-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
2-way with wide dispersion and low distortion. Linear CTA-2034. Mechano26
- Thread starter XMechanik
- Start date
- Thread Starter
- #22
For anyone interested in experimenting with this design, I planned to gradually update the leading post post with additional information. I wanted to start with LF measurements, simulations, and a spec comparison. For some reason I can't edit the leading post, so I'm puting the update here.
NF measurements and merging port with cone response
Comparision of simulated (yellow) and NF-measured response (red). Shifted to overlap at 80-90Hz where the filter response is close to 0dB
Possible reason of the discrepancy: the driver parameters given in the specification differ from the parameters of my driver copies, for example, the free air resonance seems to be around 56 Hz compared to 35 Hz given in the specification.
And BTW, note that enclosure dimensions are diffrerent from the enclosure of the previous Mechanos (the enclosure is scaled up to get close to 10l of raw volume).
NF measurements and merging port with cone response
Comparision of simulated (yellow) and NF-measured response (red). Shifted to overlap at 80-90Hz where the filter response is close to 0dB
Possible reason of the discrepancy: the driver parameters given in the specification differ from the parameters of my driver copies, for example, the free air resonance seems to be around 56 Hz compared to 35 Hz given in the specification.
And BTW, note that enclosure dimensions are diffrerent from the enclosure of the previous Mechanos (the enclosure is scaled up to get close to 10l of raw volume).
I figured as much, and I'm happy to help test anything out.
That is quite concerning, I would expect the tolerance to be much tighter than that. I would argue that explains the bass response difference. I wonder if it would be worth it to move up a model to their aluminum/ceramic sb15 drivers, I see those used a lot with success so perhaps they have better tolerances and might get you close to that 'end game' sound you mentioned. Still, that's just way off. I've run into 5-10hz off but 20hz off is nothing something I've seen yet.
I notice this design seems to have out of band response byproducts on the woofer showing up at a fairly high amplitude than I'm used to seeing, I see some small peaks at 8-10khz is about 30db down. Do you feel this is audibly adequate suppression?
Last edited:
- Thread Starter
- #24
I have to reiterate that the entire design idea is based on the MFC driver and its unique dispersion properties. Let's take a look at its characteristics around typical crossover frequency of around 3 kHz and compare it to the coresponding aluminum (NBAC) driver.
The MFC driver radiates broadly at this frequency, unlike the NBAC aluminum driver, which has noticeable directivity at even lower frequencies and would likely require a waveguide on the tweeter to match it.
Let's also look at the dispersion of the dome used – at 3 kHz it differs little from what is shown in the MFC characteristic, it looks like the waveguide won't be needed here at all!
Sorry, but the way you "braced" the cabinet shows a deep misunderstanding of how speaker housings resonate and how to reduce this problem. The edges you "x-braced" are the most solid points of a wooden box.
Your x-over is a typical result of too much simulating. You can get a very close results with 1/3 or 1/2 of the components you use. While a 18 parts crossover may look very important, in most cases it doesn't help with the end result. Every extra part degrades the drivers quality. The art of passive crossover construction is not to use as many parts as you can fit inside the box, but to get a perfect result with the least amount of components. Some commercial DIYS speaker designers, that make most of their profit from selling the x-over components, may have an interest in a huge parts count. This doesn't make a better speaker.
In your case, with such a well behaved Polypropylene woofer and this excelent tweeter, a simple crossover should not be a problem. In fact Polypropylene is used for woofer cones to make them simpler to handle.
If your real speaker measures very complicated, you should improve the baffle step and edge diffractions by modifying the cabinet and think over the driver placement.
Testing different baffles may seem complicated. I use XPS panels for test baffles. You can cut off and glue on anything you like in a few seconds, during measurements. A knife or much better, a hot wire, is the tool for this. Much simpler than wood, cleaner, silent and faster.
Do not rely on simulation only, it may lead you away from more important parts of such a construction.
There still is a real world beside of simulation, even as this soon will not matter any more, once AI has taken over any decission making.
