: 10
post #1 of 48
04-16-2008 | Posts: 72
Joined: May 2007
This was my first DIY build so any please run any comments or suggestions positive or negative my way. The speakers are already built, I'm really just looking for suggestions on how to improve them or for things I could try in my next build. In particular, I'd like to know your opinions on crossing over the drivers at such a high frequency, the driver spacing, and how to take properly take accurate measurements of the speaker's response. I have the drivers crossed over high because the ear is less sensitive in that region and the off axis response of the woofer is still very good. The driver spacing is purely cosmetic. They're lined up centerwise but I think they should be closer spaced when crossed over at such a high frequency. I also see a dip in the response at 1kHz in the measurements. I'm not sure if that's something inherent to my design or if it's because I'm holding the mic in my hand with the speakers on a large desk in a small room. Or maybe it's a null due to both drivers playing at the same time... hey they're rough measurements. Don't kill me.

Speakers are a 2-way design and utilize Vifa P17WJ-00 6-1/2" woofers and Vifa XT25TG30-04 tweeters.

Cabinets are about .6-7 cubic feet in volume and moderately stuffed.

Port is tuned to around 55 Hz which gives a slightly peaky response before falling off.

Drivers are crossed over at roughly 3kHz with a third order butterworth filter on both high and mid.

Additional crossover components include attenuation to adjust the tweeter sensitivity and impedance compensation for the woofer's rising voice coil inductance.

This is the crossover simulation schematic including the inductor DC resistances and modeled driver impedances.

Here is the actual simulation.

Here is a quick measurement using WinMLS.
This is a quick (rough) near-field measurement using Speaker Workshop.

Plots removed, see later posts for better plots

Picture of the final speaker build.

And finally a picture of the finished crossover.

Thanks for looking.
AlexE's Avatar AlexE
04:30 PM Liked: 10
post #2 of 48
04-16-2008 | Posts: 440
Joined: Feb 2008
Wow they look great!

How did you finish them? Looks like sanded + clear on bedliner?
formfactor's Avatar formfactor
05:44 PM Liked: 10
post #3 of 48
04-16-2008 | Posts: 72
Joined: May 2007
Actually just sanded the cabinets down, primed them, and then put a few coats of satin black paint on. It turned out matching the mounting flanges of the drivers really well. Here are a few more pics.

augerpro's Avatar augerpro
06:10 PM Liked: 29
post #4 of 48
04-16-2008 | Posts: 1,715
Joined: Aug 2005
Hmmm...don't really know where to start. You're measured responses are not reliable if you are holding the mic in your hand. They are definitely not BW3, and cross at more like 2.2k Hz if we are to believe the measurements. You just simulated the electrical network, which means nothing without accounting for the actual response of the drivers. I've never seen a driver represented on a schematic as some sort of electrical model. How did you come up with that model?
formfactor's Avatar formfactor
06:53 PM Liked: 10
post #5 of 48
04-16-2008 | Posts: 72
Joined: May 2007
I agree, my measurements are probably somewhat (if not totally) out of whack with the mic in my hand, the speakers on the desk, and the AC running in the background. I'd like to learn how to take better measurements but I haven't really come across anything that describes how to do it very well.

The crossover schematic I have in the first post contains the 3rd order butterworth filter in addition to:

-the DC resistances of the inductor windings (the resistor in series with the inductor coils on the schematic),

-the L-pad for a 3.5 dB tweeter attenuation in the high pass network,

-impedance compensation (the capacitor and resistor in parallel) for the rising impedance of the woofer's voice coil at higher frequencies,

-and the modeled DC resistance and inductance of each drivers voice coil along with its mechanical response (the right most portion of the high pass and low pass networks)

So you're right, it's not a pure 3rd order butterworth filter, but it's in there.

The parameters used to model the impedance presented by the drivers at different frequencies were derived using this website:

The site describes how the drivers can have their impedances modeled using the DC resistance of the driver's voice coil along with its inductance in addition to the back EMF calculated by the driver's moving mass. The program he uses is Orcad/Pspice, which is a circuit simulation program that electrical engineers (like myself) are very familiar with. So instead of modeling the crossover with a constant 8 or 4 ohm load, you can substitute that model in and get the exact response of your filter to the changing impedance of the driver.

