3 way with Dayton and SB drivers.

[Ktacos]

A recent build that lacked some foresight has had me back in my shop to work on a new speaker that has been planned out better. It really is true what the research says, bass plays a huge role in how we perceive a speaker and my last build made that apparent. This one has plenty of bass thankfully. I did a driver purge not too long ago and kept just enough to do one speaker. The left overs were an RS180, RS100, and SB26STCN.

Woofer cabinet is probably ~15L, mid cabinet volume is ~1L. Vcad says my port tuning is around 37hz or so with a 2"x6" port, measuremetns will tell the trurth. Main material is BB with some red oak trim on the sides.

Only thing left to do is give a final sand and finish and develop passive filters. Running some quick basic active filtering for testing I am quite pleased with the sound. Will probably be 2026 when I post again on these so have a lovely holiday and a happy new year!

 

[Ktacos]

Got a small window of time to measure the speakers. It was very windy out. Probably could have gotten better DI with an offset on the drivers but I think this ok. It's not doing the typical 2 way narrow mid range into omni 3-5k tweeter thing that annoys me. Will tweak filters some more later.

 

[Ktacos]

I came up with a few xover iterations and settled on trying one that came out to about 60 usd in parts per speaker. This isn't the final xover but what is here does sound pretty good. I've done all post sim tuning with a single speaker. I spent quite a bit of time just listening to music with the single speaker and at times it was very engaging so I think that's a good sign. Currently my only complaint is that is lacks a bit of body in the lower mids which I hope reducing the 5mh value should alleviate.

The resistor on mid range inductor and the notch cap on the woofer inductor have been omitted. I am not sure if I need dimensional driver offsets as I've mostly ditched the vertical data I took, I didn't feel it was accurate enough, doesn't really show lobing on vertical axis like one would expect. The Z offsets when applied seemed to make almost no difference so I just left them out. I am under the impression Z offset isn't needed if you're using a fixed mic measurement process.

This is the first measurement that I did with the crossover shown above compared to my sims estimated in room response. Kind of cool to see everything come together.

 

[a4eaudio]

Crossing the RS100 at 3500Hz is a good place. I have not used that tweeter, but if you use an L-Pad you might be able to drop the 0.20mH inductor and 3.9uF capacitor.

Currently my only complaint is that is lacks a bit of body in the lower mids which I hope reducing the 5mh value should alleviate.

1) Why are you crossing the RS100 to RS180 at close to 750Hz? The RS100 can easily cover down to 350Hz so right now your mid is only covering about 2 octaves.
2) You should try measuring nearfield driver and port and merging with the farfield, you may not be getting what you think from your simulation.

I have a three way with the RS180-8 and RS100-8. I had to use a notch filter (with large 6mH inductor and 330uF capacitor) to reduce the upper impedance spike on the RS180. When I do, I have a 5.0mH inductor with a 125uF shunt capacitor to reach a xo point of 385Hz. BUT...without the series notch filter, I have a huge hump in SPL. If I don't have accurate merged nearfield and farfield I would not see this in my measurements. Measuring the actual low-end was VERY challenging and I ended up taking ground-plane measurements to finally have confidence it was I was actually getting. (As a note, I had a hard time getting the merged farfield, nearfield, port results to match VituixCAD but in fact it did match very accurately to the ground plane measurements.)

Below is the actual measurement of the speaker.
Below that is if I would NOT have included the notch filter. I would get a 6dB peak at 150Hz and would NOT know this without accurate low-end measurements.

 

[a4eaudio]

The Z offsets when applied seemed to make almost no difference so I just left them out. I am under the impression Z offset isn't needed if you're using a fixed mic measurement process.

You do not need Y-offset if using fixed mic position.
You do not need Z-offset if you are using a dual-channel measurement setup that accurately measures phase.
An easy test to see if you have your phase information is correct is to measure with the drivers in reverse polarity and compare the reverse-nulls to the simulation. If your measurements are accurate, they will match VituixCAD closely. If they don't match closely, then your data is not accurate (i.e., it is user-error not VituixCAD error. VituixCAD is very accurate.)

 

[a4eaudio]

One other comment on Z-offset. The most obvious place you will see a difference is the phase rather than the on-axis SPL.

Below, the black lines are what I might expect based on measurements and simulation, not correctly taking into account the Z-offset.

But the red lines correctly take into account a Z-offset of the midrange being 3/8-inch behind the tweeter and the woofer being 1-inch behind the tweeter.
The on-axis difference is very small around the crossover point at 3.5kHz. But you can see in the reverse-nulls that I while I would have thought I had good phase alignment based on the black (dashed-dot) line, in reality it wasn't as good.

For the crossover point at 385Hz, if you don't have accurate merged nearfield driver, port, and farfield, then you probably don't have accurate phase information. If you cross a three-way at 750 Hz, or a two-way at 2.5kHz it won't matter much, but if you design a three-way crossed around 350Hz (a typical point), then you need to get the phase information correct. (At least theoretically, it may or may not audibly matter.)

