Considerations for designing a speaker cabinet with sim software (4 ohm Purifi 6.5 woofer target)

[Rick Sykora]

What did you assume for in line resistance and which TS params did you assume? No way either program is this far apart unless the driver or design assumptions differ. I listed mine in my previous post. How did your assumptions differ?

I pulled the T/S params from the Purifi website recently. Pretty sure they are the same as your posted v1.10, but IME, never hurts for another set of eyes to review as I added the driver manually...

For inline resistance, here are the assumptions...

Here are the box settings...

NOTE: Bassbox does not allow you to manipulate Vb and Qtc independently. Here I chose to match your Vb, if I set Qtc instead, it lowers Vb to about 3.9 liters and f3 becomes 96 Hz.

This was done with the damping set to Heavy. This corresponds to a Qa of 5.

Full output of sim is attached.

 

[DDF]

I pulled the T/S params from the Purifi website recently.....

We used the same TS params and Unibox and bassboxpro give largely the same box alignment results. Here's Unibox:

Note that Unibox's sensitivity and efficiency numbers include the external resistor, bassboxpros don't but when I remove the resistor, they're identical.

I found the two differences:
1. Bassboxpro "heavy fill" assumes a higher Qa (less box stuffing damping) than Unibox. I estimated this by comparing the bassbox roll off at 20Hz vs 100Hz (21dB) vs Unibox's (19 dB). But this won't make that big a difference

2. The real difference is in how F3 is defined. The way to spec an alignment F3 is half space, per Unibbox (AES papers, LDC etc). Half space has no diffraction (which gets eqed out by the crossover). So you need to somehow remove that wierd high pass shown in Bassboxpro when estimating F3. When I do that, they're the almost the same normalized to 300Hz

But what is that bizarre 15 dB high pass rise in bassbox pro? Whatever it is, ignore it or remove it somehow in the your sims because it's not real. It can't be diffraction, here's an accurate sim of that:

 

[Rick Sykora]

@DDF Looking into how Vb is determined for a closed box (from Loudspeaker Design Cookbook, 7th edition):

Vb = Vas/Alpha where Alpha = (Qtc/Qts)^2 - 1

Since Vas, and Qts are inputs from the speaker parameters, you can either set Qtc and calculate Vb or vice versa.

Unibox looks as though it takes the approach of choosing a Qtc and calculating Vb for you. Anything else appears to be driving off road and not sure the results can be trusted.

 

[DDF]

@DDF Unibox looks as though it takes the approach of choosing a Qtc and calculating Vb for you. Anything else appears to be driving off road and not sure the results can be trusted.

Na, Unibox calculates Qtc from your given box volume. Its dead on accurate if Ql and Qa are right. Please see my prior post, it explains the differences. Mystery solved.

 

[Rick Sykora]

Na, Unibox calculates Qtc from your given box volume. Its dead on accurate if Ql and Qa are right. Please see my prior post, it explains the differences. Mystery solved.

Sent you detail on a PM, but the shelving shown on the Bassbox graph is just one of a set of graphing options (BSC, etc.).

None of the options, affect the calculated f3.

We are not there yet...

 

[Rick Sykora]

Just in case others might be interested, @DDF and I have been working on this offline. He has a personal family matter to address, so thought I would share our progress. Using different approaches, we both are able to align Bassbox and Unibox for a basic box calc. This resulted in a 3.5 liter box with an f3 around 87 Hz. The difference in the models appears to have been the default for QL. Bassbox uses value of 20 and it appears Unibox uses a value of 7.

As I start to add back inline resistance, more differences appear. Unibox‘s model lowers f3 much more than Bassbox for the same resistance. The greater difference is in the box damping. With a roughly 6 liter tightly sealed box, Unibox manages an f3 in the upper 60s. Under the same conditions, Bassbox will not produce a result. The most it will do is increase f3 when you would expect a decrease. I have seen a more sensible result from Bassbox when the driver EBP indicates the driver is suited for a closed box, but seems to not like to fully model a closed box for a driver that is more suited for vented box.

At this point, if anyone has modeled this Purifi woofer and can replicate DDF’s design, would be useful to know. I trust his judgement, but Unibox is old enough that would help to have independent verification of its results. Thanks!

 

[MZKM]

 

[ctrl]

This resulted in a 3.5 liter box with an f3 around 87 Hz. The difference in the models appears to have been the default for QL. Bassbox uses value of 20 and it appears Unibox uses a value of 7.

Compared your specifications with the software I use.

Come to slightly different values f3 is about 80Hz with a Qtc of 0.72 and the assumption that much damping material is used, without series resistor of 0.7 Ohm. With a series resistor of 0.7 ohms a slight hump results.

Due to the assumed dense damping of the inner volume I calculated with 22% virtual volume increase. If the inner volume of the cabinet is less tightly stuffed with damping material, the virtual volume is reduced accordingly.

With VituixCAD I get about the same result. With 0.7Ohm series resistor f3 is about 77Hz and a slight hump.



In order to find out how well the used simulation software approaches reality, its results should be compared with a real example.

I used the SB17NAC35-4 as midrange driver in a project. The gross internal volume of the speaker enclosure was 3.5L with tight damping of the inner volume.
For the simulation I would not use the manufacturer TSP, but the one from a review magazine.

The near field measurement of the woofer showed the following (near field measurement - green, target function Q 1.05 - red, measurement with baffle step correction 0.3x1m - yellow):

The resonance frequency was 96Hz and Qtc about 1.05.

