Seas 3-way wide baffle

[norcad]

Yup true (beware - a phrase you don't often see on boards follows) - my mistake.

David.

Well, now we have seen it twice already!

With new measurement, onaxis to mid, a mug of tea and a lot of sigarettes, the times was come (ones again) to make some new crossovers.
After a couple og hours fighting with Speaker Workshop, I came up with this:









And I can tell you right away, this sounds good!
(But not great )

[Dave Bullet]

Hi norcad,

Can you tell what part of the spectrum doesn't sound great? According to Zaph's HD sweep, the CA18 (presume you are still using that for the mid?) has a 3rd order HD peak around 1.5KHz.

Are you listening to a stereo pair or just the one speaker in mono?

Have you tried a lower crossover point? My excursion modeling shows the CA26 / L26 will give up before a Seas 27TDFC does at around 1.6KHz. A lower point will help off axis too.

Could you please also post phase response for each driver? Although your on-axis looks nice and flat maybe phase starts deviating from target BW above or below XO.

Last but not least, since you have designed a removable back panel, if you have some scrap, try putting a scrap piece in half to form the CA26 enclosure and running the Mid OB. I know this might cause more problems than it solves, but that would eliminate any backwave issues through the CA18.

And one more thought... have you done any HD testing in SW?

Cheers,
David.

[norcad]

Thank you for your reply Dave!

I actually have a hard time explaining what that doesnt sound great.
In english its imposible!

I always start with one speaker, and after constructing a new network, I always listen in mono to check the overall frequency response first, before connecting the second speaker.

I did try xo points down to 1,8khz before I increased the size of the midenclosure and without this new baffle. But I'm not sure if I agree with you that a lower point will help the off axis response. A small tweeter playing down in frequency, will have a "poor" power response IMO.

This I dont understand, could you please explain?
Although your on-axis looks nice and flat maybe phase starts deviating from target BW above or below XO

Are you talking about minimum phase? If so, I dont have it. The measurements is taken from 1m, onaxis to the mid for both drivers, without moving the mic. Im sure you know how the phaseplot looks like then?

I have tried the CA18 OB, without the woofer connected. As you said it made more problems so I didnt try it any further.

And once again; can you explain HD testing?

[augerpro]

I think Dave is asking if you've done off axis measurements and plugged these into the XO model. One on axis measurement is not reperesentative of the speaker response off axis. Everything may look ok on axis but off axis there may be some gremlins-and you hear this off axis sound just as strongly as the on axis sound. I typically take several off axis measurements between 15-45degrees and plug those into teh XO model to see what is happening.

Another thing is teh art of voicing the speaker. Just a frequency response measurement (particulalry a single on axis measurement) is not always a good indication of how a speaker sounds. This is probably where you need to focus some attention. I typically use a 31 band EQ to see what response changes help or hurt the sound, then look at the XO model and see how it relates, and how I can accomplish the EQ (if positive) in the XO. Getting two drivers to integrate well can have a large impact in the sound that is not obvious in the response alone.

And it just may be that you don't like the sound of these drivers. Have you heard them in a commercial design you liked? Have you heard the Sonus Faber speaker this is modeled after?

[norcad]

I did some off axis measurements of the drivers alone, when I first startet this. And I measure it off axis with the xo in place, using Arta. Its only the demo ver. So I cant save those plots.
I havent tried to use any off axis respons in SW modeling XO.

Voicing the speaker is a tricky part! At least for a noob like me. I do have an eq, and I have experienced that I do like an BBC dip in the 3-4khz area.
Another important thing, atleast for me, is that SW uses a low resolution in the measurements that hide to much information IMO.

This project is a learning prosess, and its fun to try out different solutions and see (hear) whats happening.
I have listen to speakers with one of this drivers, and liked it. But none with this driver combo.
The Sonus Faber Stradivari? Yes I have listen to them, several times, but also seen its pricetag! I will never dream of comparing my DIY speakers with anything close to those! In Norway they cost about 52,000.-UDD 8O

[Dave Bullet]

I did try xo points down to 1,8khz before I increased the size of the midenclosure and without this new baffle. But I'm not sure if I agree with you that a lower point will help the off axis response. A small tweeter playing down in frequency, will have a "poor" power response IMO.

