DIY inspired Buchardt S400 – Crossover design

[Madslide]

Dear all,

After hours of reading on the web and especially on the ASR forum, I’m finally decided to start designing my first DIY speakers.
Initially I wanted to start with a tower speakers but I reconsidered my plan by first trying to take experiance first with a 2 Way Bookshelf speaker.

This project will be not very original as I want to make an inspired version of the Buschardt audio S400.

In that regard, I was planning to use :
-SB17NBAC for the woofer ;

Speaker SB Acoustics SB17NBAC35-4, impedance 4 ohm, 6 inch

en.toutlehautparleur.com

-SB26STAC for the tweeter with a 3D printed wave guide ;

Dome tweeter SB Acoustics SB26STAC-C000-4, impedance 4 ohm, voice coil 26 mm

en.toutlehautparleur.com

-SB15SFCR for the passive radiator.

Speaker passif SB Acoustics SB15SFCR-00, 5 x 8 inch

en.toutlehautparleur.com

The woofer SB17NBAC as an Sd of 118cm² and a Xmax of 7.17mm (so 84.6cm^3 of displaced volume)
The passive radiator as a Sd of 178cm² and à Xmax of 11mm (so 195.8cm^3 of displaced volume)
So the passive radiator can displaced 2.3 time the air volume of the woofer which looks in the rules.

I was planning to use a 18L box and tried to simulate that in WinISD and it look like I could obtain a 50hz low at -3db.
So first, what do you think of this driver selection ?

Then come the hard part for me as I have no skills in crossover design and I know it have a huge rôle in the final result (I only know how to read a crossover schematic and good at welding components)
By looking at the frequency response of the drivers, I was thinking of using a crossover point somwhere beetween 1500 and 1800hz.

This leads me to ask you few questions :
-How do you select the nominal speakers impedance as the woofer is available in 4 and 8 ohm ?
-Does a generic crossover always lead to a bad result ?
-Is there any active crossover module that can be used with numerical output for 2 FDA « budget » amplifier ? ;
-Othrewise, where to start from and which software do you recommand for passive crossover design ? (Go back to school is an acceptable answer ;-) ;

I also plan to order a UMIK-1 as I know it will be necessary to adjust parameters in every cases.

Thanks !

 

[D!sco]

Without the waveguide, your design will end up quite different from the S400. Everything matters. Spacing, baffle dimensions, depth of cabinet. You don't need to know it all to make a good speaker, but it keeps you from chasing your tail around. I assume you've read some about directivity being on this site? The waveguide is huge for that. Otherwise you would be best off putting the two drivers as close together as you can stand, which changes the crossover from the original. None of this is to discourage you, just to say it ain't all that easy to copy another person's work.

-How do you select the nominal speakers impedance as the woofer is available in 4 and 8 ohm ?

8 ohm is standard, 4 ohm is for specific driver arrangements as necessary. Most amps don't like lower ohm ratings and thus the standard.

-Does a generic crossover always lead to a bad result ?

Yes, no exceptions.

-Is there any active crossover module that can be used with numerical output for 2 FDA « budget » amplifier ? ;

If there was I would've bought it by now. It's all Hypex for HiFi. Some Dayton boards no one has tested, primarily for portable sound. I actually made a fake company account to contact the manufacturers of the Buchardt A500 amp, but they're not interested in small batches.

-Othrewise, where to start from and which software do you recommand for passive crossover design ?

VituixCAD 2.0 is a bit to really get your head around, but it's damned near perfect when used correctly.

Drivers and dimensions look good otherwise. The hard part is simulating the elements in software so you can accurately model a decent crossover. Good luck.

 

[HammerSandwich]

So first, what do you think of this driver selection ?

Where did you find >7mm Xmax for the SB17? Factory specs list 5.5mm, and @hardisj's Klippel results for the (very similar) SB17CAC showed only 2.8mm, limited by the suspension. That said, distortion results at Dibirama & Hificompass hint that SB17s are comparable to similar drivers, except Purifi's (obviously). Perhaps Erin's sample had an issue, bass distortion is tricky to measure, etc. Just be aware.

