Open Source Monkey Box

Member
Joined 2005
Paid Member
This is a long thread with many twists and dead ends. Before diving in, take a look a this first:
  • The documentation files of the finished Open Source Monkey Coffin loudspeaker. I am (trying to) keep this up to date!
  • The notes at the end of this post, which have some useful hints to navigate the thread.

-----------------------------------------------------------

Following up on the earlier Open Source Speaker thread (part I and part II) I am starting a new thread to discuss the develpment of the "Open Source Monkey Coffin".

Here's an update of my earlier Monkey Coffin ideas. Sorry it's so long.

POINTS OF DEPARTURE (as I see them):
  • The first poll says it should be a three way, and it's okay to cost more than $500 per box.
  • The second poll says that the Monkey Box should be "amplifier friendly" (work well with ACAs or small tube amps, etc.), and it should not be too big (not more than 60-80 L).
  • The speaker should be straight forward to make for DIY novices (we don't want to scare them, we want them to run to the hardware shop and build stuff on their kitchen table). There is no point in designing for expert builders, because they will design their own speakers anyway.

MY IMPLEMENTATION IDEAS (I admit these are partially biased by my own preferences, because I need some motivation to actually put some work into this):
  • The box shoud be easy enough to make on a kitchen table. Dimensions roughly 30 x 65 x 30 cm or so (will depend on the requirements of the drivers and box tuning).
  • For "weak amp compatibility" I set the efficiency target to 92 dB/2.83V/1m, with bass extension to 45 Hz (-3 dB). Achieving this will not be very easy.
  • The impedance curve should be flatish and qualify as "8 Ohm". This means we can't "cheat" with the efficiency target by using low-impedance drivers.
  • Use high-quality drivers/parts with the right tech specs for the task. The look of the drivers has to be "right" for a HiFi system in a home environment, but is second priority after the tech specs (people may not want to build a speaker that does not look "right").
  • Woofer: The woofer will likely be the limiting factor that determines the size/bass-extension/efficiency compromise. I guess a 10" unit would be the right choice here. The earlier threads already mentioned a few units that might be interesting for this design (Fane Sovereign Pro 10-300SC, Ciare FXE 10-2.5, Volt BM251.3, some Beymas, etc.). We'll have to look at this some more.
  • Midrange: Needs to keep up with the SPL requirements, so many of the conventional "HiFi" midranges are out. I have some experience with the ATC SM75-150 dome, which is VERY good, expensive, and not available to DIYers anymore. When ATC stopped selling to the DIY market, Volt released the VM752 dome, which is a very good clone of the ATC driver. I really want to use the Volt VM752 dome. It can be used from its fs (400 Hz) up to about 3 kHz or so.
  • Tweeter: There are many good tweeters out there. I am a big fan of Be dome tweeters (ScanSpeak D3004/6640, SB Acoustics Satori TW29BN). I also read a bit about waveguides, which do good things. They help with matching the dispersion of the tweeter to the midrange, they reduce the effects of baffle diffraction, they increase the on-axis efficiency, and they reduce non-linear distortion of the tweeter (by reducing the drive voltage for a given SPL). At the moment I can't think of a reason why we should not use a waveguide.
  • Cross over filters: the x-over frequencies will depend quite a bit on the driver choices. However, if possible, I like to set the woofer/midrange x-over frequency such that it fits the baffle-step centre frequency. This makes it easier to compensate the baffle step by fiddling with the woofer SPL vs midrange SPL. If the drivers allow, I like using 6 dB/octave filters because it's easier to get the time-domain response of the loudspeaker right. In my experiments, my ears preferred 6 dB filters over 12 dB filters.
  • I will openly admit that costs are not my first priority. If the right parts cost a lot of money and there are no alternatives, so be it. The earlier threads indicated that it's ok if the parts cost is more than $500 per speaker. Some said $1000. My priority is to make this a very good speaker, and suitable parts may cost a bit of money. In the end, there will always be substantial flexibility with costs related to x-over parts and enclosure materials + finishing.