 

[Ktacos]

You do not need Z-offset if you are using a dual-channel measurement setup that accurately measures phase.
An easy test to see if you have your phase information is correct is to measure with the drivers in reverse polarity and compare the reverse-nulls to the simulation. If your measurements are accurate, they will match VituixCAD closely. If they don't match closely, then your data is not accurate (i.e., it is user-error not VituixCAD error. VituixCAD is very accurate.)

I didn't use timing reference, but I went ahead and redid the measurements with one. I'm not entirely sure how to manage phase in measurements. When I measure it's just endless wraps that are nigh impossible to work with. I measured the tweeter on axis then did estimate IR delay and applied the offset to all further measurements I took. Is this the correct way to do it?

I did do some reverse null tests and they looked fine but weirdly vituixcad wanted me to invert the polarity of my tweeter as well as the mid, which I felt was unusual. It also just didn't sound right.

) Why are you crossing the RS100 to RS180 at close to 750Hz? The RS100 can easily cover down to 350Hz so right now your mid is only covering about 2 octaves.

Just seemed like it offered easy summation. I find small mids going lower tend to make things sound kind of, small to me.

I'll try and get a merged NF+Port response. Those never go well for me so I must be doing it wrong.

I'm also having trouble getting phase to look as predicted, but also not sure if I'm looking for the right things. I was under the impression that you're typical 2nd order LR filter will show a 180 flip at the xover point but I don't see that. What I do get are some continuous traces that I can line up. Am I looking to line up the traces as best as possible or should I do the typical phase flip at the xover point?

FWIW if you need any .mdat's of anything let me know, still working on getting this measurement process down.

 

[Ktacos]

Below that is if I would NOT have included the notch filter. I would get a 6dB peak at 150Hz and would NOT know this without accurate low-end measurements.

That's also another reason for my higher xover point, I noticed that the further down I went I would get the same boost. Wasn't aware one could bring it down passively.

 

[a4eaudio]

That's also another reason for my higher xover point, I noticed that the further down I went I would get the same boost. Wasn't aware one could bring it down passively.

I wondered if that was the case, which is why I asked. I didn't want to assume to know what issues you ran into, but it is a coincidence that I happened to use the same mid and woofer so other than different baffle diffraction there are probably a lot of similarities.

I had happened to read a thread about using an LCR notch to flatten the upper impedance peak in some ported builds. At the time I didn't really get it, because I had never experienced the issue, but was lucky that when I was getting weird results I remembered the thread and reached out to some people for help. The advice I got was to add the LCR notch and determine the component values needed to flatten the peak before any other components are added. I had a 4mH parallel inductor on the midrange, 7mH series on the woofer, and 12mH in the LCR filter. I then developed a prototype xo. Once I was satisfied I went back to fine tune the component values (12 of 16 xo component changed at least a little) and was able to reduce the inductors to 4, 5 and 6mH (raising the xo point from 300 Hz to 385 Hz).

First filter...

Most recent...(I had to work hard to keep the minimum impedance above 3.2 ohms with the 6mH inductor)

 

[a4eaudio]

...I'm not entirely sure how to manage phase in measurements. When I measure it's just endless wraps that are nigh impossible to work with....

What is your mic setup, USB mic or XLR mic + USB interface?

I did do some reverse null tests and they looked fine but weirdly vituixcad wanted me to invert the polarity of my tweeter as well as the mid, which I felt was unusual. It also just didn't sound right.

My xo topology looks a lot like yours, except the parallel inductor and capacitor at the end of your tweeter filter. My tweeter and midrange are inverted.

...I'm also having trouble getting phase to look as predicted, but also not sure if I'm looking for the right things. I was under the impression that you're typical 2nd order LR filter will show a 180 flip at the xover point but I don't see that.

Each additional inductor or capacitor affects phase. So, for example, you may be able to achieve acoustic 2nd order LR on the tweeter with just a cap and inductor (2nd order electrical), or by adding the second capacitor in series (3rd order electrical but tweak values to maintain 2nd order acoustic). This is why it is common to add the second capacitor in series as it adjusts the phase to align better when using a flat faceplate tweeter on a flat baffle with no slope or step to physically align the drivers.

This is why the Z-offset matters. Right click on any curve (phase curves, SPL or SPL when polarity is inverted) and "Save selected as overlay." Set Z-offset on the mid to +10mm and on the woofer to +28mm. As you change xo parts you will be able to see the difference between the new results and what you had (most pronounced in the phase and most visible when looking at reverse polarity results).

...FWIW if you need any .mdat's of anything let me know, still working on getting this measurement process down.

Let's talk about your mic and setup first and see if you need to change anything in your measurements and then post the .mdat.

 

[Wolf]

Honestly, I feel that 2.5" cone RS100 will have issues at higher output levels with xover at 350. My gut would say 500-600Hz is ideal. If it falls into place at 700, then maybe that is what it wants.

The bump in magnitude for the woofer that requires the LCR is a result of the woofers' higher in box impedance spike in relation to the high inductance lowpass coil with low DCR. That is why the bump is there.