How the values for the model came to be I'm not sure about. But what I do know is that once calculated, the results are strikingly similar to what's measured from the driver in real life.

Using the tweeter I used, you can model it like below:

You can then perform an AC sweep and then dividing the voltage by the current running through the circuit you can derive the impedance of the driver:

You'll find that it looks strikingly similar to the published specs from the manufacturer:

All that's left to do is to plug that impedance response into filter/crossover you'd like to model and you'll get a close to real life transfer function of the filter with the particular driver across the frequencies of interest.

It makes sense to me as an engineer, I hope that clears things up for you.
Vinculum's Avatar Vinculum
07:24 PM Liked: 12
post #6 of 48
04-16-2008 | Posts: 944
Joined: Dec 2001
While they look great, I would have used nicer fasteners than those cheap looking philips heads!

Dr V
augerpro's Avatar augerpro
07:28 PM Liked: 29
post #7 of 48
04-16-2008 | Posts: 1,715
Joined: Aug 2005
Ah, I figured it was just the impedance model and that confirms it. You now need to include the actual driver response into that model. If you want a BW3 acoustic response you need to include the driver response, using a third order network with text book BW3 calculations will not result in an actual BW3 response. It most likely will not need a 3rd order network to accomplish this. Also the zobel is not needed on the woofer if you have the actual impedance and frequency response.
formfactor's Avatar formfactor
07:41 PM Liked: 10
post #8 of 48
04-16-2008 | Posts: 72
Joined: May 2007
Oh hmm, interesting. So 3rd order electrical response /= 3rd order acoustic response. By 3rd order acoustic response you mean 18dB/octave falloff? Am I getting this right?

Can you elaborate on not needing the zobel network? Without it, there's an awful peak that I heard and measured around the crossover point. Is there a way around it?


While they look great, I would have used nicer fasteners than those cheap looking philips heads!

Aiy. It was a choice between a few cents worth of screws and $10 plus shipping for a nice pack of 100. Any good places to get them from? I tried but I didn't really find anything I thought would look good. I just have #6 and #8 metal screws right now with fiber washers for dampening.
TheEAR's Avatar TheEAR
08:29 PM Liked: 12
post #9 of 48
04-16-2008 | Posts: 3,464
Joined: Nov 2006
Very clean work on the crossovers! The cabinets are simple but well done,the screws holding the drivers...just for looks get some hex ones from PE for example.
hossage's Avatar hossage
10:35 PM Liked: 10
post #10 of 48
04-16-2008 | Posts: 42
Joined: Feb 2008
Why not use some kind of adhesive to glue them on instead?
formfactor's Avatar formfactor
06:46 AM Liked: 10
post #11 of 48
04-17-2008 | Posts: 72
Joined: May 2007
You guys just won't give me a break on the screws. Ok, I'll find something better.

Any suggestions on tuning the crossover or taking measurements? Anything?
augerpro's Avatar augerpro
12:08 PM Liked: 29
post #12 of 48
04-17-2008 | Posts: 1,715
Joined: Aug 2005
What modeling software are you using? That should be able to do measurements. Speaker Workshop does measurements and modeling and is free, but kind of user hostile. Soundeasy and LSPCad are great packages but both cost a bit more than $200. I think LSPCad has a free version that does everything you need, but you can't save the files. So you would have to model the crossover and write down the values and remodel it every time you open the program. Any of those have guides you can checkout for measurement basics since they all use similar methods (MLS signals). Here is good SE wiki:
AlexE's Avatar AlexE
01:51 PM Liked: 10
post #13 of 48
04-17-2008 | Posts: 440
Joined: Feb 2008
As for screws anything will work, I like the R-bits (square hole) for looks and utility
devani's Avatar devani
02:02 PM Liked: 10
post #14 of 48
04-17-2008 | Posts: 37
Joined: Jan 2008
well, the problem with Orcad simulation is's simulation...the hardest part is finding the actual parts that are pretty close to the values of the simulation...and how close were you to the simulation values? C, L, R values...did you run another simulation with actual values?? if you did what % difference?
I use Filter Solutions 9.1 to design butterworth filter...and ORCAD 15.7 Capture well for work anyway....I am also an EE and have been using ORCAD for 10 years now...
augerpro's Avatar augerpro
02:41 PM Liked: 29
post #15 of 48
04-17-2008 | Posts: 1,715
Joined: Aug 2005
If all it does is model electrical networks then it's worthless. You need the actual driver response and impedance included in the model. If the response and impedance of the drivers was perfectly flat then the sim would match reality, but no drivers like that exist.
formfactor's Avatar formfactor
05:19 PM Liked: 10
post #16 of 48
04-17-2008 | Posts: 72
Joined: May 2007
Devani: The values in the simulation are already the ones that can be bought off the shelf. The ones I couldn't buy within tolerance I made by combining elements.

