Newb...2-way bookshelf X-Over design... 15W/8530K00 & Morel ST1108 Supreme

[Jopldangla]

I am working on a x-over design for the Revel105 copy, however I am using the Scanspeak 15W/8530K00 and Morel ST1108 supreme tweeters. I am going to mount the Morels in a waveguide as well, by using some of the guidelines from the huge waveguide thread I found on the internet somewhere

Supreme ST1108 datasheet
15W/8530K00 datasheet

I used tracing tools to create ZMA and FRD files from the datasheets, these files were imported both into the VituixCAD software and Passive Crossover Designer 8.0
I will use roughly the same cabinet design and port sizing.

Here are my results from both tools....I am having trouble figuring out why they don't match. I used the same crossover values from the Revel105 copy build for the scanspeak and matched the crossover frequency for the tweeter but used a higher order filter. I also needed to add an L-Pad to drop the sensitivity of the Morel to match closer to the Scanspeak.

I really appreciate any tips from more experienced folks. I haven't been able to find much information on the Morel ST1108 tweeter online (maybe nobody is using it).

Passive designer followed by VituixCAD

 

[alex-z]

They match better than you think. VituixCAD is using 40dB scale by default, PCD is 70dB scale.

Copying the datasheet info is doing nothing for you. The drivers will behave much differently in a real cabinet, especially the tweeter once you add a waveguide. The goal of a speaker is a combination of linear on-axis response and consistent radiation pattern.

The Morel ST 1108 is a good quality tweeter with low Fs. You should crossover around 2000Hz, not 2700Hz. This will give you better vertical dispersion.

 

[Jopldangla]

I was under the impression that the datasheet data is the best starting point. I guess then I would need to follow up with some measurements and hope to understand them and also update x-over.

I realize the scale of the two images are different but in 1 case the values peak above 90 and in the other they don’t. I am sure I am doing something stupid.

How do speaker designers (DIYers) work out crossovers? Crossover components are pricey, so do people just have a bunch of random components kicking around?

 

[Jopldangla]

I made some updates... I think I figured out my issue from before. In the excel sheet calculator I put the L-Pad resistor before the crossover (oops) and that was the reason for the difference. Once I fixed that issue, the graphs matched much better.

I am hoping for any/all feedback on this x-over design. I kept the notch filter ("the filter of L3+C4+R4 is used to tame the characteristic peak Scan Speak paper woofers tend to have around 800-900 Hz"-P.Carmody...but I used the one from the Revel105 copy - McFly version.)
I posted both pictures...just incase it is helpful in anyway (not sure that it is).

 

[PatentLawyer]

In case you don’t already know, this sort of modeling is only a starting point. You have to measure your drivers in your cabinet (on and off axis) and then tweak the design iteratively until you get close enough to your target response.


Good luck, this is a big project to choose as your first!

 

[alex-z]

How do speaker designers (DIYers) work out crossovers? Crossover components are pricey, so do people just have a bunch of random components kicking around?

I don't use datasheets at all. I put the drivers in a cabinet, and measure them with the following method.

How to make quasi-anechoic speaker measurements/spinoramas with REW and VituixCAD

Quasi-anechoic measurements are a way to capture the 'real' frequency response of a speaker without the influence of a room. They are called 'quasi-anechoic' because they have the potential to get you really close to the results one might achieve with a non-reflective environment like an...

www.audiosciencereview.com

After importing the data to VituixCAD, I design a crossover. I have a small collection of cheap caps and resistors for tweaking after. An active crossover is even easier, as you can tweak settings with no extra cost or soldering work.

For the cabinet design, I measure the T/S parameters of the woofer. Manufacturer data isn't always accurate.

 

[Jopldangla]

I don't use datasheets at all. I put the drivers in a cabinet, and measure them with the following method.

How to make quasi-anechoic speaker measurements/spinoramas with REW and VituixCAD

Quasi-anechoic measurements are a way to capture the 'real' frequency response of a speaker without the influence of a room. They are called 'quasi-anechoic' because they have the potential to get you really close to the results one might achieve with a non-reflective environment like an...

www.audiosciencereview.com

After importing the data to VituixCAD, I design a crossover. I have a small collection of cheap caps and resistors for tweaking after. An active crossover is even easier, as you can tweak settings with no extra cost or soldering work.

For the cabinet design, I measure the T/S parameters of the woofer. Manufacturer data isn't always accurate.

This was super helpful. Up until this point it hasn’t clicked that the first thing to do is to collect the driver data. In my head it was always… build crossover … then collect… then tweak. It seemed like an expensive endeavor. What you proposed is much more reasonable (just need to now sort out how to collect the data).