I'm mainly basing that comment on other designers who have used the 27TDFC with good effect down that low. Be they MTM, 2 way or 3 way designs. I always thought power response was mainly affected by directivity / beaming. My knowledge of off axis behaviour / horizontal lobing with various crossover topologies is non-existant so I can't add any more. All I was suggesting is if things weren't working with the CA18 (assuming it is a midrange sound issue - I've got no idea as you haven't been able to say), is that the 27TDFC offers some flexibility in XO point.

This I dont understand, could you please explain?
Although your on-axis looks nice and flat maybe phase starts deviating from target BW above or below XO

Are you talking about minimum phase? If so, I dont have it. The measurements is taken from 1m, onaxis to the mid for both drivers, without moving the mic. Im sure you know how the phaseplot looks like then?

I have tried the CA18 OB, without the woofer connected. As you said it made more problems so I didnt try it any further.

And once again; can you explain HD testing?

As Brandon noted. It is good to also take off axis measurements, especially on the angle you intend to listen to in room. It is not unusual to see +/- 3dB variations in power response. Of course I assume you are listening to the one speaker on axis anyway, so your curves are indicative of what you hear.

My comment was around seeing how phase tracks well above and below the crossover point. I remember reading an article by John Krekovsky who explained why this was important. I know with BW target slopes you aren't going to be phase aligned, I'm talking about phase deviations within the passbands of the drivers (ie. upto 40dB on either side of the XO). This would be indicative in amplitude summing anomalies.

HD = harmonic distortion testing. SW can do this. Not as nice as what Zaph produces out of SE and I don't think it will do the 5th harmonic.

At the end of the day it might just come down to voicing. Many designers tend to prefer one type of voicing. Troel's designs tend to taper down over the FR spectrum. Jay seems to prefer a 1 - 4KHz dip. Zaph states he prefers a flat reponse with low odd order harmonic distortion.

Get some cheapy electrolytic caps and small value inductors. You can then series / parallel these to tweak the response and voice to your preference.

When I eventually get around to buying the first crossover for my 3 ways - I am going all electrolytics (since my parts supply is pretty limited at the moment). I will only go poly when I finalise the XO. I just hope my partner doesn't mind wires and a naked crossover out the back of the speaker for a few months while I go through that process (I haven't told her yet :roll: )

Anyway.... there is nothing wrong with the drivers you have. No one has faulted your cabinet design so you will get there. It's just a matter of tweaking to get it sounding the way you want. :T

Cheers,
David.

[norcad]

I'm mainly basing that comment on other designers who have used the 27TDFC with good effect down that low. Be they MTM, 2 way or 3 way designs. I always thought power response was mainly affected by directivity / beaming. My knowledge of off axis behaviour / horizontal lobing with various crossover topologies is non-existant so I can't add any more. All I was suggesting is if things weren't working with the CA18 (assuming it is a midrange sound issue - I've got no idea as you haven't been able to say), is that the 27TDFC offers some flexibility in XO point.



As Brandon noted. It is good to also take off axis measurements, especially on the angle you intend to listen to in room. It is not unusual to see +/- 3dB variations in power response. Of course I assume you are listening to the one speaker on axis anyway, so your curves are indicative of what you hear.

My comment was around seeing how phase tracks well above and below the crossover point. I remember reading an article by John Krekovsky who explained why this was important. I know with BW target slopes you aren't going to be phase aligned, I'm talking about phase deviations within the passbands of the drivers (ie. upto 40dB on either side of the XO). This would be indicative in amplitude summing anomalies.

HD = harmonic distortion testing. SW can do this. Not as nice as what Zaph produces out of SE and I don't think it will do the 5th harmonic.

At the end of the day it might just come down to voicing. Many designers tend to prefer one type of voicing. Troel's designs tend to taper down over the FR spectrum. Jay seems to prefer a 1 - 4KHz dip. Zaph states he prefers a flat reponse with low odd order harmonic distortion.

Get some cheapy electrolytic caps and small value inductors. You can then series / parallel these to tweak the response and voice to your preference.

When I eventually get around to buying the first crossover for my 3 ways - I am going all electrolytics (since my parts supply is pretty limited at the moment). I will only go poly when I finalise the XO. I just hope my partner doesn't mind wires and a naked crossover out the back of the speaker for a few months while I go through that process (I haven't told her yet :roll: )

Anyway.... there is nothing wrong with the drivers you have. No one has faulted your cabinet design so you will get there. It's just a matter of tweaking to get it sounding the way you want. :T

Cheers,
David.