auger: Could you elaborate on the response of the drivers and what else needs to be taken into account for the simulation? I'd like to find out what else can be done.

Oh, and the only modeling programs I've used are winISD and orcad/pspice. I have a bunch of other software installed including speaker workshop and demos of winMLS and other measurement programs. Part of what DIY means to me is spending as little money as possible so I frowned at spending $200+ for a modeling program. I suspect that feeling will change when I start learning more about what I can do.
formfactor's Avatar formfactor
08:40 AM Liked: 10
post #17 of 48
04-18-2008 | Posts: 72
Joined: May 2007
Oh I think I know what you're getting at now auger. I think all I have to do is scale the electrical output from the simulation to the actual acoustic frequency response plot of the driver using an FTABLE in pspice.

I think that's what programs like soundeasy and speaker workshop do by nature. You measure the parameters of the raw driver and then the program and simulate the acoustic output of the entire network using the actual driver characteristics.
augerpro's Avatar augerpro
11:48 AM Liked: 29
post #18 of 48
04-18-2008 | Posts: 1,715
Joined: Aug 2005
The programs I have use the actual measured frequency response of the driver. I don't what yours do. The plot you posted earlier is merely the transfer function of the electrical network. Obviously that does not represent the actual frequency response of the driver hooked to the crossover, since the driver has it's own contributing transfer function composed of the impedance and frequency response on its own.
formfactor's Avatar formfactor
08:26 PM Liked: 10
post #19 of 48
04-18-2008 | Posts: 72
Joined: May 2007
I'm not clear on what you're trying to say auger. Both driver's impedances have been more or less modeled in the CAD simulation when connected to the crossover network. It's the same thing speaker workshop and soundeasy does when you hook up the 'driver' element in the crossover design editor only mine is based on an equivalent model of what you actually measured. The only thing that's left is the variance in acoustic output of the drivers which according to the spec sheets for both, remains relatively flat in the frequencies of most interest to me.

Am I missing something? I don't see how a simulation like this could be totally useless.

Just to clear things up, the impedance of the 'driver' in my simulation is not flat. It's a model of the actual impedance of the driver (which is very close to what you would measure from it yourself.) The transfer function on the plot I made is the actual signal being passed to the driver from the filter network given the driver's respective impedance at a given frequency.
augerpro's Avatar augerpro
11:55 PM Liked: 29
post #20 of 48
04-18-2008 | Posts: 1,715
Joined: Aug 2005
SE, LSPCad, Praxis, Leap and SW use an impedance AND frequency response models for the drivers, yours is only using impedance. Don't you wonder why the transfer function crosses at over 3k Hz yet the plots indicate the driver crossing at 2.2k hz? No driver is so flat that you can assume it will behave ideally according to just an electrical network sim. And those drivers are only flat on an infinite baffle-on those little ones you have I guarantee the woofer is far from flat, in fact about 6-8dB from flat. And how do you know drivers are in phase? Again you are only modeling the electrical filter, NOT the loudspeaker as whole. I really don't know any other way to explain this. Get a real loudspeaker modeling program and check it out.

EDIT: and go buy a couple loudspeaker design books
augerpro's Avatar augerpro
12:06 AM Liked: 29
post #21 of 48
04-19-2008 | Posts: 1,715
Joined: Aug 2005
Here is example of what I'm trying to explain, from Zaph's slim wall mounted speaker here is the response of the individual drivers:
Roughly LR2 slopes crossing at 3.5K Hz. Now here is the transfer function (sim) of the electrical network:

Clearly NOT LR2 @ 3.5kHz, but it's just a means to an end-the acoustic response is what matters.
formfactor's Avatar formfactor
12:19 AM Liked: 10
post #22 of 48
04-19-2008 | Posts: 72
Joined: May 2007
Sounds good. Any books you'd recommend in particular?
augerpro's Avatar augerpro
04:45 PM Liked: 29
post #23 of 48
04-20-2008 | Posts: 1,715
Joined: Aug 2005
Speaker Building 201 & Loudspeaker Design Cookbook to start with. Master Handbook of Acoustics if you finish the other two. High Performance Loudspeakers is alright too.

