-
SCAM WARNING! See how this scam works in Classifieds.
The Bud Toaster - (currently: Model 14, version 3)
- Thread starter Hippie Dickie
- Start date
nothing new to share yet. i plan to take all my work into some maker space. i think i have some useful information to share.
couple of days ago i was vaping on the balcony and thought of a new way to fabricate the cube body. i realized that my cnc router makes it possible to revisit the lessons learned on the cylinder model, with a channel milled for a metal midbody. it eliminates some boring steps and buying new woodworking equipment ... or working with a woodworker. and needs fewer milling steps. 2 or 3 times faster to fabricate.
so, i have to do some drawings, generate some GCODE and make some samples.
i have been consumed by studying biology for the last 5 years - as it turns out, a life or death pursuit. and i MUST wrap up some computer projects.
current cube is still going strong. use it 5 to 6 times per day, every day. but this new idea for the cube is burning a hole in my mind. soon, he says, soon.
couple of days ago i was vaping on the balcony and thought of a new way to fabricate the cube body. i realized that my cnc router makes it possible to revisit the lessons learned on the cylinder model, with a channel milled for a metal midbody. it eliminates some boring steps and buying new woodworking equipment ... or working with a woodworker. and needs fewer milling steps. 2 or 3 times faster to fabricate.
so, i have to do some drawings, generate some GCODE and make some samples.
i have been consumed by studying biology for the last 5 years - as it turns out, a life or death pursuit. and i MUST wrap up some computer projects.
current cube is still going strong. use it 5 to 6 times per day, every day. but this new idea for the cube is burning a hole in my mind. soon, he says, soon.
D
Deleted member 30392
Guest
I have been following this for nearly a decade... 
thank-you for your interest. two decades for me.
update time ...
my cube was starting to act wonky (technical term - meaning some type of operator assistance (i.e. whack on side with knuckle) was needed for functional operation). so for the first time in several years i took it apart and found the copper connector holding one end of the heater coil had unsoldered from the pcb - must have been a cold-ish solder joint when fabricated - and a thousand heat/cold cycles let it loose. resoldered, reassembled and now running as expected.
i was happy to see there was no infiltration of cannabis, or even dust, inside the cube. ordered and received some 1.7mm router bits for the CNC router for the next iteration of the cube design. this is very exciting - eliminates some drilling steps, wood working steps, detailing steps - could be a game changer.
ordered some 6mm x 6.5" x 15" cherry sheets to try the new cube design. getting back into doing design updates, writing a lot of documentation, and some prototyping.
sometimes X is useful. the AI showed me a new knot: the constrictor. i have been waiting for the perfect knot to appear. it is necessary to bind the two wires of the power cable. zip ties are too cumbersome - bad esthetic. i use WS Deans 16ga ultrawire - 720 strands and wonderful silicon insulation so it is super flexible. but this makes tying the two wires together, well, impossible until the X post ... appears to be problem solved.
how the AI knew to show me this knot, now, i don't know. it's a tease - could have shown me this years ago.
details to follow.
my cube was starting to act wonky (technical term - meaning some type of operator assistance (i.e. whack on side with knuckle) was needed for functional operation). so for the first time in several years i took it apart and found the copper connector holding one end of the heater coil had unsoldered from the pcb - must have been a cold-ish solder joint when fabricated - and a thousand heat/cold cycles let it loose. resoldered, reassembled and now running as expected.
i was happy to see there was no infiltration of cannabis, or even dust, inside the cube. ordered and received some 1.7mm router bits for the CNC router for the next iteration of the cube design. this is very exciting - eliminates some drilling steps, wood working steps, detailing steps - could be a game changer.
ordered some 6mm x 6.5" x 15" cherry sheets to try the new cube design. getting back into doing design updates, writing a lot of documentation, and some prototyping.
sometimes X is useful. the AI showed me a new knot: the constrictor. i have been waiting for the perfect knot to appear. it is necessary to bind the two wires of the power cable. zip ties are too cumbersome - bad esthetic. i use WS Deans 16ga ultrawire - 720 strands and wonderful silicon insulation so it is super flexible. but this makes tying the two wires together, well, impossible until the X post ... appears to be problem solved.
how the AI knew to show me this knot, now, i don't know. it's a tease - could have shown me this years ago.
details to follow.
wonky again ... so i did a deeper dive into the internals to really locate the problem ... sometimes, when looking at this 1.75" pcb though the magnifying light on my desk i swear it seems so much bigger - like the Tardis ... anyway, upon removing the nichrome80 ribbon coil one of the copper mounting clips fell right out of the pcb. and was very oxidized - as opposed to the nichrome80 ribbon which is pristine - so filed all the oxide off, resoldered to the pcb and now the heat up is much, much better going 72°F to 400°F in 10 seconds. it was only hitting 30°F per second with that cold solder joint.
and this cube has been used for 1000s of sessions and the battery pack has been recharged probably 3000 times. LiFePO4 for the win! (Tesla calls it LFP).
and this cube has been used for 1000s of sessions and the battery pack has been recharged probably 3000 times. LiFePO4 for the win! (Tesla calls it LFP).
now, some interesting new performance results ...
i am thinking the cold solder joint has been there for a long, long time. now, i am getting about 20°F overshoot on warm up. and the temp climbs about 2 degrees with a strong inhale. haven't seen this since earlier units. looks like the PID parameters need some tuning. these are stored in a table so trivial to adjust.
