First DIY speaker project thread! - Page 2

And about the Z, their tiny difference on Zaph's plots may be just a sampling variation. The Re of P18 measured by Zaph is 5.9 Ohms and the Re of CA18 is 6.0 Ohms; but on the Sea's table, now the P18 has Re of 6.1 Ohms and the CA18 has Re of 5.8 Ohms. And as I said, their voice coil inductances are virtually the same (1.18 mH vs 1.2 mH). But again the old spec sheet of the CA18 on Madisound tells it's 1.1 mH. Looks like we only see unit to unit variations of the same driver.

Good choice. Considering the quality of drivers and design, you'll be certainly impressed. As for the XO design, LR2 design (Likwitz-Riley 2nd order; basically means a large overlap between two drivers' responses for a more coherent blend of drivers) is rare and if it's implemented well, it tells a lot: real high-end! If Zaph did it, you can be sure that he did it well.

OP: Original Poster --- it means you in this thread.
SW: Speaker Workshop --- a popular free software for speaker measurement and design.
Z: Impedance measured in Ohms --- speaker impedance changes across frequency; so a plot of impedance vs frequency is used to see the impedance characteristic of a speaker.

People on a budget usually use Eagle metal oxide from Madisound or Dayton non-inductive from Parts Express. Since you'll buy from Madisound, order Eagle resistors. Some people use more expensive ones like Mills, but many people think that Eagle or Dayton is more than adequate even for a high end design.

Use a single layer Sonic Barrier or any kind of sound absorption foam you can find cheap. This will work only for sound absorption but not be effective for damping cabinet resonance. Instead of using 3 layer Sonic Barrier or other expensive damping material, how about adding another brace or two?

People usually fill heavily in a sealed design, but not for a vented design. Nevertheless, light filling could be helpful.

sphynx,

I know you alreay ordered crossover parts. But do you still have a chance to change some items in the order? I just looked at Zaph's XO design carefully and found he used an LCR notch filter in the woofer network, which is to attenuate a bump in the 1 kHz - 2 kHz range of the CA18's response (which is somewhat exaggerated as well by diffraction effect). But looking at both Zaph's measurements and manufacturer's spec sheets, I found P18 behaves a bit differently in this regard. The bump is less severe and occurs towards 2 kHz rather than towards 1 kHz. Sorry that I noticed it a bit late.

I simulated an XO for P18 in my Speaker Workshop and determined notch filter component values customized for P18. The LCR notch filter consists of C3, L4, and R5 in Zaph's crossover schematic. Current values are 6.8 uF, 2.0 mH, and 7 Ohms. Change these to 6.2 uF, 1.5 mH, and 10 Ohms, respectively. For a 1.5 mH inductor, use Madisound standard 19 ga. Don't worry about the internal resistance (DCR) of the inductor. Just use a 10 Ohm resistor for R5.

Jay can you post that sim? How did you model diffraction? Does the simmed diffraction (using the CA18) agree with Zaph's measured response?

I'll just show the results of raw driver simulation on the baffle. I used FRDC tools---UniBox, BDS and FRC spreadsheets---to simulate box and baffle diffraction effects. Here are the results (P18's respons is lowered by 10 dB):



And Zaph's measurement of CA18's in-box response is (shown in red below):




Just look at the midrange and ignore bass and treble responses in my results. I did not pay careful attention to these ranges. Also, note that this sim can't take into account a baffle cavity effect and the woofer's off-axis response, which should be apparent in Zaph's actual measurement. Considering this, the sim result is quite similar to the actual one. And I found that the notch Zaph used is centered at 1350 Hz, which seems appropriate. I moved the notch a bit to the higher frequency side (1500 Hz) and made the notch depth a bit shallower to accommodate the P18 substitution. Seem reasonable?

Of course, this will only be an approximation because we have no actual measurement. But considering the usual accuracy of Zaph's driver measurement, this approximation will work fine, IMO.

sphynx,

In my post #39 above, the suggested L4 value should be 1.75 mH, not 1.5 mH. I tweaked the value several times, so being confused I made a mistake when I posted it. I checked it again to confirm the correct value of 1.75 mH. To recap, the C3, L4, and R5 values I got from my sim are 6.2 uF, 1.75 mH, and 10 Ohms, respectively.

Yeah. I would definately keep it shallow (higher resistor), but highish Q. Wouldn't want to overshoot and cause a big dip because of the cavity effect from 1600-3500 hz.

