Acoustic Offset, WinPCD vs. XSim

[wogg]

The "Play with design software..." thread showed me the Z axis offset settings in XSim that was unaware of, so I tried to replicate a design I've been playing with in both programs to see how they compare. This is a 3 way with DCS165-4 woofer, RS52AN mid, and a ND20FB high. Both simulations are using the same FRD and ZMA files run through response modeler for the intended cabinet. The crossover values in PCD looked pretty good, decent but not awesome reverse nulls and flat response. The response was near identical in XSim, until I applied the offset delays, then it went a bit wonky. I noticed the "Polar Active" setting in WinPCD makes the curves far more similar with the same offset settings in both programs, are the Z offsets ignored until you activate the polar response?

The PCD image has an overlay without the polar active, while the black plot and phase is with polar active. The XSim plot is with the Z offsets. Note the response is similar but not exact, and the most obvious differences are in the system phase plot.





Follow up question... How do you guys typically model for phase tracking when designing on paper? It's obvious that Z offset has a huge effect on phase tracking, particularly at the higher frequency / shorter wavelength transition to the tweeter.
If using WinPCD, should I tweak with Polar Active, or should I use XSim to tweak?

[dlr]

The response was near identical in XSim, until I applied the offset delays, then it went a bit wonky. I noticed the "Polar Active" setting in WinPCD makes the curves far more similar with the same offset settings in both programs, are the Z offsets ignored until you activate the polar response?

That is correct. If you are using actual measured responses, that is with excess-phase, then you should not use polar settings as these are intended to be used only with minimum-phase measurements so that the x,y,z spatial settings can provide the appropriate delay for each driver in order to be able to examine off-axis points.

The PCD image has an overlay without the polar active, while the black plot and phase is with polar active. The XSim plot is with the Z offsets. Note the response is similar but not exact, and the most obvious differences are in the system phase plot.

You should be able to get a nearly identical summed response from the two programs, assuming both implement the calculations correctly. I'm not aware of problems with WinPCD, but I never discount that someone may find a problem that has remained undetected.

If using WinPCD, should I tweak with Polar Active, or should I use XSim to tweak?

In WinPCD the issue with tweaking is one of using the correct method which that is determined by the measurement files used, either minimum-phase (HBT generated) or excess-phase (direct measurement files). I can't speak to the case of XSim. Which one you choose to use, though, should come down to preference.

dlr

[wogg]

Thanks for the insight! I am using min phase extracted files, so I'm thinking the polar active mode should be accurate. I need to adjust to line the phases back up. I think PCD is a little easier since I can have the system response and schematic overlay up on screen and see in real time the effect of component changes. XSim takes a couple clicks in and out of the tune dialog for the same effect.

[jhollander]

With simulated min phase files and you can take your sumed response from win PCD (with the x, y, and z drive locations) into XSim as an overlay and then adjust the woofer offset so the curves match.

I've also used the right triangle calculation (pythagorean theorem) to get an a offset distance between the tweeter and the woofer in Xsim

[ErikSquires]

I'm so glad to see that my thread is yielding even more fruit! Glad to see the discussion happening.

jhollander, what do you mean about right triangle method? I don't know it, but if you are just measuring the C to center distance, won't you miss the correct depth?

Best,


Erik

[jhollander]

As you know XSim looks for the offset as a number of inches. Assuming you are designing on the tweeter axis, the delay in inches is added to the woofer. The distance difference between tweeter and woofer can be roughly calculated.

Distance from mic to tweeter is 36 inches; distance from center of tweeter to center of woofer is 10 inches. Distance from mic to woofer is square root of (10 x 10 + 36 x 36), equals 37.63 inches Difference is 37.63-36 = 1.63 inches add a rough z offset for the woofer of say .75 inches, use a total delay for the woofer of 2.38 inches.

Of course this is just for simulations, if you can measure you should to the overlay the tweeter plus woofer response with the tweeter paralleled with the woofer summed response and calculate the correct offset.

[wogg]

I see what you're saying. Instead of doing just the Z axis offset, use the full 36" mic point and calculate the actual distance to the driver points. That could make a big difference. I ended up tweaking the crossover to get it to transition flat mid to tweeter in PCD, using the polar active setting. That actually ended up making the match between PCD and XSim worse.... I'll keep playing. Playing is free!

[wogg]

As you know XSim looks for the offset as a number of inches. Assuming you are designing on the tweeter axis, the delay in inches is added to the woofer. The distance difference between tweeter and woofer can be roughly calculated.

Distance from mic to tweeter is 36 inches; distance from center of tweeter to center of woofer is 10 inches. Distance from mic to woofer is square root of (10 x 10 + 36 x 36), equals 37.63 inches Difference is 37.63-36 = 1.63 inches add a rough z offset for the woofer of say .75 inches, use a total delay for the woofer of 2.38 inches.

Of course this is just for simulations, if you can measure you should to the overlay the tweeter plus woofer response with the tweeter paralleled with the woofer summed response and calculate the correct offset.

That was exactly it John, thanks!

I calculated relative offsets using the dimensions I had in WinPCD, applied those in XSim and the result matches exactly. Pulling in the curve from WinCPD helped confirm.