Interesting stuff here: Geddes also has a book that I have to check out.
KyleLee's Avatar KyleLee
06:21 PM Liked: 12
post #24 of 48
04-20-2008 | Posts: 1,129
Joined: Feb 2007
Things for future builds: Very nice passive xovers btw, so dont take this personally when I suggest to consider active eq. You can do a lot more with active filters and you can make adjustments and fine tune as you go. Also very important: measure the off axis response!!!! That can tell you alot about your xover point, either active or passive. off axis response is critical in acoustics!

here is also a little tip that no one ever talks about... with active EQ, you can possible account for this with more elegance, i'm a programmer so i will spare you of the algorithmic aspects of dynamic EQ, but even if you're using a fixed analog passive filter(s), consider NOT using the TSP's given by the mfr, but rather accounting for a new set of TSP's when the drivers are at their average DC resistance. This means when they are warm. TSP's can shift dramatically with just a few minutes of use and its really unfortunate so much effort goes into measuring the speaker when its cold when in fact, we never listen to a cold speaker. Try upping the DCR a bit on your TSP's and see how that shifts everything in the frequency domain, it could make a worthy difference.

Orcad software... brings back horrible memories of my ECE labs, yuck!
Jay_WJ's Avatar Jay_WJ
07:15 PM Liked: 10
post #25 of 48
04-20-2008 | Posts: 841
Joined: Jan 2006
There's an alternative simulation method that considers both the FR and impedance of drivers, and results in reasonably accurate system responses when used properly. Take a look at these web pages:

shujin's Avatar shujin
07:44 PM Liked: 10
post #26 of 48
04-20-2008 | Posts: 57
Joined: Nov 2006
Oh got to love pspice.
KyleLee's Avatar KyleLee
08:04 PM Liked: 12
post #27 of 48
04-20-2008 | Posts: 1,129
Joined: Feb 2007
...better than Xilinx tho
formfactor's Avatar formfactor
11:37 AM Liked: 10
post #28 of 48
04-22-2008 | Posts: 72
Joined: May 2007
Thanks for the guys.

I'll check those books out auger. I've actually read through some of those sites before. Should be interesting.

Kyle, I'll try out electronic crossovers eventually. Just waiting for some money to spare. I'm trying to pace myself since I've already spent $400 on a table saw and router on top of the cost of the speaker parts.

Jay, those sites look like they do something similar to an FTABLE for pspice. I'll have to read through them to see what they actually do though. Thanks for the heads up.

I did some real measurements over the weekend and it looks like the response is very good. I'll have to make more off axis and far field measurements but I'm happy so far. I'm already thinking of putting in some baffle step compensation.

Here's one of the near field measurements I took:

augerpro's Avatar augerpro
12:33 PM Liked: 29
post #29 of 48
04-22-2008 | Posts: 1,715
Joined: Aug 2005
Those measurements look a little better. What was the mic distance? The port resonances look a bit scary though. Can you make a measurement with both the woofer and tweeter at the same time? Currently you can't determine how well they sum in the crossover region if you take each driver's measurements individually.
formfactor's Avatar formfactor
01:12 AM Liked: 10
post #30 of 48
04-23-2008 | Posts: 72
Joined: May 2007
The woofer's measurement was taken 1/4 inch away from the dust cap and the tweeter's was taken roughly the same distance away from its dome. The port response was measured with the mic (ECM8000) approximately at the entrance of the port but not inside of it.

The port resonances are probably due to the backwave of the woofer. Would more dampening fix that?

And how should the combined measurement be taken? 1m away on axis with the tweeter and somewhere outside my house? Would a phase measurement be useful? I have options for wrapped, unwrapped, minimum, and excess phase measurements.

Reply DIY Speakers and Subs

Subscribe to this Thread

Powered by vBadvanced CMPS v3.2.3