Mechano23 Open-source DIY Speaker Review

[Maiky76]

I have played with an active version of the Xover i.e. just fit the passive XOver transfer functions with a DSP.
Here are the results FYI. It could be used on a MiniDSP HD.
The 44.8Hz biquad can be omitted, it was easier to fit the TF with it...

And the difference on axis

Edit:
For the more advanced users I would probably advise the following:
- keep a notch filter on the Woofer in the analogue domain. That can improve distortion; a good illustration can be found here
Another benefit is to maintain a small equivalent parasitic series resistor that will help to keep the box tuning in the ballpark.
- keep a capacitor on the tweeter for protection against "pop" at start-up of the amp
- Refit the digital filter with the resulting system.

example, I did not check if the components were available

 

[Chris*42]

Thanks for the info! It will be my first DIY speaker ever. How do you prevent the stuffing from touching the rear of the driver/PR? Or doesn't that matter at all.
And do you think the improved distortion result has to do with the inceased baffle thickness, the use of the PR or lining the inner walls with the foam?

Polyester wool is available in mats with different thicknesses. You may cut these pieces to the size you need, so this point doesn‘t matter.

IMO the use of passive radiators is the better (but more expensive) choice for small cabinets with low tuning frequency. You avoid the problems with the tube design (resonances, air speed may be to high,…), see also the discussion in the mechano23 thread.

The other changes do not have such a big influence. This is rather my personal style in the design of speakers...

 

[Chris*42]

How do you keep the wire from rattling or buzzing?

I fix each wire in a separate slit inside the damping material (polyurethane foam is stiff enough to hold the wire). I have made bad experiences with flexible wire in small cabinets. Flexible wire always finds a way to hit somewhere during operation.

 

[marc_n]

…and this is the result. I build up these speakers with a little variation (passive radiators SB15SFCR-00 on the left/right side panel instead of BR vent on the back panel) to avoid the resonance (vent at 1 kHz) and let me easily change the tuning frequency. Without the additional mass, the system (PR and cabinet of ca. 7.5l) has a tuning frequency of 53Hz. Adding ca. 31g will result in a Helmholtz frequency round about 45Hz - this will be the next step in my experiments.

My instant impression of these speakers was, that they sound so very natural and absolut correct! Even the upper/mid bass is fine, as long as you don’t set the volume too high. These are really amazing speakers at this price level!

I can only repeat the congratulations to @XMechanik for sharing his construction and @amirm for making and publishing the measurements!

Nice build – and great woodwork! I have some cherry wood waiting ...

How did you cut the slot for the PR? I understand how to sink the round drivers into the baffle with a router and a circle jig, but am not sure how to do this with the elongated PR slot.

BTW: Really inspiring to see this community effort!

 

[Ktacos]

Looks like a hair got on the baffle when finishing, might as well just throw it away.


I kid of course.

 

[Chris*42]

Nice build – and great woodwork! I have some cherry wood waiting ...

How did you cut the slot for the PR? I understand how to sink the round drivers into the baffle with a router and a circle jig, but am not sure how to do this with the elongated PR slot.

BTW: Really inspiring to see this community effort!

The shape of this PR is composed of two semicircles. So you can mill them with the necessary distance and connect them with a straight line.

Good luck with your woodwork. Cherry wood will certainly look very good too!

 

[marc_n]

I have played with an active version of the Xover i.e. just fit the passive XOver transfer functions with a DSP.
Here are the results FYI. It could be used on a MiniDSP HD.
The 44.8Hz biquad can be omitted, it was easier to fit the TF with it...

View attachment 373274

And the difference on axis
View attachment 373273
Edit:
For the more advanced users I would probably advise the following:
- keep a notch filter on the Woofer in the analogue domain. That can improve distortion; a good illustration can be found here
Another benefit is to maintain a small equivalent parasitic series resistor that will help to keep the box tuning in the ballpark.
- keep a capacitor on the tweeter for protection against "pop" at start-up of the amp
- Refit the digital filter with the resulting system.

example, I did not check if the components were available
View attachment 373315

View attachment 373314

Thanks for posting this version with an active XOver – quite an interesting option.