Using a higher xover, lower mH value coil, a higher DCR or smaller gauge coil, a full LCR conjugate, a split 6dB filter at the bump, or a large high power 40-50 ohm resistor across the woofer can help mitigate it.

 

[Ktacos]

What is your mic setup, USB mic or XLR mic + USB interface?

EMM6 into motu m2, loop back for timing reference.

 

[a4eaudio]

Honestly, I feel that 2.5" cone RS100 will have issues at higher output levels with xover at 350. My gut would say 500-600Hz is ideal. If it falls into place at 700, then maybe that is what it wants.

I understand where you are coming from in principle. In this case, the RS180 hits xmax at about 60w and max SPL is about 106-7dB in 4-pi space. The RS100 can hit about 106db with 22 watts barely hitting 1mm if crossed at 350Hz. But I don't disagree that higher may be better and the RS100 would not be able to keep up with something more capable (in terms of max SPL) than the RS180. In my case, my crossover points were somewhat predetermined as I was using the tweeter and mid from the Barefoot01 and knew they had used crossover points of 250Hz and 3.6kHz.

The bump in magnitude for the woofer that requires the LCR is a result of the woofers' higher in box impedance spike in relation to the high inductance lowpass coil with low DCR. That is why the bump is there.

Using a higher xover, lower mH value coil, a higher DCR or smaller gauge coil, a full LCR conjugate, a split 6dB filter at the bump, or a large high power 40-50 ohm resistor across the woofer can help mitigate it.

This is all very informative, thanks! What do you mean by "split 6dB filter at the bump?

 

[a4eaudio]

EMM6 into motu m2, loop back for timing reference.

If you want to post your .mdat file I will take a look - I'm no expert, but I'll let you know if anything jumps out at me.

 

[Ktacos]

If you want to post your .mdat file I will take a look - I'm no expert, but I'll let you know if anything jumps out at me.

If distortion is bad in random measurements it's because I live next to a highway.

This is why it is common to add the second capacitor in series as it adjusts the phase to align better when using a flat faceplate tweeter on a flat baffle with no slope or step to physically align the drivers.

Idk why but this made something click for me. I was just getting my phase to line up at the xover frequency but it appears I want the phase to track throughout the drivers bandwidth as well as outside of it.

Here's a zip of mdat, vcad project, near field woofer and port.

Ktacos 3 way data pack.7z

drive.google.com

 

[LightninBoy]

"SB Drivers" is a strange name for a gal, but congrats anyways!

 

[Wolf]

I understand where you are coming from in principle. In this case, the RS180 hits xmax at about 60w and max SPL is about 106-7dB in 4-pi space. The RS100 can hit about 106db with 22 watts barely hitting 1mm if crossed at 350Hz. But I don't disagree that higher may be better and the RS100 would not be able to keep up with something more capable (in terms of max SPL) than the RS180. In my case, my crossover points were somewhat predetermined as I was using the tweeter and mid from the Barefoot01 and knew they had used crossover points of 250Hz and 3.6kHz.



This is all very informative, thanks! What do you mean by "split 6dB filter at the bump?

Essentially 2 parts, L in parallel with C (maybe added R to C for balance, not additinally in parallel), space Fc of each (lower on L, higher on C) to lower the bump. I had to do this on my EMP build for a 200Hz inherent warmth. I think it was a 3.0mH laminate core in parallel with a 680uF and a small resistance.

 

[Ktacos]

This is a little bit of unorthodox development but I changed the speakers entirely. After spending some time with them, the honeymoon phase wore off and I was kind of bothered by their dispersion width and less than stellar DI. I think I'm just spoiled by speakers like the mechano23, just perfect tonality out of the box, nothing stands out as a problem. I had one waveguide left over that I never used that I felt could get me something similar.

Finally used the somasonus 6.5" circular waveguide my friend 3d printed me awhile ago. Put the SB26ADC in it. Sounds wayyyyy better than the three way configuration. It was just large enough to terminate right at the roundovers edge and cover the original tweeter and mid holes. Was a challenge routing it but I got it. Gained a good bit of bass extension no longer having the mid cabinet take up volume. Hopefully get some data soon with some tips from a4eaudio. I guess the thread could be renamed or a new one made? Not sure. Perhaps keep it as it so people can see what ridiculousness can transpire with a build?

 

[Salt]

Any measurements available? Would be intersting to compare 2-way and 3-way.

 

[Ktacos]

I have the DI, on axis is a little boring just being a fairly neutral line. Took some tips from a4eaudio and my data looks better, reverse nulls appear to lining up pretty well. They sound pretty good with passive filtering. My brain does kind of miss that time alignment on the tweeter, but at least I'm not hearing the whole "something is off" phase problems of the 3 way. I had a bit of a scare with my DSP setup (one interface into another) when reaper lost connection and got it back, sent a pop into the drivers that was too loud.

My PR mechano23 have spoiled me, I can definitely hear the port tube resonance on these speakers. I might splurge in the future and install some PR's and omit the port.

Here's the DI. Definitely more in line with what my ears like. These are definitely the best sounding waveguides I've used so far.