First DIY !

[Nordic]

And when listeing, it is even harder to agree on that point.

 

[gareth]

i wouldnt base my choice of crossover point on a certain sounds frequency range. base it on the drivers and their optimum operation ranges, everything else would be an intentional design error. especially with those shallow analog crossovers, it is difficult enough to find a good point, where both drivers add just fine.

Hi MaVo, do you think that 6dB/octave have too shallow attenuation at the crossover point? Perhaps you think I should move to a higher order
maybe 18dB/octave?

I was hoping to achieve as smooth an integration as is reasonably possible with these drivers and also, I wonder if the steeper roll-off will be more noticable to my ears?

What do you think?

Thanks
Gareth

 

[gareth]

i wouldnt base my choice of crossover point on a certain sounds frequency range. base it on the drivers and their optimum operation ranges, everything else would be an intentional design error. especially with those shallow analog crossovers, it is difficult enough to find a good point, where both drivers add just fine.

Also, do you think that bi-amping with electronic crossovers is a better idea?

I know that may sound a little crazy for a beginner but I am well aware and have heard the benefits of active speakers. The reason I say this is that I have a lot of BJt's at my disposal (for which there is another thread titled 'Headphone amp help') and would really like to build this sort of system.

There is a saying about learning to walk before trying to run, I know but I am optimistic.

Thanks
Gareth

 

[MaVo]

I dont have much experience with good analog crossovers, only with bad premade ones, since i converted to digital crossover and active amplification before i had a chance to learn more about the subject. That beeing said, there probably are really good passive crossovers, which can achieve the same as digital ones. But the design of something like that is for the pros among us. I would not dare to try it. Maybe if i had some months to spare for learning the background and playing alot in simulation tools.

From my experience, the less drivers you have, the steeper the crossover should be, since each one allready has to handle quite a lot of the spectrum. I decided to go 4 way with my own speakers to have each driver handle its optimum range, and plan to use 24db crossover. I mean theoretically, you could even do 2 way with 6 db crossovers, its just that the steeper the crossover are, the cleaner the sound will be, since less excess energy goes into the driver and less driver to driver interference.

If you would go active, you have the advantage of beeing able to experiment alot. Thats good when you want to learn and frees you of the hassle of designing a passive network. And a behringer crossover with two amps is not so expensive after all.

If you just want good sound, go with a prooven design, since there will be problems, which you can only realize after building your own design. But building your own is definitely more fun, as it is completely yours. But you probably wont stop with this design, since there is allways something to improve (it is also an advantage of active speakers, that you can reuse the crossover for the next project) 🙂

 

[hermanv]

I dont have much experience with good analog crossovers, only with bad premade ones, since i converted to digital crossover and active amplification before i had a chance to learn more about the subject. That beeing said, there probably are really good passive crossovers, which can achieve the same as digital ones. But the design of something like that is for the pros among us. I would not dare to try it. Maybe if i had some months to spare for learning the background and playing alot in simulation tools.

From my experience, the less drivers you have, the steeper the crossover should be, since each one allready has to handle quite a lot of the spectrum. I decided to go 4 way with my own speakers to have each driver handle its optimum range, and plan to use 24db crossover. I mean theoretically, you could even do 2 way with 6 db crossovers, its just that the steeper the crossover are, the cleaner the sound will be, since less excess energy goes into the driver and less driver to driver interference.

If you would go active, you have the advantage of beeing able to experiment alot. Thats good when you want to learn and frees you of the hassle of designing a passive network. And a behringer crossover with two amps is not so expensive after all.

If you just want good sound, go with a prooven design, since there will be problems, which you can only realize after building your own design. But building your own is definitely more fun, as it is completely yours. But you probably wont stop with this design, since there is allways something to improve (it is also an advantage of active speakers, that you can reuse the crossover for the next project) 🙂

Several points:
1. I agree that fewer drivers demand more from each one and higher order crossovers seem like a better choice.
2. The more octaves you ask a driver to deliver the worse the phase performance will be at the extremes.
3. An active crossover/biamp scheme makes adjusting the crossover frequency fairly easy, but tends to ignore the phase problem.
4. Some speaker software includes phase in predicting the acoustic output of the complete design, the behavior of these crossovers will probably not seem intuitive to a newbie.
5. Good passive crossovers use many parts and aren't cheap, on the other hand biamping can hardly be considered cheap either.