NEXT STEPS AND QUESTIONS (everyone has his/her own approach to loudspeaker design, but here's my suggestion):
  • In the first step we'll have to narrow down the driver choices. This will be a forward-and-backward process between driver suggestions and model runs using these drivers, always keeping the design targets in mind. I'd suggest choosing a woofer first, because that will define the box size, bass extension, and sensitivity. Let's say the midrange is set (yes, you are allowed to complain, but... :mallet:). The tweeter comes last.
    Question: What input data do we need for these model runs? Do we need actual measurements of the drivers in their baffle/box? I am not big with modelling tools, and I tend to replace detailed model runs by measurements from prototypes. Others might be more efficient than this by using their modelling tools.
  • The second step is to build an actual prototype. Now we need money to buy parts. If the design is still attractive to me (i.e., it's something that I'd like to build and use myself), I'd be willing to put in some money and go for it. I could also do measurements.
    Question: How will the funding work? Anyone experienced with raising and managing money for open source projects? Anyone willing to put some cash on the table? Who will manage the money, and how?
  • The third step is to tweak the prototype. Again, this will be another forward-and-backward process. And we may need more money.


---------------------------------------------

Notes and updates:


Important note 31 Oct. 2019: DIYers may profit from this open-source project for their own private purposes, for example by building and enjoying a copy of the Open Source Monkey Coffin speaker. Please do not use the information developed in this open source project on a larger scale (for example by selling speakers based on the Open Source Monkey Coffin design or substantial parts of it) without written permission.

Note 4 Oct. 2018: This project has received financial support from LORDSANSUI, Paul Vancluysen, George Wright, KaffiMann, Charles Bueche, zimmer64, John Barbor, and others (anonymous) here. Thank you!

Note 10 Nov. 2019: This project has come a very long way, and the OSMC design is now completed. The design goals have been largely met, but some of the early implementation ideas have changed a bit along the way (x-over topology, woofer size, tweeter details). The full documentation of the OSMC design is here: https://github.com/mbrennwa/osmcdoc

Note 4 Oct. 2018 and 18 Nov. 2019: Final choice of drivers:
  • Tweeter: ScanSpeak R2904/7000 ring radiator (post 295 and follow ups) with a custom waveguide as described in post 881 (available via group buy or by DIY 3D printing or CNC machining).
  • Midrange: Volt VM752 3" textile dome
  • Woofer: FaitalPRO 12PR320 12" paper cone (see post 162 and post 244).

Note 18 Nov. 2019: Final enclosure is in post 925 (drawings) and post 930 (photos, showing stuffing).

Note 18 Nov. 2019: Final x-over is in post 931.

Note 18 Nov. 2019: The polar response diagrams (horizontal and vertical) are in post 895.

Note 18 Nov. 2019: The power response is in post 901.

Note 18 Nov. 2019: The electrical impedance is in post 900.

Note 19 Feb. 2020: A bunch of graphs with test results as compiled from the documentation on GitHub (on-axis SPL response, step response, polar response, cumulative decay spectrum (CSD) are in post 917.

Note 9 May 2020: If you don't know where to look, here's a list with some shops and retailers for OSMC parts:
  • diyAudio group buy for the OSMC tweeter waveguides
  • Falcon (Europe): Scan 2904 tweeter, Volt VM752 midrange driver, x-over parts, etc.
  • Solen (Canada, US): Scan 2904 tweeter, Volt VM752 midrange driver, x-over parts, etc.
  • Audiohobby (free shipping everywhere!?): Scan 2904 tweeter, x-over parts, bass-reflex ports
  • US Speaker (US): Faital 12PR320 woofer
  • Blue Aran (Europe): Faital 12PR320 woofer
  • Acoustical Surfaces (US): Melamine foam. Contact person: Tom Ewert +1 (952) 466-8229. They also sell in small quantities!
  • Ebay: Melamine foam
  • Buttinette (yes...): lambs wool
  • Deer Creek Audio: Custom speaker stands
 
