Hi I started working on a simplified body for the Mark II that can be easier on an FDM 3D printer as I had some very frustrating hours trying to print the original STL files.
@guhl I don’t like the BoM because for some reason its so dictated to specifics.
I think its quite possible to create an extensible design that allows an array of equipment and budget all from the same base.
That you can start simple by firstly getting just a simple enclosure and build up in steps or if you want you can purchase complete all bells and whilstles, ready made.
My first issue with I posted on Github wasn’t very focussed but took some time to make a clearer proposal.
Opensource design isn’t easy because the materials and purchase qty you have often means everything is a compromise.
I am bemused at very specific solutions that seem at odds with a opensource collaborative project that is extremely modular with a plethora of options as that would seem to be a polar oppiste of what the project is.
It also narrows interest rather than increasing it, as either you want that very specific BoM or its of no interest at all.
I think its possible to create a 4.5" puck that is extensible, modular with a range of function and much choice of budget.
That apart from some of the enclsoure parts there isn’t really a BoM as its about choice.
That you have a choice from a $7 PS3 eye usb cam, $10 ReSpeaker 2-Mics Pi HAT, $25 ReSpeaker 4-Mic Array for Raspberry Pi or $99.00 ReSpeaker Core v2.0 with the last one also being your SBC.
Which even though they say the are Pi hats actually its any SBC that is the original ‘credit card’ profile or smaller.
Pi/ReSpeaker Core v2.0 is prob best option but some may already have a tinker or rock64 or even a Pi zero.
That this can start as a speakerless Bluetooth AI, be just a Bluetooth speaker or be AI, Pan/Tilt, Cam Display, Speaker Mic deluxe, personal assistant robot.
All from the same collection of extensible and modular community designs.
That as a community people can say hey anyone make ‘spi cables’, ‘cuts vinyls’ to ‘print backplates’ or even does anyone provide finnished loaded and ready to go ‘recipe’.
So I am hoping I can twist your arm and wet your appetite to start the ball rolling with an AIX format specification that like ATX did for PC we have a basic format that can provide a plethora of options.
Also that as a community we have a look at some common engineering materials on maybe how we can molest them for others uses.
Well this is a very far reaching topic that you are laying out here.
First I will definitely give this a thorough thought an see if I can and want to work on a modular chasing platform if I have time (and thanks to a certain virus this might well be the case).
I really like the idea of modular designs and a variety of options but I also would like to tell you that I am not so convinced any more that these are the keys to success.
There is always the risk of ending up in a GNU HURD situation if modular design is taken to far.
In the last 20 years I have been involved in too many community and startup projects that spread out too far until there was no focus and no coherent strategy left. In German we would call this trying to create the “Eier legende Woll-Milch-Sau”.
And regarding a restrictive BOM I would like to add something.
When the 3D-printing thing started and I got involved in the RepRap project there where plenty of options and modular designs and it was all just done by hard core tinkerers and DIY specialists. But the first really successful model was the availability of a complete rather restrictive BOM and in addition some, then small, companies that picked up the idea and basically sold the complete BOM as a DIY kit.
Not everybody wants and has the ability to source the different parts separately or even think about individual setups.
And as the Mark II in principle is a commercial product (at least in the end I really hope that it still will be one) this community driven version could also be something like the RepRap Mendel Prusa.
Don’t take this as a turn down of your strive for a highly modular and flexible platform, I just wanted to point out a different view.
What I think would be necessary to make a Mark II community version successful would be the availability of a DIY kit that you can purchase from one source. Maybe even including the a 3D-printed case (probably not my version) so that people interested in building and tinkering with a Voice Control Device can have an easy starting point.
Works OK, up to now.
I messed up the size of the carrier of the power supply, it is to small but works for me. Fixed that in CAD and will update the git soon.
But otherwise the bottom and the audio chamber look good.
I think you are right Gruhl “successful would be the availability of a DIY kit”.
But part of the modular thing is to be that carrot to get you started and once hooked you end up buying the lot anyway
To be honest its not that modular as Seed have by far the best array of mic kits for price.
There are a few options with SBC but max Pi3 format size and min perf level gives a few choices.
Its more you just don’t have to buy them all in one go or at all modular, rather than trying to adopt all.
I understand what you mean about have your eyes set so far you lose focus.
There is a huge difference with the specific job a 3d printer does than some ad hoc components for a great array modules/software provided by Mycroft.ai and community.
Mycroft doesn’t do a specific job like a 3d printer does.
That is why it needs to be far less specific and more general but this is at such a low level.
There isn’t even something as simple as a decent Pi enclosure that can house a speaker and mic array.
Even Respeaker Core2 don’t have an enclosure and the core 1 doesn’t fit a speaker.
I really agree with the idea of a DiY kit and and slightly bemused when the absolute basic blocks are absent and bemused that Mycroft are not rubbing there hands at the possibility of some simple solutions and the traffic that will garner as elsewhere it doesn’t exist, so it would come.
So much so that I am one of those without a 3d printer and poor old Myford is going to reside in a 4" shit pipe as its called here in the UK.
So yeah I am the same and have a strong wish they would do a DIY kit, but if you think shit pipe will provide vastly better audio quality and generally overall solution, like I do.
Your not going to buy but just wish someone somewhere would create a kit and set a format in a similar way they have a framework rather than just a binary blob.