Similarly, I'd consider the SB26ADC or CDC instead of the soft dome. You'd sacrifice some sensitivity, but distortion & Fs are significantly lower. The hard domes look a little better on Augerpro's waveguides as well, at least to my eye.

re: DSP, you might find a used Minidsp Nanodigi, and the Flex with digital IO is supposed to arrive soon. God knows in 2022...

And +2 for VituixCAD. I have 2 thumbs, and it's great. Especially for the price! Learn it, and send Kimmo a few bucks.

 

[Madslide]

@D!sco , thanks for the feedbacks.
I’ve mentionned the S400 but I could also have mentionned the Revel or the Directiva R1 as an inspiration source (but the budget is out of my scope for the R1).

I assume you've read some about directivity being on this site? The waveguide is huge for that.

Yes I’ve read few topics about the waveguides and especially the Augerpro's waveguides topic which led me to think it would be good to use. Also I have access to different type of professional 3D printer so that easy for me to test few iterations.

If there was I would've bought it by now. It's all Hypex for HiFi. Some Dayton boards no one has tested, primarily for portable sound. I actually made a fake company account to contact the manufacturers of the Buchardt A500 amp, but they're not interested in small batches.

Actually a french audio craftsman (the shop name is Atelier Acoustik) have made such full digital DSP system but it’s a custom work and it have a cost I can't afford. (It’s important to read the comment above the picture in both link to understand what as been done/modified)

https://www.facebook.com/atelieracoustik/posts/2111432842473177

https://www.facebook.com/audiophonicsfr/posts/1068016616710291

Thanks @HammerSandwich ,
I found the Xmax of 7.17mm in my link above but you are absolutely right, on the SB acoustics website they mention 5.5mm so there is a mistake in the technical specs in my link. This lead to a corrected ratio of 3 times the air volume displaced for the passive radiator vs the woofer. And may be more if it’s only 2.8mm in the real world (may be the tested unit was not « break-in » ?)

Similarly, I'd consider the SB26ADC or CDC instead of the soft dome. You'd sacrifice some sensitivity, but distortion & Fs are significantly lower. The hard domes look a little better on Augerpro's waveguides as well, at least to my eye.

Also it’s true that a lot of work as been done on the waveguide for the SB26ADC and CDC and that would be a mistake to do not take this into account.

Will go for VituixCAD so...
I emailed SB acoustics in order to know If they could share there frequency/impedance graph files. I doubt but who knows…

 

[HammerSandwich]

I emailed SB acoustics in order to know If they could share there frequency/impedance graph files.

Please let us know if SB sends more than the datasheets include. If not, Vituix's SPL Trace tool works pretty well with the charts at Hificompass. You won't get past 60 degrees, but that will give you a reasonable start with your sims.

 

[D!sco]

Those measurements are often smoothed to a high degree, or aren't necessarily what you would get in the box. A better brand like SB Acoustics would probably have decent QC model to model, but ultimately measuring it in-box or in-baffle in free field is your best bet. Especially since you plan to have a UMIK-1 already.
If you have the time and space, that is.

 

[Robovox]

SBA always quote peak to peak Xmax values on the data sheets, so 11mm/2 = 5.5mm is the correct figure to use.

If you are looking for a budget active solution, I have used the Hifiberry Beocreate board to build two-way active speakers - it gives you enough DSP power to implement a two way crossover and bafflestep / diffraction compensation (and an assisted bass alignment) and has four channels of amplification for about £180 in the UK. The amps are TPA3118d2 class D chips - the sound quality is OK if you don’t turn the volume up too high. Passive crossovers are a lot harder to get right than active - you can’t tinker with them easily unless you have a stock of inductors, resistors and capacitors to hand.