I still think you start with a kit. Since in most cases the box dimensions are largely controlled by the chosen woofer, you can always re-use the box for future design improvements whether they be passive or active. If you are serious about this and plan on building more than one design, seriously consider purchasing Vance Dickason's "Loudspeaker Design Cookbook".

If you only plan to do this once, stick with a kit. The learning curve is too steep to absorb in one pass.

 

[MaVo]

hermanv, i agree with your agreement 😀

i would like to add, getting phase right with passive crossovers is really hard. you have to measure the real life response, as not only the drivers have an inherend phase change, but also a delay. acording to thomas danley, woofers may be virtually a few feet behind the midranges because of intrinsic delay. you dont know by how much, thus you will have to measure. additional to that, you wont know if the drier really behaves minimum phase all over its passband, so that simulations of phase are valid. this follows, that to get a passive speaker right with phase correction, you have to iterate the process, building, measuring, building again and so on until it is right. the same goes for active speakers. only digital speakers dont need the rebuilding phase, as you can easily adjust individual driver phase and delay in software. the digital behringer crossover even has an automatic function that does this for you, if you have a measurement mic hooked up to it. this way, alot of difficult design steps are not needed anymore.

 

[PeteMcK]

“a cabinet of 42litres was specified for a tuning frequency of about 37Hz” - I get 43 L @ 34Hz, not much in it, your alignment will have a tad more solidity in the lows.

I’d be inclined to build the boxes, make a simple bracket so you can try the tweeters on top, and cross over using 1st order (an inductor and a cap), but try this at LOW level, and see how that sounds...

The KLS9 xover – can’t get simpler than this:
http://www.audiodesignguide.com/doc/KLS9_loudspeakers_2.jpg

Here’s a 1st order xover (in French), looks like it needed a notch filter to deal with the impedance peak:
http://www.gbiloba.org/hmz1725.psp

(I’m not sure this tweeter, being non-ferrofluid, will handle 1st order that well... good alternatives are the SEAS 27tdfc or tbfc, low xover point and excellent value)

An Italian design with a simple 2nd order (bit hard to read)
http://www.audiokit.it/ITAENG/KitDiffusori/dappolito/RES/Brenda-2.gif

Here’s how one guy used your tweeter 2nd order:
http://www.fdj.org.uk/loudspeakers/av_speakers/index.html
I don’t like his attenuation circuit, but it’ll give you an idea of what’s involved

Let us know how you go with this...

 

[gareth]

�a cabinet of 42litres was specified for a tuning frequency of about 37Hz� - I get 43 L @ 34Hz, not much in it, your alignment will have a tad more solidity in the lows.

I�d be inclined to build the boxes, make a simple bracket so you can try the tweeters on top, and cross over using 1st order (an inductor and a cap), but try this at LOW level, and see how that sounds...

The KLS9 xover � can�t get simpler than this:
http://www.audiodesignguide.com/doc/KLS9_loudspeakers_2.jpg

Here�s a 1st order xover (in French), looks like it needed a notch filter to deal with the impedance peak:
http://www.gbiloba.org/hmz1725.psp

(I�m not sure this tweeter, being non-ferrofluid, will handle 1st order that well... good alternatives are the SEAS 27tdfc or tbfc, low xover point and excellent value)

An Italian design with a simple 2nd order (bit hard to read)
http://www.audiokit.it/ITAENG/KitDiffusori/dappolito/RES/Brenda-2.gif

Here�s how one guy used your tweeter 2nd order:
http://www.fdj.org.uk/loudspeakers/av_speakers/index.html
I don�t like his attenuation circuit, but it�ll give you an idea of what�s involved

Let us know how you go with this...

Thanks pete, I have just looked at some of the links and was kind of relieved in that there are similar figures to what I have worked out for different filters.

I am going for a bass reflex of about 42litres. These figures were suggested by WinISD. The front baffle will be 287mm wide by 500mm high and the depth of the cabinet will be about 400mm. Deducting the driver and port vulume should give me about 42litres.