Last edited:
I've started checking out some drivers, there is a slight chance we can get away with a really nice 12".
Will look into it tomorrow.

yes, someone mentioned Fane SOVEREIGN series woofer, looks good if I remember correctly, I would really like to see a very good low crossed mid range/mid woofer, like maybe Acoustic Elegance TD6M (high end, reasonably priced and high sensitivity) ? I personally prefer a good mid woofer to carry the sound (maybe over 200Hz) I don't like my woofer to have anything to do with my busy bottom heavy music and vice versa, to my ears it makes it a little boomy.
 
Last edited:
Fanatic
Joined 2009
Paid Member
What I think Mordikai is saying:
The Sovereign series lack some fancy bits like cast frame and shorting rings etc. The brand or manufacturer does not matter so much, it's more the build quality.
(I like Fane, it's good stuff, and value for money!)

I am still checking around, nothing nailed about the driver, design or anything, but I got the compact volume part at least...
It may be hard making a 30 x 65 x 30 box, mbrennwa.

... xo at 300hz?
 

Attachments

  • Faital Pro 12FH530_preliminary.jpg
    Faital Pro 12FH530_preliminary.jpg
    376.2 KB · Views: 9,262
Member
Joined 2005
Paid Member
What I think Mordikai is saying:
The Sovereign series lack some fancy bits like cast frame and shorting rings etc. The brand or manufacturer does not matter so much, it's more the build quality.
(I like Fane, it's good stuff, and value for money!)

Ok, I see. Is there a way to tell if a driver has copper rings etc? Some data sheets are not very specific about this? Is slow impedance rise at frequencies above the responance peak a reliable indicator?

I am still checking around, nothing nailed about the driver, design or anything, but I got the compact volume part at least...

I tried to reproduce your result for the Faital 12FH530 in a bass reflex box. I could not reproduce your result using AJ Horn. No matter what I tried, a 59 L box would give a nasty response peak at the low end. I can get it flat with a volume of 30 L or so, with a -3dB frequency of about 52 Hz. Quite a difference to your result. Hmm.

What about the Beyma SM-110/N or SM-112/N? I get decent results with AJ Horn from these.

It may be hard making a 30 x 65 x 30 box, mbrennwa.

Sure. I am flexible with this.

... xo at 300hz?

Not with the Volt dome midrange. Need to go up to about 400 Hz I guess.
 
It is not so easy to use bass reflex as "Type of design", I am not familiar with AJ horn, but hornresp and MJK's excel sheets are good, use horn simulation software, do not use bass reflex software. It is a slot port design.
If you want I can try and get an Akabak simulation going to verify, but I got so consistent results from hornresp that I stopped verifying, it may take some time to do this, have not used it for some years.

And I like to have the combined phase from port and driver together, it is hard to get good (PROPER!) results with bass reflex software.

Edit:
Do you mind posting the data you entered into AJ horn? Maybe the dimensions are different?
 
Last edited:
OH!
I am sorry, there is a peak at the low end, my sim was wrong.
I had somehow written 6,4ohm dc resistance instead of 5,1ohm.

I will revise the simulation.

Edit:
At any rate, it was just intended as an indication that the goal is possible, I do not have any intention of using the Faital Pro 12FH530, it was just a random driver that seemed to give "ok" results. I would rather pick a driver with different values.
 
Last edited:
It is not so easy to use bass reflex as "Type of design", I am not familiar with AJ horn, but hornresp and MJK's excel sheets are good, use horn simulation software, do not use bass reflex software. It is a slot port design.
If you want I can try and get an Akabak simulation going to verify, but I got so consistent results from hornresp that I stopped verifying, it may take some time to do this, have not used it for some years.

And I like to have the combined phase from port and driver together, it is hard to get good (PROPER!) results with bass reflex software.