 

[fluid]

I found the Xmax of 7.17mm in my link above but you are absolutely right, on the SB acoustics website they mention 5.5mm so there is a mistake in the technical specs in my link. This lead to a corrected ratio of 3 times the air volume displaced for the passive radiator vs the woofer. And may be more if it’s only 2.8mm in the real world (may be the tested unit was not « break-in » ?)

Manufacturers quote x-max in varying different ways, some determine it from where THD reaches 10% others specify it based on physical dimensions, and they can't all agree on which dimensions to use either. The figures are often inflated from the real performance that can be expected.

If you look at the Klippel results for the driver you can see the points where the driver parameters change beyond a certain point.

X C in Erin's table (or Kms (x) as it tends to be shown as in audioexpress/voice coil) shows the speaker is displacement limited to 2.8mm due to suspension non linearity, beyond 2.8mm of excursion THD is above 10%. Even though the physical x-max is higher the suspension makes it irrelevant.

Will go for VituixCAD so...
I emailed SB acoustics in order to know If they could share there frequency/impedance graph files. I doubt but who knows…

You can trace the graphs from the manufacturers data sheet very easily in Vituix to gain some insight but the method has some significant limitations.

A guide here
https://www.diyaudio.com/community/threads/vituixcad.307910/post-6841450

 

[Prana Ferox]

Your drivers don't match up but your general design sounds very similar to Jeff Bagby's 'Helios CM' design. That's a build I want to do at some point. I'm not sure if I've ever seen the crossover for that published the crossover wasn't published openly, and obviously with different drivers it's of limited help, but Javad mentioned on Facebook you may be able to email Meniscus for a schematic.

 

[hardisj]

Perhaps Erin's sample had an issue, bass distortion is tricky to measure, etc. Just be aware.

IME, SB always overrates their linear excursion. You can see evidence in other tests (Vance Dickason has some measurements in Voice Coil mag, too, IIRC).

 

[alex-z]

Will go for VituixCAD so...
I emailed SB acoustics in order to know If they could share there frequency/impedance graph files. I doubt but who knows…

Even if SB Acoustics does send you that data, it is almost useless. Manufacturers test their drivers in large IEC baffles, to reduce the influence of baffle step loss and edge diffraction. The response in an actual speaker cabinet is significantly different, and you should be measuring your own drivers. Ideally you want a full CEA-2034 plot for each driver, so that you can fully optimize the crossover.

In order to get that quality of data, you need a turntable with no front reflection, and a microphone stand + calibration file setup for free field use. Most measurement microphones ship with random incidence calibration by default. Which means you may need a third party calibration from a place like Cross Spectrum Labs.

I strongly recommend measuring outdoors, as most people have 8ft ceilings, which limits you to about 4ms gated measurements. Ideally you want 6ms or more, low resolution can easily mask resonances.

 

[fluid]

Even if SB Acoustics does send you that data, it is almost useless.

Useless for making a final crossover but quite useful for evaluating different options, baffle sizes driver positions and crossover topologies. The IEC baffle diffraction can be undone and the design baffle diffraction simulated to provide full 360 degree data. The document linked to in the post above gives instructions and warnings on the assumptions being made with limited data.

 

[hardisj]

Even if SB Acoustics does send you that data, it is almost useless. Manufacturers test their drivers in large IEC baffles, to reduce the influence of baffle step loss and edge diffraction. The response in an actual speaker cabinet is significantly different

Many programs assume an infinite baffle for the driver and the enclosure simulation accounts for baffle step & diffraction.

 

[Rick Sykora]

Regret that I am buried rn and cannot assist more. Rest of Directiva team is busy too.

Kimmo’s design checklist here is very helpful: https://kimmosaunisto.net/Software/...20.html#Checklist_for_designing_a_loudspeaker

Directiva r2 is using the smaller version of the SB woofer you bought. The Xmax spec issue is disconcerting but, in the end is price/performance question and what other driver performance is a better value. Directiva team looked at a bunch of other drivers and went with the SB. The overall value was hard to beat.