 

[Wolf]

You can use the mfr spec plots IF you adapt them to your pending cabinet dims. You'll also have to have a Z-offset or tweeter delay entered to estimate differing time of arrivals for the drivers used, as well as their respective locations in the cabinet. Then you can sim the system responses. However, garbage in = garbage out. If you are less than accurate with your adjustments, this will skew the pending results.

As it sits, the tweeter will be too loud.

 

[D!sco]

VCAD has an awful lot of tools that need to be understood and used in the correct order to simulate appropriately, and you may still have a totally different result when you plug in real drivers to real boxes. Baffles have responses of their own, boxes resonate, impedance affects crossovers and vice versa, enclosure sizes and shapes affect bass response. This isn’t a bad starting point but it is kind of like trying to climb a hill from its highest face. Building and measuring a box with raw drivers in it would probably be better. Measuring the impedance with DATS and checking basic T/S parameters is a great start for a crossover. Copying one for different drivers will not get desirable results. Passive components affect impedance to change response, unlike digital crossovers, which alter the source. Driver design will also affect optimal enclosures. We can help you with all of this but it’s a ton of work and reading to really get it.

 

[voodooless]

It seemed like an expensive endeavor.

Well, it doesn’t have to be. For designing the crossover you don’t need expensive parts. Just use the cheaper kind of caps, and you can parallel them to mix and match values. This way, you don’t need too many of them. Coils are a bit more difficult, because a coil is just some copper. If you have an inductance meter, you could unwind a coil until you have the value you need though.

Generally, I would argue against buying expensive capacitors. I haven’t seen anyone reliably distinguish them from cheap ones in controlled testing.

And all of the above is the reason many of use abandoned passive crossovers in favor of active systems.

 

[Jopldangla]

I just received the scanspeak drivers and can now finish the enclosures. I created a custom waveguide for the morel tweeters to get better off axis performance and better alignment of the drivers. I definitely need a measurement microphone (still trying to decid on going XLR with some interface to the PC or just go USB… all while not breaking the bank). The plan is to finish the enclosures, load up drivers, take some measurements, change waveguide if needed and final 3d print it, then design crossovers, build those and final test everything. At times it seems like I have bitten off more than I can chew… hoping for some expert guidance on here (I have read so much… but there is so much more to read) to help me through any confusion I may have (seems like a good bunch of contributors on this forum). Finally…. I will share the cad models, measurements and such for all others to use (if they are even any good. Ha ha).

 

[kemmler3D]

I just received the scanspeak drivers and can now finish the enclosures. I created a custom waveguide for the morel tweeters to get better off axis performance and better alignment of the drivers. I definitely need a measurement microphone (still trying to decid on going XLR with some interface to the PC or just go USB… all while not breaking the bank). The plan is to finish the enclosures, load up drivers, take some measurements, change waveguide if needed and final 3d print it, then design crossovers, build those and final test everything. At times it seems like I have bitten off more than I can chew… hoping for some expert guidance on here (I have read so much… but there is so much more to read) to help me through any confusion I may have (seems like a good bunch of contributors on this forum). Finally…. I will share the cad models, measurements and such for all others to use (if they are even any good. Ha ha).

My advice is to get the UMIK-1 and go from there. Having done it both was with the ECM8000 and the UMIK, the UMIK is nicer. It will be the best $100 you spend on this project, believe me.

From there start taking measurements.

You might also want to consider getting a secondhand MiniDSP 2x4 for prototyping the crossover or even just implementing an active one. The speed difference in punching in different values to a DSP and running another measurement, vs. soldering stuff ... incomparable.

Once you have filters you like in DSP you can approximate them as a passive. This way you only need to buy once, solder once. If you are keen on passive crossovers you can sell the MiniDSP when you're done prototyping.

 

[voodooless]

Once you have filters you like in DSP you can approximate them as a passive. This way you only need to buy once, solder once. If you are keen on passive crossovers you can sell the MiniDSP when you're done prototyping.

If that is the plan, then no MiniDSP is needed. Just use the stereo output on your PC and use software DSP. You will need to measure the impedance of the drivers though for a passive crossover. And also remember that a DSP can do things that a passive crossover can’t. For instance a delay is basically a no-go in a passive system.

 

[Jopldangla]

Few quick questions.
1. I want to test the waveguide I created for the tweeter. Can I use a baffle with the same geometry of the baffle on the enclosure for testing or do I need to mount the tweeter in the enclosure? (reason for doing this is that I don't want to commit to the waveguide shape on the enclosure without first testing and confirming it is an improvement.
2. Will results be effected if I don't install the tweeter in the enclosure when testing the woofer? (I can place the tweeter in the enclosure to consume some volume).

 

[Jopldangla]

If that is the plan, then no MiniDSP is needed. Just use the stereo output on your PC and use software DSP. You will need to measure the impedance of the drivers though for a passive crossover. And also remember that a DSP can do things that a passive crossover can’t. For instance a delay is basically a no-go in a passive system.