Edit:
Do you mind posting the data you entered into AJ horn? Maybe the dimensions are different?

AJ Horn does the "generalised transmission line approach". It links the different box units by their acoustic impedances to model the system. It is not the "classical bass reflex equation" program. I attached my inputs and the outputs (SPL at 1m, sum of driver + port, free space).

The box input screen may look a bit funky when your brain is wired to bass reflex. It works by linking the driver to a box with a volume of 59 L, and this box is then linked to a tube (44cm² cross section, 10cm long). The tube looks like a "horn" on the input screen, but it really is the bass reflex port.
 

Attachments

  • monkeybox_faital12FH530_driver.png
    monkeybox_faital12FH530_driver.png
    13 KB · Views: 1,858
  • monkeybox_faital12FH530_box.png
    monkeybox_faital12FH530_box.png
    10.7 KB · Views: 1,850
  • monkeybox_faital12FH530_curves.png
    monkeybox_faital12FH530_curves.png
    15.7 KB · Views: 1,864
OH!
I am sorry, there is a peak at the low end, my sim was wrong.
I had somehow written 6,4ohm dc resistance instead of 5,1ohm.

I will revise the simulation.

I just tried AJ Horn with 5.1 Ohm (correct, black curve) and 6.4 Ohm (wrong value, red curve). The difference is not very large. I believe the difference between our simulations is somewhere else.

If you could provide your 5.1 Ohm simulation I will ask the developer of AJ Horn about his opinion on this.
 

Attachments

  • monkeybox_faital12FH530_curves_RDC_5.1_6.4.png
    monkeybox_faital12FH530_curves_RDC_5.1_6.4.png
    17.6 KB · Views: 1,811
mbrennwa:
The port I used for simulation is over 36cm long, not 10cm. And it's 33,1cm wide (divider in the middle for bracing) * 4cm tall, so 132,4cm2. It is not only tuning frequency that is important, but how the port behaves at tuning frequency.

Anyway. Using a Beyma SM212 the results are much better, group delay can be lowered to a very nice value. 400hz and maybe even up to 700hz should be just fine, if you can place some damping material on the right locations.
 

Attachments

  • Beyma SM212.jpg
    Beyma SM212.jpg
    378.5 KB · Views: 2,606
IMO it is better to use a closed sub @home as the room gain will amplify the lowest freqencies the most. BR, horns, TL and bandpass boxes have a steeper rolloff at the bottom, which makes them prone to booming.
Put the closed sub in a corner and if it is boomy, move it along a wall untill the sound is ok.

In general they are way more monkey proof. :)
 
Last edited:
Think:
It's not a sub, but a 3 way speaker suited for stands or shelves, IE not floorstanding.
Previously I had the same misconception you seem to have, that ported enclosures are inherently boomy because they function on the principle of resonance in relation to volume. This is not right, every single type of design can be done "wrong", it does not have anything to do with the type of design, but the implementation.

I set out to "prove" that you can not get a ported enclosure to sound good. I proved myself wrong and am now a convert.
 
mbrennwa:
The port I used for simulation is over 36cm long, not 10cm. And it's 33,1cm wide (divider in the middle for bracing) * 4cm tall, so 132,4cm2. It is not only tuning frequency that is important, but how the port behaves at tuning frequency.

Oh man... thanks for pointing out this stupid mistake. I guess there was a reason for my earlier comment that I am not very good with those simulation tools. I will try again this evening.
 
Hahah! No worries, I had a *facepalm* moment myself when I noticed I had put the wrong parameter in hornresp. :-D

Can we call it even?

Regarding port loading:
Yes, there's more to it. Important to keep group delay as low as possible (completely ignored in commercial designs), often a port will cause lots of phase wrap at resonance, have to be mindful of that in the design.
Also think about the distances between the driver membrane and the internal sides of the enclosure and what frequency modes can be excited, it is easier to use dampening material to reduce high frequencies than low frequencies, just like with filters.
 
Last edited: