I'd be looking for criticism of this technique...
A crossover is designed using the acoustic phase of the drivers measured on the listening axis, but the response associated with each driver is swapped as needed between its response on the listening axis, and the magnitude of the driver power response calculated from measured SPLs. In this case the speaker can be designed for flat power, but the power of the drivers will sum incorrectly.
So, making the following assumptions: an in-phase style crossover with 6dB summing is used, the power will be down 3dB compared with the SPL in the middle of the crossover, and reasonable phase matching is achieved in the design before continuing.
Then shift one driver to force quadrature between them and achieve quasi power summing?
A crossover is designed using the acoustic phase of the drivers measured on the listening axis, but the response associated with each driver is swapped as needed between its response on the listening axis, and the magnitude of the driver power response calculated from measured SPLs. In this case the speaker can be designed for flat power, but the power of the drivers will sum incorrectly.
So, making the following assumptions: an in-phase style crossover with 6dB summing is used, the power will be down 3dB compared with the SPL in the middle of the crossover, and reasonable phase matching is achieved in the design before continuing.
Then shift one driver to force quadrature between them and achieve quasi power summing?
I believe it is true, if the drivers are not close together, that an artificially forced quadrature summing would represent the power response. If they are close together then the random summing that 90 degrees represents may not occur.
There was a thread a while back where I helped a guy who was plotting constant in phase and constant out of phase summing to plot the envelope of the extremes of off axis response variation. You could add the 90 degree sum to the mix.
David
There was a thread a while back where I helped a guy who was plotting constant in phase and constant out of phase summing to plot the envelope of the extremes of off axis response variation. You could add the 90 degree sum to the mix.
David
Yep, the main plot shows sim'd power response overlay.
Boxsim as well, I think.
Both are free.
Taking a look at this it seems to be able to show the equivalent of power from an axial slice. Can this be used to create a full radiation power plot?
I've haven't used Boxsim. Often freeware from a commercial source is limited, looks like I should try this though.
Do you mean the response is not always proportional to the voltage seen at the terminals?
I'm not in a position to predict this with accuracy. If a waveguided HF driver's primary wave acts as a point source, its other modes both distributed and displaced, and a piston woofer.
I'm just trying to predict the directivity index around a crossover without resorting to further measurement, or coding which I'm only partially confident of the result of. Like the example screenshot below (DI at the bottom), the crossover is between 1-2k. If the crossover is pushed higher with the same flat power (ignore the common peak at 1k), then DI at the crossover region would decrease but the hole at 2k would fill.
At this level though I need better resolution than a guess. One other technique I was using was to create target Butterworth responses on the power magnitudes, match phase throughout, export this and mow the peak flat in a text editor.
Attachments
A 6dB drop in SPL should relate to a 3dB drop in power where radiation gain was a factor.
I verified that using a power plot in a crossover sim can work by recalculating the driver(s) power response after applying the transfer function of a sample crossover to each axis and comparing the result to the original calculated power response. The only reason I feel it would be valid is because the axial responses used for the power calc don't coincide and radiation gain wouldn't apply.
[I'm labouring under a few quiet refreshments this evening, there's a question coming, I promise
]
So where the drivers are crossing, radiation gain may apply and so I feel I should extract the "filtered" powers and add them as powers rather than let the sim deal with it?
Is it also valid to survey the lobing cancellations and convert that to an across the board, crossover region, power response modifier?
@Sreten, I'm seeing your point now.
I verified that using a power plot in a crossover sim can work by recalculating the driver(s) power response after applying the transfer function of a sample crossover to each axis and comparing the result to the original calculated power response. The only reason I feel it would be valid is because the axial responses used for the power calc don't coincide and radiation gain wouldn't apply.
...and maybe if the sources are very close and in phase? (point being no lobing issues either very close or very far?)
[I'm labouring under a few quiet refreshments this evening, there's a question coming, I promise
]So where the drivers are crossing, radiation gain may apply and so I feel I should extract the "filtered" powers and add them as powers rather than let the sim deal with it?
Is it also valid to survey the lobing cancellations and convert that to an across the board, crossover region, power response modifier?
@Sreten, I'm seeing your point now.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.