If the simulation software gives roughly the same result, the simulation for the Purifi chassis should also be correct.

 

[Rick Sykora]

@ctrl thanks for sharing! As usual, you make some great points. The sim is just an approximation but is useful for helping determine whether to commit funds to a purchase and build a box. Still need to get the real driver and try it out.

Noticed you used a QL of 100. Implies a tight box. Did you choose or is that sim software default?

 

[ctrl]

Noticed you used a QL of 100. Implies a tight box. Did you choose or is that sim software default?

This is the default in VituixCAD, but can be changed.


Here is a second example. WF118WA-02 in seal box as midrange driver in 1.9L speaker enclosure with dense damping.
Measurement, nearfield, no baffle step. Resonance frequency was 116Hz, Qtc 0.85.


If your simulation program does not approximately reach the values of the real measurement, it is crap useless
...and if it does, you can use the settings as a blueprint for the Purifi chassis.

The tool I mainly use comes pretty close to the real measurement for the second example. Simulated with 1.9L (1.9L gross volume - 0.3L chassis plus 22% virtual volume increase because of the dense damping --> 1.9L), the resonance frequency hits the measured frequency exactly.

The Qtc is with 0.785 too low, but it was also simulated without series resistor. If you take 0.3Ohm series resistor as a basis for the measurement (6m loudspeaker cable), the result is an Otc of >=0.8, which is okay.



Used TSP from magazin:


UPDATE:
The volume increase of 22% I assumed (in both simulations) is very much and only applies to CB when the inner volume is completely filled with damping material. This is the case in both examples.
The value itself is a pure empirical value of mine, so it is not generally valid, like the number 42

 

[bennybbbx]

the phase look realy good for 6,5 Zoll speaker .what should it cost ?. I meausure my speakers and they are much worser. the best in phase is the 10 zoll Celestion guitar speaker but this is no bass. it begin at 100 hz. i see not much phase measure. I do also in this thread https://www.audiosciencereview.com/...li-lp-6-defect-canton-look-much-better.17854/ screenshots so can see the transient in 1 cycle. the speakers with the better phase i see in my tests always have better transients and have more stereo width. but there should more speakers have phase measure or do test transients . my measure seem correct because 9 euro headphone have good phase and transients.

 

[Chippy]

Hi all,

one has to be carefull when comparing excursion within two alignments and/or enclosures. You have to compare them for same frequency response, otherwise you get a total wrong picture. I will show you an example.

But first I have question for DDF:

The text says it is tuned to 34Hz, but the excurion says it's tuned more like 23Hz? Whatbis wrong here?

Here is a better comparison, which does not comapre apples and peaches. First you see left the SPL and on the right the excursion. Then we have three sets of curves, the closed box with 6l, then the BR 18l tuned to 34Hz but also the BR box but EQ'ed to have the very same SPL response as the CB box has.

You clearly see the advantgaes of the BR in terms of excursion, although one should use another, more optimized tuning when comparing it to the CB enclosure, but I just wanted to compare the enclosures discussed before.

Rock on
Chippy

 

[Rick Sykora]

After checking with BassBox support, I summarize their response as follows:

1. It appears that Unibox ignores gamma (param for different types of damping material), so use Bassbox Classical model (for equivalency).
2. Bassbox's classical model is more accurate for this driver. Given this...
3. The inline resistance lowers the normalized response curve about 1.24 dB. Since Bassbox calculates f3 from 0 Db, you get an f3 of about 80 Hz. If you follow the normalized response to -4.24 db, you get to about 70 Hz (matching Unibox).

As for my takeaway, the results still follow a typical lower Q 6" woofer. You can get extension of f3 to about 50 Hz vented or 70 Hz in a smaller sealed box. As I posted in the review thread, Purifi could have kept the diagonal brace and used a simpler port by swapping the port location with the input jack. This also avoids the wiring draping over the port on the outside of the cabinet. Ofc the transient response is not as good as the sealed box, but likely can do without subwoofer(s) too. There may be other good arguments in favor of the sealed cabinet (aside from size), but will have to wait for @DDF to return.

As a final note in defense of Purifi, complex port aside, a vented box seems the best compromise for their demo box. You get the low end extension without the additional spend for their passive radiator. As we have seen in other reviews, a passive radiator can have resonance issues too.

 

[DDF]

@amirm doesn't measure speaker SINAD, but he could. It would be a meaningful advance in public audio testing, providing a single easy to understand metric vs the difficult (for most) to interpret near field port measure.

An example related test is the Incoherency Distortion method in MLSSA here. It measures all extraneous sonic junk from rub/buzz to vent noise. Calculating "Incoherency SINAD" wouldn't be much more difficult than calculating a traditional SNR. A number of other test options exist but this is the one I use the most.

Why is this important? It seems at cross purposes to add a measurably coloured component (a vented speaker) with an undeclared and compromised SINAD of 10 to 30dB (over a constrained frequency range, dominated by port noise) to a system carefully assembled for >100dB SINAD elsewhere.

Port noise is noise. Its effect is largely audible as illustrated by the body of @amirm tests showing ragged mids with cause proven in the near field port measures.

Getting back to sealed vs vented discussion, port noise applies much more equally at low, and normal sound levels (worsening with chuffing) than does distortion. Sealed typically has notably better SINAD over the mid range at normal listening levels (I see 60 dB in modest designs with well chosen drivers).

Its possible, but not typical (more like the exception), to avoid the measurable effects of port noise in a vented.