I would now like to build suitable crossovers and I think I will try a first order to begin with and listen and see. A similar approach was used in the KLS9 which uses the same tweeter but slightly different bass unit, HM210 instead of AP210.

Also I would like to build some compensation circuit for the tweeter resonance but the supplied manufacturer data is a little conflicting. The quoted figures say that fs is at 976Hz but the corresponding graph shows a peak at approximately 2700Hz. I think I should go for the written data as the 2700Hz peak seems odd.

In one of the links, KLS9, it talks of the distance between drivers on the baffle. It says about 30mm driver difference at a crossover fequency of 3000Hz. How did he arrive at this distance?

 

[gareth]

I have been looking around at measurement microphones. There are quite a few available and using on of these with some software sounds like a good idea. There is behringer available for around £30 which combined with some software seems like a worthwile investment.

I have even seen a DIY microphone using a Panasonic electret condenser which should cost less than £10.

This is becoming a mission!

Any suggestions?

Thanks
Gareth

 

[Jay_WJ]

gareth,

Below is my crossover modeling result. In the following I'll explain how this has been obtained.

An externally hosted image should be here but it was not working when we last tested it.

To simulate in-box responses of the AP210Z0 woofer, I used Parts Express CLIO data (both amplitude and impedance), which are usually very accurate. Bass box and baffle diffaction/loss simulations were performed and added to the data. For the tweeter FR modeling, I traced the manufacturer's SPL plot. But I had to use my modeled impedance curve since the plot in the manufacturer's data sheet is obviously wrong. Then, baffle diffraction/loss effects were added to the FR data, too.

For the baffle diffraction/loss modeling, I used the same cabinet design as this:

http://members.shaw.ca/rtowsley/towers/cabinet1-Model.png

This cabinet will work well in your case, too. According to my Unibox simulation, 31 Hz tuning in a 45 to 47 liter box is suitable for your woofer. Note that for this modeling I used the average T/S parameters from both the manufacturer's and Parts Express sample data. So, my suggestion should be a safer bet. The above cabinet has a 57 liter net volume. So you'll need to reduce it by making the cabinet depth a little shallower. Low, 31 Hz tuning will give you lean but deep bass that will be flat in a moderate sized room. Too high or flat tuning can give boomy bass in an actual room. This 31 Hz tuning is obtained by using a 2" diameter, 4" (or 10 cm) long tube. Don't use a 3" diameter tube---2" is sufficient. The woofer only has a 2.6 mm xmax. You'll hear the woofer distort before hearing the port chuffing.

Now to the crossover. This is a basically Linkwitz-Riley acoustic 4th order design. Why 4th order? Take a look at the woofer and the tweeter's raw, in-box responses obtained from my simulations:

An externally hosted image should be here but it was not working when we last tested it.

Focus on the woofer's natural rolloff above 4 kHz. This is already an almost 3rd order slope, certainly greater than 2nd order. And note that a speaker's crossover filter is not "brick-wall" type. Its rolloff begins before the xover point and continues to roll off gradually far after the xover freq. In fact, the order of a filter's acoustic rolloff is defined by its rolloff rate at the curve's tail, not at the xover point. This means that even if you use an electrical 1st order filter for the woofer, you'll end up having an acoustic slope between 3rd and 4th order if your xover point is around 2 kHz. So, the design should aim either Butterworth 3rd order (BW3) or Linkwitz-Riley 4th order (LR4).

A BW3 design is not so popular these days since it results in a tilted vertical polar response around the xover frequency. Also, this has weaker tweeter protection than LR4.

So, our chice here is LR4. But an exact, theoretical LR4 design cannot be implemented in your case due to the size of a woofer. This is because the driver's acoustic offset (i.e., how far behind the driver's acoustic center is located) is quite large for an 8" woofer. An usual solution, when an flat baffle is used, is to use a bit shallower acoustic slope for either (or both) the woofer or the tweeter's rolloff. As you can see in the plot of individual driver rolloffs via crossover, each driver's rolloff around the xover point is a bit shallower than normal LR4 targets. But the curves reach 4th order slopes as they go farther from the xover freq.

The xover point occurs at 1.8 kHz. Around this frequency, phase tracking is very good with a deep reverse null.