In my S-pipe version I wish to adapt what the Google engineers did with a cone facing a single speaker, its a slight curved cone also but on the right I just drew that awfull triangle to represent the side view of a cone .
My drawing skills are so bad that the pointless effort of centering the left V shape means its too low, the bottom of the V should be the center of a circle.
You will have to forgive the drawing as said and forgot the solidworks term which is oppisite to extrude but if you started with that standard cone and shaped it into that V that is something like what I am expecting.
The badly worded oppiste extrude creates a chamfer so its a wider and more directional, also harder to explain but basically that and blended to create a final form.
The Myford puck I am playing with uses schedule 40 4" pipe with a 3.3" full range driver.
So the imaginery speaker cutout is 75mm in an inside diameter Sched 40 4" tube of 102.2604 mm.
The cutout is hidden which is good for the likes of me as we can be quite brutal with a jigsaw or dremmel.
The actual shape is far from formalised but its an impossible shape and profile to describe or at least it was for me.
It just makes more sense to describe how I am going to assemble.
On top of the tube I have a 6mm thick circular acylic disk as I can get those cut pretty cheap.
Its 114.3mm diameter and sits on 4 hex spacers that go to another acrylic disk that is the internal diameter of 102.2604 mm.
The spacers raise the top disk so there is 10mm gap from tube to the lid forming top disc that does a similar thing as the google design but rather than omnidirectional it creates an unidirectional sound pattern.
Its actually a really complex shape that 3D tools would have no problem making.
As well as creating the dispersion pattern of sound it acts as an acoutic cover to the microphone array.
Also the V shape is to house the possibility of a tilt mechanism and provide a path for a Pi Cam style ribbon cable to pass down to the lower non acoustic compartment of housed electronics.
@guhl if you ever have any spare time then please give it some thought as you seem to be really talented that way.
I am a terrible hack with code and also the same when it comes design assembly but part from that one 3D printed part I can accomplish much with just a drill and some simple tools.
The top disk as said fits on of 4 spacers that on the underside hold mic array and this wierd and wonderful sound dispersion cover.
They go to another acrylic disk that I forgot about which is the speaker cut-out (75mm) but underneath the spacer pillars continue to the bottom acrylic disc that continue again to a FRB board that makes the base of the SBC/Amp/Buck housing.
Those 2 disk should really be sealed with resin, hot glue maybe silcone or something but they make quite an adequate acoustic cabinet once sealed.
If you are going to go all audiophile you might want to line with some rubber sheeting.
What I am proposing is super low tech that I am wondering how you can produce such a great and complex design and then have an opinion and belief people might get lost in modularity complexity.
The modularity is so simple that the choice and position of the drill holes allows you to mount and array of Mic boards, SBC and amps that can be printed out on a paper guide.
The only thing that is complex is the shape form of “dispersion cone” and is perfect for 3D printing as apart from bespoke molded parts there isn’t really a way to do it with conventual engineering or materials.
Like I say if you ever do have any spare time or anyone else in the community who is a solidworks or whatever and 3d printing guy then please do.
I have ordered a few bits of plastic just to build one up minus those essential bits of finnese.
I will post some pics when its assembled as you never know it might perk some interest.
So the S-pipe is on its way.
Yeah that brill @Dominik to be honest what I am doing is just an example and even hobnailed myself by picking US 4" schedule pipe puely for example, as with the respeaker core2 it might be a tight squeeze on what I am doing.
110mm standard EU tube might of been a better option.
But that is exactly the sort of thing where you can make modular systems and its not much more than hex pillars in some sort of enclosure.
Its what @guhl said about an offcial kit and what I m trying to demonstrate that if one was available all that is needed is a couple or parts from simple precut tube, some discs and you have that modularity.
The respeaker core2 is pretty cool as its would allow the lowest profile design, but there is definately a common set of a few parts that would allow some simple modular choices.
I am trying to work out what they are doing in that PDF is it something similar to a Pibow case?
But yeah with a common tube size you can make an array of modular pucks.
Someone could make decent $ as the engineering materials are really cheap so they could be offered at quite a decent price and still make good $
With a vinyl kit to finish.
The main drawback of the respeaker core2 is the closed source Alango routines that for opensource is slightly taboo but as a bit of kit its a great bit of kit.
The onboard amp can only drive 8 watt, but it would also fit the exact same orangement of the Pi version but just have a but more space.
You could add an amp then, but its just a stack on some common kit materials which currently are lacking for all.
S-pipe is extremely cheap for a very dense and ridgid external tube. https://www.plasticdrainage.co.uk/plain-ended-pipe-3-mtr-4-8297.html
3 meters costs £7.50 in the UK and could be cut into 18 - 20 sections that cost price is £0.40
The material costs of cut discs if sourced is equally cheap.
You could easily create a £10 - 20 kit with quite a whopping markup that incudes some printed parts and vinyls.
Or as seperate items.
I had some delay because my 3D-printer broke down (it is a very old RepRap Mendel Prusa) and I had to rebuild some parts of it. (but I managed to print the necessary spare parts before it broke totally).
Unfortunately the threaded inserts were not delivered yet (and who knows when they will be) so I can not finally assemble the body.
But so far everything looks fine.
Hi,
from my perspective this is complete. I did some fixes to the CAD and added the sides.
All STLs are complete and the repository is up to date.
Have fun - Guhl
as they are not so important).