What software would you recommend for software DSP?

 

[voodooless]

What software would you recommend for software DSP?

EQ Software for Windows, Linux, macOS, iOS/iPadOS and Android.

Any suggestions are welcome. Post suggestions here. I will gather the details and add them. Thanks. :D This will be organized as results come in. @sweetchaos made a guide for PEQ filter data input and operation. For crossover software go here. Windows: AutoEQ four easy steps to make your...

www.audiosciencereview.com

 

[Jopldangla]

I don't use datasheets at all. I put the drivers in a cabinet, and measure them with the following method.

How to make quasi-anechoic speaker measurements/spinoramas with REW and VituixCAD

Quasi-anechoic measurements are a way to capture the 'real' frequency response of a speaker without the influence of a room. They are called 'quasi-anechoic' because they have the potential to get you really close to the results one might achieve with a non-reflective environment like an...

www.audiosciencereview.com

After importing the data to VituixCAD, I design a crossover. I have a small collection of cheap caps and resistors for tweaking after. An active crossover is even easier, as you can tweak settings with no extra cost or soldering work.

For the cabinet design, I measure the T/S parameters of the woofer. Manufacturer data isn't always accurate.

When measuring a raw driver with REW, do you run the full 0-20khz sweep on all drivers? Is there a chance that the full sweep can damage the driver (I feel the answer is yes)? I am thinking of running the sweep at the same starting/ending frequencies shown on the individual drivers datasheet db/freq curve.
Good news is I picked up a UMIK-1 and am ready to follow the quasi-anechoic chamber guide and have the cabinets all fabricated and are waiting for glue-up (waiting for acoustic foam to arrive).

 

[Paweł L]

The tweeter doesn't or shouldn't be tested from 10Hz. One possibility is burned voice coil when running such low frequencies and mechanical damage. I usually limit the test to 200Hz, and when doing such measurements inside places like living room the room modes and reflections cause errors below 1kHz.
As for testing the waveguide mounted on the baffle, I would make a removable baffle for testing/prototyping and glue final one to the cabinet. The tweeter/waveguide volume has somewhat minimal effect.
As for the freq. sweeps, additional In phase and Out off phase, with correct min phase/distance, plus offset between drivers are a must for successful X-over CAD simulation which confirm with real life summation and measurements.
If possible make outdoor measurements, far off the ground using tall ladder to avoid room reflections.
As for X-over itself, you might have to change the crossover's frequency/slope a bit to get the driver's phase in line if you're after textbook slopes and phase matching. Also don't forget that tweeter even in the waveguide may not match woofer's directivity, therefore ideal flat freq resp is not always what the ear is after.

BTW, the guide how to make proper measurements linked by Alex-z, is something what I would do, but not always have chance to. Nice write up.

 

[Jopldangla]

The tweeter doesn't or shouldn't be tested from 10Hz. One possibility is burned voice coil when running such low frequencies and mechanical damage. I usually limit the test to 200Hz, and when doing such measurements inside places like living room the room modes and reflections cause errors below 1kHz.
As for testing the waveguide mounted on the baffle, I would make a removable baffle for testing/prototyping and glue final one to the cabinet. The tweeter/waveguide volume has somewhat minimal effect.
As for the freq. sweeps, additional In phase and Out off phase, with correct min phase/distance, plus offset between drivers are a must for successful X-over CAD simulation which confirm with real life summation and measurements.
If possible make outdoor measurements, far off the ground using tall ladder to avoid room reflections.
As for X-over itself, you might have to change the crossover's frequency/slope a bit to get the driver's phase in line if you're after textbook slopes and phase matching. Also don't forget that tweeter even in the waveguide may not match woofer's directivity, therefore ideal flat freq resp is not always what the ear is after.

BTW, the guide how to make proper measurements linked by Alex-z, is something what I would do, but not always have chance to. Nice write up.

Thank you for the tip. I will make sure not to sweep frequencies which are too low for the tweeter. I think I may even start later than 200hz. I am planning to test outside and follow that measurement guide.

 

[Jopldangla]

I have glued up the enclosures, mounted the drivers, and taken near field, far field, horizontal and vertical measurements of both raw drivers in the enclosure. Here is a sampling of my progress.

The measurements were taken indoors in a large warehouse. The drivers were approx 5' from the ground. Far fields were taken at 22" and near field at 0.2" (tried to follow Jeff Bagby white paper). Near field port and woofer were summed, adjusted for SPL level, adjusted for baffle diffraction, then they were merged with the far field measurements at/around 350hz. I used gate timing of 5ms, which is 200hz. Any idea what is going on with the scanspeak in the 250hz range?

Here is the six-pack from VituixCad with my first go at a crossover. (I actually just used about the same crossover as shown in my first post).

@PatentLawyer you were right...this is a big project. its been fun so far though.