A dip around 2.2 kHz in the system's frequency response is not from the crossover, but from the drivers' natural responses, as you can see from each driver's raw, in-box responses. Baffle diffraction effects also contribute to this dip. This cannot be filled in unless you want to use a really complex crossover. In fact, a dip around 2k to 4 kHz is good for non-fatiguing listening.

I suggest using an MT driver layout (i.e., tweeter is below the woofer) instead of the usual TM layout because the design axis of this crossover is the woofer axis, not the tweeter axis. Due to the deep acoustic center of this 8" woofer (and also due to the "bumped" flange of the tweeter), the acoustic offset between these two drivers is very large. Compensating this by the crossover is somewhat limited. So, listening at the woofer axis gives better phase alignment.

You'll need to experiment with the R2 value in the tweeter network to find a suitable tweeter level for you. I suggest trying values between 6 and 10 ohms. Values in this small range won't ruin intended phase tracking. So don't be afraid to tweak. A smaller value will decrease the tweeter level and a larger value will increase it.

Finally, the only issue of this design is the tweeter's power handling, which I'm not sure of since we don't have its distortion measurement. It seems that this tweeter has had good reputation. I hope it can handle a 1.8 kHz Fc in LR4. I actually tried to move it a bit higher, but concluded that this 1.8 kHz Fc is optimal for this woofer considering the 8" driver's beaming effect and steep, high-frequency natural rolloff. I also tried to use a notch filter at the tweeter's resonance (around 1 kHz) point. For this LR4 design via a 2nd order electrical filter on the tweeter, I concluded that a notch filter is not necessary. We can have the same amount of suppression simply by varing the component values in the 2nd order network.

If you're interested in how I design a loudspeaker crossover, take a look at my notes at my website:

http://www.geocities.com/woove99/Spkrbldg/DesigningXO.htm

Enjoy!

 

[gareth]

gareth,

Below is my crossover modeling result. In the following I'll explain how this has been obtained.

An externally hosted image should be here but it was not working when we last tested it.

To simulate in-box responses of the AP210Z0 woofer, I used Parts Express CLIO data (both amplitude and impedance), which are usually very accurate. Bass box and baffle diffaction/loss simulations were performed and added to the data. For the tweeter FR modeling, I traced the manufacturer's SPL plot. But I had to use my modeled impedance curve since the plot in the manufacturer's data sheet is obviously wrong. Then, baffle diffraction/loss effects were added to the FR data, too.

Thanks Jay, this is awesome! I cannot say how pleased I am that you have don this, thankyou very much.

Basically if I stick to your crossover schematic and fire them up I should have something pretty good? Do you know what sort of off- set I should be looking for between the tweeter and the bass? Is this critical? Also I feel that mounting the tweeter below will work well in my listening position as I can get the tweeters around ear level.

Also, do I need to compensate for the higher efficiency of the tweeter? 92dB as apposed to 89dB. Another point is the placement of the port. Do I mount it front or rear? My listening room is not huge so I don't want the port interacting with the wall they will be relatively close to.

Once again thankyou.
Gareth

 

[Jay_WJ]

Thanks Jay, this is awesome! I cannot say how pleased I am that you have don this, thankyou very much.

Basically if I stick to your crossover schematic and fire them up I should have something pretty good? Do you know what sort of off- set I should be looking for between the tweeter and the bass? Is this critical? Also I feel that mounting the tweeter below will work well in my listening position as I can get the tweeters around ear level.

Also, do I need to compensate for the higher efficiency of the tweeter? 92dB as apposed to 89dB. Another point is the placement of the port. Do I mount it front or rear? My listening room is not huge so I don't want the port interacting with the wall they will be relatively close to.

Once again thankyou.
Gareth

Oh I forgot to say that the tweeter had a 1" horizontal offset on the baffle I used for diffraction modeling. Other than this, I used the same (but inverted) baffle layout of the cabinet design I posted in my post. Go to this thread and see how this guy ended up:

http://www.diyaudio.com/forums/showthread.php?s=&threadid=111691

Don't worry about the tweeter efficiency. It's already taken into account in my crossover design. You'll only need to adjust the tweeter level a little bit by your ear according to my suggestion (R2 value between 6 and 10 ohms; default is 8 ohms).