i am hoping i can get AI to help with this. in my simple minded view, showing the AI a video of the temperature change with time during a session and giving the table values should figure it out. i wish. copilot in Visual Studio Code is only a little helpful. they sell it as providing so much more - but that's a bit off topic.
generally, the parameters are too large. trial and error adjustments works, too.
i am thinking the cold solder joint has been there for a long, long time. now, i am getting about 20°F overshoot on warm up. and the temp climbs about 2 degrees with a strong inhale. haven't seen this since earlier units. looks like the PID parameters need some tuning. these are stored in a table so trivial to adjust.
i am hoping i can get AI to help with this. in my simple minded view, showing the AI a video of the temperature change with time during a session and giving the table values should figure it out. i wish. copilot in Visual Studio Code is only a little helpful. they sell it as providing so much more - but that's a bit off topic.
generally, the parameters are too large. trial and error adjustments works, too.
2tiki
Well-Known Member
You may want to try the currently free google AI for that. You can show it things on your screen and then ask it questions, like you described.
XpeeN
Well-Known Member
Or even better, use Duckduckgo AIs (privacy respecting wrappers)
2tiki
Well-Known Member
Can you show your screen to that AI, and get it to help program or understand what you are working on just by looking at it? The new google one is highly advanced, and can do that, which is why I suggested it. That seemed to be the function they are looking for.
I totally understand the privacy issue. And mostly use open source AI on my computer, but that google one can do things I haven't' seen other options doing yet.
Anyway, I don't want this to turn into an AI thread. Just trying to be helpful.
4/20 update ... (was going to post this yesterday but not quite done)
finally cleared my desk enough to once again get some solid time put into this project. the focus has been on the GCODE for the CNC router.
the design has evolved from a solid 2" cherry cube, to milling the faces from a 1/4" thick x 6-1/2" x 15" cherry blank - normally used for laser engraving, but the 6mm thickness is perfect. the six "faces" of the cube are laid out on the cherry blank. both sides of the blank are milled - each face has an inside and outside. the faces are aligned on the blank. the inside is done first, the blank is flipped on X axis, the outside is milled, the tool is changed from 1/16" upcut to 1/8" upcut and the 6 faces are cut from the blank.
the vertical faces - left, right, front, back - have tabs along the top and bottom edges that fit into slots cut into the inside of the top and bottom faces. the PCB slides into grooves in the vertical faces and is screwed down to the bottom face, locking the vertical faces into the bottom face. the top face slots fit to the vertical face tabs. screws lock the top to the vertical faces. that's the plan, anyway.
here is the layout for the inside of the faces:
flip that panel across the X axis and get this layout:
after this is milled, change the tool and cutout the faces:
the milling for each of the faces is defined in an Excel worksheet. this one is for the
TOP inside face in the top image:
there are two buttons. the second one, Create gcode Sheet, runs down the rows from (START) to (END) and creates a worksheet
with the gcode instructions. here is the worksheet for TOP inside:
the first 3 TAP columns are truncated, but similar to the last TAP column which shows the full gcode.
there is an Excel module that collects the gcode for the 6 inside face gcode worksheets and outputs a single .ncc file the feeds the CNC router. same for the 6 outside face gcode worksheets.
the simulator says the milling will take 34 minutes total time.
maybe, maybe, maybe i can actually try this tomorrow. and pictures! maybe
all this information is going on a website, TheHerbalCube.com, but that hasn't happened yet. i am still organizing the source docs and looking to see which AI can generate the site.
more to follow if anybody is interested.
finally cleared my desk enough to once again get some solid time put into this project. the focus has been on the GCODE for the CNC router.
the design has evolved from a solid 2" cherry cube, to milling the faces from a 1/4" thick x 6-1/2" x 15" cherry blank - normally used for laser engraving, but the 6mm thickness is perfect. the six "faces" of the cube are laid out on the cherry blank. both sides of the blank are milled - each face has an inside and outside. the faces are aligned on the blank. the inside is done first, the blank is flipped on X axis, the outside is milled, the tool is changed from 1/16" upcut to 1/8" upcut and the 6 faces are cut from the blank.
the vertical faces - left, right, front, back - have tabs along the top and bottom edges that fit into slots cut into the inside of the top and bottom faces. the PCB slides into grooves in the vertical faces and is screwed down to the bottom face, locking the vertical faces into the bottom face. the top face slots fit to the vertical face tabs. screws lock the top to the vertical faces. that's the plan, anyway.
here is the layout for the inside of the faces:
flip that panel across the X axis and get this layout:
after this is milled, change the tool and cutout the faces:
the milling for each of the faces is defined in an Excel worksheet. this one is for the
TOP inside face in the top image:
there are two buttons. the second one, Create gcode Sheet, runs down the rows from (START) to (END) and creates a worksheet
with the gcode instructions. here is the worksheet for TOP inside:
the first 3 TAP columns are truncated, but similar to the last TAP column which shows the full gcode.
there is an Excel module that collects the gcode for the 6 inside face gcode worksheets and outputs a single .ncc file the feeds the CNC router. same for the 6 outside face gcode worksheets.
the simulator says the milling will take 34 minutes total time.
maybe, maybe, maybe i can actually try this tomorrow. and pictures! maybe
all this information is going on a website, TheHerbalCube.com, but that hasn't happened yet. i am still organizing the source docs and looking to see which AI can generate the site.
more to follow if anybody is interested.