The onaxis vs power response isnt just about listening angle. Even though you are listening onaxis, the power response will affect with the room reflections, and can make a lot of trouble. I know!
Like I said, I have done a lot of off axis measurements, with Arta. Often in "2ch spectrum mode", and just walking around the speaker with the mic, it is very interesting to see. But I have never tried the measurements in the xo calculations.

If you look at my latest posted measurements, you will see that the drivers response is 6dB down at the xo point, and even though the LP has an 3.order BW slope, I do get an reverse nill. So the phase tracking around the xo point is rather good, but not "by the book" maybe?

Are you sure SW can do distortion measurements? How? Where?
I have done some HD ( ) within Arta, but I really dont know how to read the plots. What is good, and what is bad? And were is the limits?

Caps, resistors and coils isnt any problem, I have bought a couple of every value from 3,3-12 uF, 1-12ohm and 0,3-1,2mH.

I'm continuing tweaking and testing, thats for sure.

[Dave Bullet]

Are you sure SW can do distortion measurements? How? Where?
I have done some HD ( ) within Arta, but I really dont know how to read the plots. What is good, and what is bad? And were is the limits?

You have to create a new signal - then it is under the measure menu. You can measure harmonic and intermodular distortion. SW accidentally reports the Y axis as dB - when infact it is percentage distortion. I vaguely recall -40dB from nominal output is 10% distortion and -80dB is 1% distortion. And of course our ability to hear distortion (like amplitude response) varies with frequency... so what is audible ...

[norcad]

You have to create a new signal - then it is under the measure menu. You can measure harmonic and intermodular distortion. SW accidentally reports the Y axis as dB - when infact it is percentage distortion. I vaguely recall -40dB from nominal output is 10% distortion and -80dB is 1% distortion. And of course our ability to hear distortion (like amplitude response) varies with frequency... so what is audible ...

8O 8O 8O ops:

Thank you Dave!
I just cant understand why I havent seen that before!
Thats something I must test in mye next measurement session.
BTW I do understand how to read the plot, what I dont know is what is good or bad, like you said, what is audible.........

[Saurav]

There's a Speaker Workshop manual that's about 7 or 8 large Word documents in a big ZIP file. I think Claudio's website has a link. If you don't have that, it might be worth downloading and reading. It goes into a lot of stuff that SW can do, but it's not obvious how to do it.

[Dennis H]

Distortion numbers:
-20 dB = 10%
-40 dB = 1%
-60 dB = .1%
etc.

PS, Norcad you could always use ARTA (Steps) and just save a pic. It's not like you need to save the raw file.

[norcad]

Time to move on....
An adapter ring of Lexan was made, and the MCA15RCY took the job as mid, instead of CA18RNX.
New measurements was taken, and ones again I sat down trying to build a proper xo.
I tried both 2. and 3. order, but didnt get the phasetracking right.
After a little study of the phase plot, I was surprised!

And when I simulated a first order BW, I was even more surprised!

Just look at this plots:

[norcad]

Some comments:
- The Mca15rcy has a nearly incredible flat FR, better than several fullrange drivers.
- First order network isnt my first choice, but this was a surprise that I want to investigate a little further.
- The impedanse and electric phase is just perfect.
- The sensitivity is pretty good.
- The systems FR isnt that perfect As you can see, it has a step in FR from 1,8 to 2,6khz. Measured with Arta, who has a better resolution, its even bigger, and it is on both mid and tweet. This s diffraction from the frame.
- The FR also has to much output in the 3 to 6khz region.
- Its not theoretical correct first order. The drivers are 6dB down at the xo point, with 0 degr phase between drivers. That makes the power respons a little different, but I'm not sure if its a bad thing.

The conclusion? Lets rebuild the baffle once again!

[norcad]

The extra (new!) baffle was removed, and a big hole was cut in the front baffle.



A new piece of chipboard(?) was glued into the baffle



And after a lot of dust............




The midrange is now Seas MCA15RCY, and the tweeter is 27TFFC, just because this has a little higher sensitivity than the 27TDFC.

[norcad]

Several measurements later, and with a heavy filled mid enclosure, I came up with this crossover:




And the measured FR looks like this:




The impedance and phase:




And finally, the CSD plot:

[ThomasW]

Several measurments later, and with a heavy filled mid enclosure, I came up with this crossover:

First off the XO contains way to many components...

Plots look pretty good..... :T

How does it sound?

[norcad]

First off the XO contains way to many components...