How did this change effect phase integration? Can you post the summed system model?

No, I don't have a system model. I simply traced their SPLs and modeled box and diffraction effects---was too lazy to trace impedance plots. The notch component values were obtained from a sim using a net transfer function and impedance of another driver I modeled before, which I thought similar enough to CA18 and P18.

I did a lot of XO simulations using SW. According to my experience, application of a shallow notch doesn't change acoustic phase responses significantly even if the notch is within an octave distance from the XO point. Anyway, the purpose of modifying the notch here is to make the resulting response of P18 close to that of CA18. Right? So, I don't think there'll much change to phase alignment.

1/2" Sonic Barrier all around will be fine with extra bracing. 100% coverage won't be necessary, but don't skip corners, back, top and bottom. Also, using two layers of foam at top and bottom of each chamber will be a good idea.

BTW, you're going to use the P18RNX for both top and bottom woofers, aren't you? I think you can still use the cabinet size as Zaph designed. Although it's not optimal for P18, it will work fine. But try different port tuning frequencies to find a suitable one for you. According to my calculation, a 7 cm port length (of PE port 260-402) will give you a tuning frequency of about 49 Hz and will provide balanced bass. But try a different one just in case.

- Jay

The most inexpensive way may be to use 1/2" carpet padding all around and fill lightly with polyfill. Zaph also used carpet padding to only sides like this:




But he did not use it at top and bottom. Read Zaph's writeup about damping materials: http://www.zaphaudio.com/whispermat.html

BTW, did you already build the cabinet? Or you will?

Somehow I thought you already built the cabinets. If not, consider reducing the cabinet volume to obtain a better volume for P18's. My suggestion is to shorten the cabinet's internal depth from the current 10.75 inches to 8.5 inches. This will change the volume of each chamber from 19.5 liters to about 15 liters. If you use 3/4" MDF, external cabinet depth will reduce to 10 inches.

As for port tuning, try two port lengths, 85 mm and 75 mm. Due to reduced volume, these values will give you a bit higher tuning frequencies than the original design's, but these are suitable for P18 woofer. As Zaph suggested, experiment it using an extra port. Using your favorite music, you can decide on one that suits your taste and your room.

Sphynx,

To make it sure, I modeled P18's response that would be produced by the modified XO. For comparison I did the same thing with CA18 using the original Perfectionist XO. Compare the response curves expecially in the range around the XO point (i.e., 1 kHz to 4 kHz)



What do you think? This kind of modeling, of course, can't be perfect for predicting a single design with either CA18 or P18. But the comparison of two drivers' behaviors in the designs is a different story. It makes much more sense because 1) two drivers were measured in an identical condition (on Zaph's infinite baffle) and 2) they will share exactly the same baffle diffraction effect. What I mean is that the modeled curve itself can deviate from an actual response due to prediction errors, but the RElative DIfference of two curves can be predicted much more accurately.

It is entirely possible that you may end up disappointed with the sound of your speakers, simply because every speaker has its own sound characteristics that you may or may not like. BUT you would be in the same situation even if you built them with CA18's. I believe that the above modeling result supports my reasoning.

One more thing to add. Compare the nonlinear distortion characteristics of the CA18 and P18 (CA18 on top of P18):






Again, very similar to each other. I really want to find some reason(s) that the design with CA18 replaced by P18 would sound very different from the original design. But I can't find any.

I assume you have a good soldering skill. If this is your first speaker XO build, read:

http://www.partsexpress.com/resource...-crossover.cfm

And if you think there'll be any possiblity of moving speakers in the future, don't just hot glue XO components. Cable tie big components in addition to hot gluing them.

is there an even more basic thread than this for a first time DIY speaker build? I have a good grasp of all the A/V stuff, but #1 have not done much woodworking, #2 don't know much about drivers, crossovers, etc, but I do have a lot of soldering experience and electronics experience. Say if I wanted to build a two fronts, a center, and four surrounds. A couple questions I would have (that may sound noobish in this sense) would be:
1. What is the advantage to having say, 8 drivers in the fronts vs 2-3?
2. How is the wiring different for 1-2 drivers and a tweeter, vs the elaborate setups?
3. Is a crossover simply to protect the tweeter from lower frequencies?
4. Can I simply have a speaker inside a box and solder up the +/- ?

They are inductors. If they are positioned in a certain way, their values of inductance are changed. You should avoid this.

There is a part about inductor placement in the article I posted above.

thanks for the direction, sphynx