 

[Rick Sykora]

This might help too…

Designing an active 2-way speaker crossover

So, while I wait for Amir to start testing his speaker with the supplied crossover (we will post once we validate the VituixCAD sim and his measurements match), thought I would share a bit more of my testing and a simple crossover development for r1. I am still iterating as I have looked at...

www.audiosciencereview.com

 

[Rick Sykora]

Manufacturers quote x-max in varying different ways, some determine it from where THD reaches 10% others specify it based on physical dimensions, and they can't all agree on which dimensions to use either. The figures are often inflated from the real performance that can be expected.

If you look at the Klippel results for the driver you can see the points where the driver parameters change beyond a certain point.

View attachment 199712
X C in Erin's table (or Kms (x) as it tends to be shown as in audioexpress/voice coil) shows the speaker is displacement limited to 2.8mm due to suspension non linearity, beyond 2.8mm of excursion THD is above 10%. Even though the physical x-max is higher the suspension makes it irrelevant.


You can trace the graphs from the manufacturers data sheet very easily in Vituix to gain some insight but the method has some significant limitations.

A guide here
https://www.diyaudio.com/community/threads/vituixcad.307910/post-6841450

This is an excellent point. Unless conditions are stated, it is difficult to know how comparable drivers specs are. As Xmax is a key one for sims, not surprised that all drivers may not yet based on Klippel metrics. If a driver was developed prior to Klippel are they going to go back remeasure and restate Xmax to (a likely lower value)? Probably not.

 

[D!sco]

re: DSP, you might find a used Minidsp Nanodigi, and the Flex with digital IO is supposed to arrive soon. God knows in 2022...

If you have a set, please send it in to Amir for testing. I have a strong need to know about the performance of more DSP solutions.

 

[Madslide]

Thanks for the feedbacks and your advices.

I just ordered the UMIK-1.
Confidance in the speaker data provided by the manufacturers looks to be a complete mess and as an R&D guy, I’m a bit amazed that there is no standard ASTM or a standardized protocol for such mesurements ?

I played a bit with the VituixCAD software and eaven if now I have the two graphs using the picture tool, it doesn’t looks really accurate or equal to the technical spec sheet. Nevertheless It look to be a good starting point.

I’m not decided yet If I will go for passive or active crossover but my thought of the day is when I see how complexe, costly and uncertain a passive crossover looks to be I wonder if I should not buy a second FDA and a MiniDSP NanoDigi (apparently discontinued – If you know where to find one for sale ?). It’s an investment but I have the feeling that once done, you can test almost everything.

 

[HammerSandwich]

Ack! Sorry to miss this before.

If possible, cancel the UMIK order. You want an analog mic + a 2-channel interface, so that you can use loopback for proper phase in your measurements. UMIK is fine for room correction, but you're looking at XO design.

 

[fluid]

Confidance in the speaker data provided by the manufacturers looks to be a complete mess and as an R&D guy, I’m a bit amazed that there is no standard ASTM or a standardized protocol for such mesurements ?

I played a bit with the VituixCAD software and eaven if now I have the two graphs using the picture tool, it doesn’t looks really accurate or equal to the technical spec sheet. Nevertheless It look to be a good starting point.

Most manufacturers test their drivers on an IEC baffle which is standardized for frequency response, but the distance of the measurement can vary or not be stated. The scales used can make a big difference in how something looks. A compressed vertical scale will make the driver look flatter, the same data viewed in Vituix will look different because the scale is more equal between horizontal and vertical.

I’m not decided yet If I will go for passive or active crossover but my thought of the day is when I see how complexe, costly and uncertain a passive crossover looks to be I wonder if I should not buy a second FDA and a MiniDSP NanoDigi (apparently discontinued – If you know where to find one for sale ?). It’s an investment but I have the feeling that once done, you can test almost everything.

For experimenting and for anyone not well versed in passive crossover design or someone who already has a large stock of passive parts an active crossover is considerably easier to get right, and easily changed if it turns out you made a mistake along the way.