Your x-over is a typical result of too much simulating. You can get a very close results with 1/3 or 1/2 of the components you use. While a 18 parts crossover may look very important, in most cases it doesn't help with the end result. Every extra part degrades the drivers quality. The art of passive crossover construction is not to use as many parts as you can fit inside the box, but to get a perfect result with the least amount of components. Some commercial DIYS speaker designers, that make most of their profit from selling the x-over components, may have an interest in a huge parts count. This doesn't make a better speaker.
In your case, with such a well behaved Polypropylene woofer and this excelent tweeter, a simple crossover should not be a problem. In fact Polypropylene is used for woofer cones to make them simpler to handle.
If your real speaker measures very complicated, you should improve the baffle step and edge diffractions by modifying the cabinet and think over the driver placement.
Testing different baffles may seem complicated. I use XPS panels for test baffles. You can cut off and glue on anything you like in a few seconds, during measurements. A knife or much better, a hot wire, is the tool for this. Much simpler than wood, cleaner, silent and faster.
Do not rely on simulation only, it may lead you away from more important parts of such a construction.
There still is a real world beside of simulation, even as this soon will not matter any more, once AI has taken over any decission making.
- Thread Starter
- #27
Well, I'm no expert in enclosure bracing, that's true. I had some scraps and I thought that either I throw them away for waste or use them to improvise some kind of reinforcement. The final bracing is slightly different from that on the photo and turned out to be surprisingly effective. The enclosure resonance problem is practically nonexistent in this design (comparing to previous ones).
Regarding the complexity of the crossover, it all depends on what you want to achieve. I made some assumptions in this regard. The circuit increasing the filter output impedance in the breakup region consists of three components. This allows for a THD reduction of <1% at 96 dB. The equalization circuit, which allows for bringing the PR slope closer to the target value of 1.2 dB/oct, consists of another three components. Based on measurements (not simulations), the woofer's response was not at all easy to manage, see the screenshot below.
I doubt that moving the woofer a centimeter this or that way would make much difference (unlike the tweeter, which position was carefully chosen for DI equalization). The frequency response equalization in the pass-band requires another couple of components, and breakup attenuation to the target level of 30 dB below the system spl - additional components.
But I don't want to convince anyone that this is the only approach. I'll just repeat that it depends on what you want to achieve. I made certain demanding assumptions, and to achieve them, such a complicated crossover doesn't seem unreasonable to me. And assembling a dozen or so components, is this really a problem if you get what you expected?
On a different note, let me just again ephasize that I chose polypropylene drivers for this design not because of the ease of crossover application, but because of their dispersion properties in the crossover frequency region, which provide the potential for linear directivity matching without the need for a waveguide on the tweeter (and therefore with wide dispersion). The idea explained in #24.
Regarding the complexity of the crossover, it all depends on what you want to achieve. I made some assumptions in this regard. The circuit increasing the filter output impedance in the breakup region consists of three components. This allows for a THD reduction of <1% at 96 dB. The equalization circuit, which allows for bringing the PR slope closer to the target value of 1.2 dB/oct, consists of another three components. Based on measurements (not simulations), the woofer's response was not at all easy to manage, see the screenshot below.
I doubt that moving the woofer a centimeter this or that way would make much difference (unlike the tweeter, which position was carefully chosen for DI equalization). The frequency response equalization in the pass-band requires another couple of components, and breakup attenuation to the target level of 30 dB below the system spl - additional components.
But I don't want to convince anyone that this is the only approach. I'll just repeat that it depends on what you want to achieve. I made certain demanding assumptions, and to achieve them, such a complicated crossover doesn't seem unreasonable to me. And assembling a dozen or so components, is this really a problem if you get what you expected?
On a different note, let me just again ephasize that I chose polypropylene drivers for this design not because of the ease of crossover application, but because of their dispersion properties in the crossover frequency region, which provide the potential for linear directivity matching without the need for a waveguide on the tweeter (and therefore with wide dispersion). The idea explained in #24.
Last edited:
You are very convinced of your construction, I have no problem with this. On the other hand, when others copy your design uncritically, they may be very disapointed after spending quite some money. Your measurements are not complete, so there is no way to understand what lead you to this construction. A clean documentation shows raw chassis response on the planed cabinet and the result of different stages of crossover construction.
It is very questionable to simulate the last decibel out of a response with tons of passive parts, just because you can.
In a real room situation, very little of this theoretical linearity will be left anyway.