Just to verify the component list for the second variant (it is a bit hard to decipher from the image):

• notch filter CLR
C = 1,1 µF
L = 470 µH, R = 0,28 Ω
R = 91,0 Ω
C = 11,0 µF
R = 1,2 Ω

• tweeter protection
C = 5,6 µF

Are these indeed the values?

The parts can be sourced according to the above specs, except for the coil (from what I can see).
Would a coil with R=0,26 Ω be still within specs?

 

[Mr.Christiee]

The baffles are quite narrow . edge rounding should be at a fairly large radius to be worthwhile ?

Buts it’s DIY someone will surely do this to with slightly altered cabinet dimensions and subsequently tweaked filters to accommodate that .

But the original design smartly ticks so many of the 1000’s compromises needed to make a speaker at a certain price point and size . This a very attractive proposition. Impressive !

Add a sub and some cheap amp and go very reasonable priced hifi

Hey, could you please elaborate on the edge rounding/altered cabinet dimensions. I'm thinking about building these very soon. Thanks!

 

[Mr.Christiee]

What frequency would you set the high-pass filter for these?

 

[Salt]

Don't bother with roundings all around, flatten the area around the tweeter at 25 degree like this:

 

[Mr.Christiee]

Don't bother with roundings all around, flatten the area around the tweeter at 25 degree like this:
View attachment 376211

Is this sonically better than rounded edges?

 

[Salt]

Yes.
It gives a wider dispersion of reflections on the front and thus flattens FR.

 

[Mr.Christiee]

Yes.
It gives a wider dispersion of reflections on the front and thus flattens FR.

Okay I see. Bevel the edge as close to the tweeter waveguide as possible?

 

[Salt]

Yes, but not beyond (*as Einstein would have said*)

 

[Salt]

Yes, but not beyond (*as Einstein would have said*)

I use a grinder (band rsp. swing) for this job, no milling.

 

[Mr.Christiee]

I use a grinder (band rsp. swing) for this job, no milling.

I guess you'd need to use an extra thick front baffle to do this? 2-3 inches?

 

[Salt]

2" is more than enough.
The example in the image uses 25 mm (approx. 1")

 

[Maiky76]

Thanks for posting this version with an active XOver – quite an interesting option.

Just to verify the component list for the second variant (it is a bit hard to decipher from the image):

• notch filter CLR
C = 1,1 µF
L = 470 µH, R = 0,28 Ω
R = 91,0 Ω
C = 11,0 µF
R = 1,2 Ω

• tweeter protection
C = 5,6 µF

Are these indeed the values?

The parts can be sourced according to the above specs, except for the coil (from what I can see).
Would a coil with R=0,26 Ω be still within specs?

I hope this is more readable
The 44.8Hz EQ can be removed.

And a simplified version

The original Vcad project is available, somewhere.
I encourage you to do the simulation and see for yourself the impact of any variation vs these designs and re-optimize the fit with what you have available.
WRT 0.26R vs 0.28R IMO, the difference will be buried under the tolerances of the drivers and the actual values of the passive components.

For the active part you'll have to double check the actual response of the DSP vs the simulations; not all DSPs have the same definition for the biquads.
Vcad should be able to export in MiniDSP advance format.

 

[kimmosto]

From VituixCAD user manual:

 

[marc_n]

I hope this is more readable
The 44.8Hz EQ can be removed.

View attachment 376307

And a simplified version
View attachment 376308

The original Vcad project is available, somewhere.
I encourage you to do the simulation and see for yourself the impact of any variation vs these designs and re-optimize the fit with what you have available.
WRT 0.26R vs 0.28R IMO, the difference will be buried under the tolerances of the drivers and the actual values of the passive components.

For the active part you'll have to double check the actual response of the DSP vs the simulations; not all DSPs have the same definition for the biquads.
Vcad should be able to export in MiniDSP advance format.

This is very useful.

Thanks a lot for clarifying this!