Plots look pretty good..... :T

How does it sound?

Yeah, it will be very expensive with all this components, but good sound costs, right?

With this xo its sounds a little better than good! It sounds great until I play (very) loud, then its a little harch, and the ears get tired after a while.

After a while, I replaced the 4,7uF cond. with 3,9uF, and now my ears like it even better.

Here you can see how its dampening the 3 to 6khz area a little more:

[Winter]

Norcad,

If you are seriously considering a 1st order electrical crossover, I recommend installing a tweeter resonance impedance compensation network, a series LCR circuit in parallel with the tweeter. This should provide a smoother sound.

You may try a 25 ohm or so resistor in parallel with the tweeter for improved damping. You may need to lower the 1 ohm resistor to 1/2 ohm or so. Back in the old days of testing speakers with a function generator and oscilloscope (before LSPcad and other computer programs), getting a higher voltage out of tweeter connections in the upper frequencies than you were putting into the well damped (low Q) crossover circuit was a little dis-concerning. Any parallel resistance across the tweeter would remedy this.

[norcad]

Norcad,

If you are seriously considering a 1st order electrical crossover, I recommend installing a tweeter resonance impedance compensation network, a series LCR circuit in parallel with the tweeter. This should provide a smoother sound.

You may try a 25 ohm or so resistor in parallel with the tweeter for improved damping. You may need to lower the 1 ohm resistor to 1/2 ohm or so. Back in the old days of testing speakers with a function generator and oscilloscope (before LSPcad and other computer programs), getting a higher voltage out of tweeter connections in the upper frequencies than you were putting into the well damped (low Q) crossover circuit was a little dis-concerning. Any parallel resistance across the tweeter would remedy this.

I'm not considering, I'm using..........
Thanks for your tip!

When simulating in SW, I didnt see the benefit of this, it messed up the phase, so I didnt try it. But maybe it still will sound better?
I will try it as soon as the speakers are finsished. I'm painting them right now.

[norcad]

Finally! The speakers are now finished. (I hope) After a lot of "boiling" paint, and me trying to use a paint brush for the first time 8O . I finally was able to get a result that could stay in my living room. I also did some tweaking to get the sound that I wanted. And now I am pretty close to the goal I think.

Lets start with the inside, where you can see the bracing, mid-enclosure, vent and the drivers in place




I did several attempts with different dampening material, where to put it, and how much. I ended up with a lot! The mid-enclosure got 5cm thick "house insulation" on both sides, bottom and the back plate. On the back plate, right behind the bass driver I also put 5cm. And on both side of the mid enclosure, and both side of the BR tube, I filled it up. I forgot to take pictures, but you can see what I try to explain here:




The back plate also got 5mm bitumen, like this:




And finally, after the paintjob was done, it looks like this:

[A9X]

Very nice looking speaker there Norcad. Congrats.

[norcad]

With everything in place, I took some new measurements, and did a little adjustment on the 1.order network. I was trying to get the flattest on axis frequency response, and so close to first order roll off as possible, without using to many components.

The xo ended like this:


A bigger inductor on the mid will give a lower sensitivity, and the little inductor on the tweet takes down the peak around 14-15khz


Measured FR at one meter, with 4mS gating:




This did sound good, and I played music with it a couple of days. But as before it wasnt to good when I played really load. And that is what I often do!
So once again I sat down and tried to simulate a new xo!
And after a while I finally did come up with something interesting.
With only 6 components, (like the 1.order xo) Ive got a very good 3.order xo! 8O




Measured on axis FR with 4mS gating:




Reversed tweeter:

[norcad]

And some final measurements:

FR at one meter, 4mS gating without any smoothing:




Impedance and electric phase:





The XO and measurements are just mid and tweet, because I'm using an active xo between mid and bass. The lower xo point are about 340hz, and all the plots are 200-20,000hz
Because of the wide baffle that pushes the baffle step down to the lower xo point, and the use of CA26RFX with a higher sensitivity, I don't need any BSC at all.
The speakers are powered with two Denon POA-2800 stereo power amps, who pushes 200w@8ohm.
The bass drivers handles the power very well, and with Fb at 28-29hz, I've got -3dB at 37hz. Even that the vent is a little on the small side (diam 75mm) I have never heard any whistling from it.

After about a year of trial and error, I finally can sit down, relax and enjoy the music.

[Ray Collins]

Congratulations on the birth of your new baby...

Ray