The 100 Ohm resistor you use is something you only find in a simulation that went too far. It has no audible impact. Just try.
If you don't understand how a crossover works, simulations make you do stupid things.
It is very questionable to simulate the last decibel out of a response with tons of passive parts, just because you can.
In a real room situation, very little of this theoretical linearity will be left anyway.
The 100 Ohm resistor you use is something you only find in a simulation that went too far. It has no audible impact. Just try.
If you don't understand how a crossover works, simulations make you do stupid things.
a4eaudio
Addicted to Fun and Learning
The 100 Ohm resistor in that spot is not that uncommon. There is a Purifi application note that shows the parallel notch filter in that position, taming the cone breakup, reduces harmonic distortion whereas the commonly used series notch filter that tames the breakup to a similar degree does not affect HD. Below is a similar filter (from someone else's design) with a 30 ohm resistor in blue and the actual 82 ohm resistor that was used in pink. It reduces the cone breakup by an additional 6 dB. Whether or not this is actually need, or how audible, in this case I don't know, but there is nothing wrong with using that filter and the 100 ohm resistor.
eddantes
Addicted to Fun and Learning
- Joined
- May 15, 2020
- Messages
- 952
- Likes
- 2,089
Um guy... this ain't for sale... this is just some person sharing their project. He has no obligation to post anything... at all... What ever he does post is purely out of kindness and to share in a mutual interest with all of us.
If you are so inclined to have this design taken to the Nth level - feel free to build and measure as you deem fit. You can then share it here, or not - as that would also be only done out of kindness and not obligation.
Obligation - thats they key word here - he owes nothing to anyone.
Cheers!
One more thing to add - he already has a design fully measured on the Klippel. And that sim turned out quite alright in reality.
Last edited:
All I wanted to point out is that you do not need 18 x-over parts for a simple 2-way with two extremely linear and well behaved drivers.
Some playing around with baffle dimensions and driver positions should reduce the parts count, which should usualy be better for a dynamic speaker. Or simple: too many parts, no good. Make your homework with the baffle and don't make the crossover correct it.
Some playing around with baffle dimensions and driver positions should reduce the parts count, which should usualy be better for a dynamic speaker. Or simple: too many parts, no good. Make your homework with the baffle and don't make the crossover correct it.
eddantes
Addicted to Fun and Learning
- Joined
- May 15, 2020
- Messages
- 952
- Likes
- 2,089
Ok - of course. Thats a message one can receive... But that original one - sheesh.
If there any evidence that lower part counts for xovers result in anything more than a xover that costs less and might make for a more appealing build to anyone who might want to build the speaker?
Seems that high xover part count works really well for speakers like ascilab. I don't really see any issues with their speakers that would come from their xovers.
Just seems like another hold over from the old days. No shortage of that in the diy world.
Jim Taylor
Major Contributor
- Joined
- Apr 7, 2024
- Messages
- 1,236
- Likes
- 4,045
Are you not free to take the same drivers and design a box and crossover that aligns with your stated parameters? Please do. You could teach us all a great deal here. I'm sure we all would appreciate it.
Those drivers are very promising. You sure can build a very good speaker from the pair. I'm simply not 100% convinced the complicated crossover is neccesary as it relies too much on simulations. You could improve it by using a passive radiator instead of a vent.
I know they were doing a site upgrade a few days ago. I think there are still a few bugs but I was just now able to put the tweeter and woofer in my cart.
Good to know.
I only looked as I'm going to attempt to use the aluminum version as I already have it, usually tweeter swaps are a bad idea but in this case it seems fine. Sims are telling me that simply adding a 1ohm resistor in series before the stock xover components should account for the 1db sensitivity difference. I'll be sure to measure to find out.
Hopefully have the cabinets done this week, woofers here next week. Have to source a few xover parts like that huge 560uf cap. My mechano23 build involved layering 9mm ply as that's all I had. It was an awful process so I scrapped what I had already cut and bought some new 18mm ply. Not exactly a budget built at around $500 for me in all, but I'll have enough wood left over to make another bookshelf. Eager to share subjective comparisons between the 23 and 26. Will I be the first person to have two mechano designs?!
Similar threads
- Poll
- Poll
