Crossover Recommendations

Passive crossovers deserve a section to itself. It is part science and part art.
Post Reply
User avatar
Chad Gray
Site Admin
Posts: 46
Joined: Wed Dec 02, 2020 7:26 pm
Contact:

Crossover Recommendations

Post by Chad Gray »

Hello, i don't want to mess this up.

I seem to have found a really nice pair of drivers. This is an Usher 8945A 7" and a ScanSpeak R2604/8320. This is the raw responses and the woofer has diffraction added in to the Near measurment.

They appear to match up really well! Does anyone have recommendations on the type of crossover to put on each? They are soo close in SPL level! I want to add a minimal amount of parts. A crossover point of 2500-3000 seems to be a good spot

I attached the project and driver files if anyone wants to play with it.
Usher SS.zip
(55.37 KiB) Downloaded 35 times
PA.jpg
PA.jpg (293.67 KiB) Viewed 606 times
User avatar
Chad Gray
Site Admin
Posts: 46
Joined: Wed Dec 02, 2020 7:26 pm
Contact:

Re: Crossover Recommendations

Post by Chad Gray »

Here is my first crack at it. I let the tweeter rise instead of trying to bring it down.

Not sure if i should let the tweeter get full signal though.
PA2.jpg
PA2.jpg (562.12 KiB) Viewed 598 times
dcibel
Posts: 43
Joined: Sun Dec 13, 2020 6:39 am

Re: Crossover Recommendations

Post by dcibel »

Oh dear, please do not run a tweeter without a filter. Even the cheapest speakers made have a single capacitor on the tweeter and let the woofer run full range. Even with that, what you've shown above is a bright speaker, 5dB rise from 100Hz to 10kHz.

Once you bring the woofer down to compensate for BSC, you'll find you have plenty of headroom on the tweeter. Use the SPL at 100Hz as your baseline, in this case around 85dB, so aim for flatness along 85dB to start. The drivers should be easy to work with, start with 2nd order filter on each and simple L-pad on the tweeter and go from there. I'd recommend trying to cross in the 2kHz range.

I recommend using CAD frequency domain for determining crossover slope only. Build the filter schematic here and use the optimizer to hit your target. Then move on to the system module so that the physical location on the baffle is considered, then use the system frequency domain module to optimize for overall response. Follow John K's guide for reference.

As always, its important to be certain the measured data is accurate, any error in the measured data, splicing, diffraction, etc will result in errors in the simulated result. I have suspicion that you've spliced a nearfield response at 300Hz without applying diffraction sim to it first.
User avatar
Chad Gray
Site Admin
Posts: 46
Joined: Wed Dec 02, 2020 7:26 pm
Contact:

Re: Crossover Recommendations

Post by Chad Gray »

I am going to re-measure everything in a different room today just to make sure my measurements are good. I do question the data in my near field measurement. It did seemed a little strange.

I am following John K's guide and i did add the diffraction to the near field response. I will look up what you mean about bringing the woofer down to compensate for baffle step compensation. I do not remember anything about that. I took far, near then added simulated diffraction to near and combined them at 300 Hz. This time i will try to remember to save the IR's so i can always redo my steps.

This time I will start with an Lpad on the tweeter and a 2nd order on each driver. I did not have the lpad the first time.

As always thanks for your time and input! It has been great re-learning all of this. I hope everyone can learn from my mistakes and questions.

I probably wont have time today, but i would like to do a live stream with Zoom video and chat to invite people in to a measurement session. People can watch post/ask questions and contribute to my errors and successes.

Chad
dcibel
Posts: 43
Joined: Sun Dec 13, 2020 6:39 am

Re: Crossover Recommendations

Post by dcibel »

The flag raised is simply that the slope of the response of the nearfield does not match that of the far field measurement. The responses should overlap when you splice them - I should not be able to tell where the splice occurs so obviously if done correctly.

Here's a quick and dirty example of Wavecor WF152.
2020-12-27 19_08_45-SoundEasy V25.0 - C__Users_Reid_SynologyDrive_speaker files_G8 WF152 + Motus WG_.png
2020-12-27 19_08_45-SoundEasy V25.0 - C__Users_Reid_SynologyDrive_speaker files_G8 WF152 + Motus WG_.png (38.05 KiB) Viewed 580 times
If I were to ask you where this response was spliced, you should have a hard time as there was sufficient overlap between the responses, so this is a fairly accurate representation of the anechoic chamber response.

How did we get here? The same process as outlined in most tutorials, including John K's SE Design Guide.
1. Far field is gated to the first reflection.
2. Near field is not gated.
3. Diffraction is added to near field response
4. Near field with diffraction applied is adjusted in amplitude to match up with far field.
5. Near field and far field are spliced and saved to driver file.

The following screenshot shows the far field, near field, and diffraction adjusted nearfield overlapped. (Steps 1,2,3)
2020-12-27 18_59_35-SoundEasy V25.0 - C__Users_Reid_SynologyDrive_speaker files_G8 WF152 + Motus WG_.png
2020-12-27 18_59_35-SoundEasy V25.0 - C__Users_Reid_SynologyDrive_speaker files_G8 WF152 + Motus WG_.png (46.24 KiB) Viewed 580 times
Here we've adjusted the amplitude of the neafield with diffraction to match up with the far field, and spliced together at 450Hz.(Steps 4,5)
2020-12-27 19_02_23-SoundEasy V25.0 - C__Users_Reid_SynologyDrive_speaker files_G8 WF152 + Motus WG_.png
2020-12-27 19_02_23-SoundEasy V25.0 - C__Users_Reid_SynologyDrive_speaker files_G8 WF152 + Motus WG_.png (43.92 KiB) Viewed 580 times
The result is what you see in the first screenshot above. For more detail I would suggest starting at John K's design guide,

All I mean on the baffle step is to apply the compensation. At low frequency the speaker operates in 4pi space which relates to 6dB of loss vs the high frequencies that operate into 2pi space, so the SPL at 100-150Hz range becomes the baseline maximum SPL for a passive design, everything else above this point should be attenuated to meet this SPL if your aim is for a flat result. In your first screenshot, that ends up being about 85dB for what you've shown. 5dB of rise from there is going to end up bright and thin.
Post Reply