Front porting seems to be good in your situation.

 

[gareth]

Oh I forgot to say that the tweeter had a 1" horizontal offset on the baffle I used for diffraction modeling. Other than this, I used the same (but inverted) baffle layout of the cabinet design I posted in my post. Go to this thread and see how this guy ended up:

http://www.diyaudio.com/forums/showthread.php?s=&threadid=111691

Don't worry about the tweeter efficiency. It's already taken into account in my crossover design. You'll only need to adjust the tweeter level a little bit by your ear according to my suggestion (R2 value between 6 and 10 ohms; default is 8 ohms).

Front porting seems to be good in your situation.

Excellent thankyou very much. I now need to order the parts for the crossover.

I am also wondering if it is worth getting a flush aluminium faceplate for the tweeter. Any ideas?

Thanks
Gareth

 

[Jay_WJ]

I strongly recommend flush mounting the tweeter in a recessed cutout.

 

[gareth]

I strongly recommend flush mounting the tweeter in a recessed cutout.

Do you mean to do this with the plastic faceplate supplied with the driver? It has chamfered edges so this would be awkward to achieve.

/`````````````````\
/___________________\

Does it matter what type of inductors are used? Should I use air/iron/ferrite? Or does it not make too much difference?

Thanks
Gareth

 

[Jay_WJ]

Do you mean to do this with the plastic faceplate supplied with the driver? It has chamfered edges so this would be awkward to achieve.

/`````````````````\
/___________________\

I looked at the spec sheet again. Indeed, the face plate was designed to work for surface mounting.

This doesn't help to reduce the large driver offset on the Z axis. Again, in this case, your ear level should be at the woofer axis, not at the tweeter axis. That's the reason why I suggested the inverted driver positions. But this doesn't mean that listening at the tweeter axis gives a horrible result. I just mean the ideal, in-phase listening axis will be the woofer axis.

Does it matter what type of inductors are used? Should I use air/iron/ferrite? Or does it not make too much difference?

I used .4 ohm DCR for L3 in my simulation. This is the DCR of a 1.5 mH 15 guage air core inductor. But you can use an iron core for even lower DCR. Unless you use a very thin (e.g., 22 guage) coil, you'll be fine. For L1, DCR doesn't matter much. Just use anything you can get cheap---my suggestion is 20 guage air core.

 

[gareth]

I looked at the spec sheet again. Indeed, the face plate was designed to work for surface mounting.

This doesn't help to reduce the large driver offset on the Z axis. Again, in this case, your ear level should be at the woofer axis, not at the tweeter axis. That's the reason why I suggested the inverted driver positions. But this doesn't mean that listening at the tweeter axis gives a horrible result. I just mean the ideal, in-phase listening axis will be the woofer axis.




I used .4 ohm DCR for L3 in my simulation. This is the DCR of a 1.5 mH 15 guage air core inductor. But you can use an iron core for even lower DCR. Unless you use a very thin (e.g., 22 guage) coil, you'll be fine. For L1, DCR doesn't matter much. Just use anything you can get cheap---my suggestion is 20 guage air core.

Thanks for your time Jay. I think I will stick to air-core as you suggest. Do cap's 1 and 3 need to be non-polarised ? Axial types maybe?

Thanks
Gareth

 

[Jay_WJ]

Thanks for your time Jay. I think I will stick to air-core as you suggest. Do cap's 1 and 3 need to be non-polarised ? Axial types maybe?

Thanks
Gareth

No components used in passive speaker crossovers have polarity. Use polypropylene caps (not expensive ones, of course) for C1 and C3. Use a cheap non-polarized electrolytic cap for C5. It doesn't have to be exactly 33 uF. Anything between 30 and 33 uF will do.

 

[gareth]

No components used in passive speaker crossovers have polarity. Use polypropylene caps (not expensive ones, of course) for C1 and C3. Use a cheap non-polarized electrolytic cap for C5. It doesn't have to be exactly 33 uF. Anything between 30 and 33 uF will do. [/QUOTE


Excellent, thank you. I will let you know the results when I get the components.

Thankyou
Gareth

 

[Jay_WJ]

Something like these for C1 and C3:

And for C5:

Let me know your listening impressions when you finish your build.