interesting indeed. I wonder if theres any way to just blow dry ice powder into blend space sometimes to keep temps low? or in least just pipe outgas from it
Of course, cooling to temperatures when PLA gets really brittle is an option, but you must; (1) Source liquid nitrogen or solid CO2, which isn't available in any hardware store. (2) add the energy needed to get nitrogen from room temperature to liquid state or CO2 to solid to the energy bill. Piping out gas or add an intercooler is what I mentioned at the end of the video, didn't I?
did you think about a setup more like flour grinding with rotating rock. the rotation wheel should have a taper and a little oscillation. I think it could be more efficient then swinging plastic balls
There's another process used to make fine powders industrially-- melt blowing... something like making 'cotton candy' Might work for low melt plastics like PLA?
Surely no easy task. You will very likely end with some sort of "cotton PLA" rather than with a powder. Melt blowing (atomizing) works fine with metals, but long plastic molecules have different properties.
Hello, I make this comment regarding the software to control the CNC with DC motors and optical encoders, which other program could be used? I only have Windows
They become a bit more brittle. If your freezer is really cold, the PLA becomes really brittle. But: (1) you have to add that freezing to the energy bill (2) really low temperatures are not that common in a household and I'd like to keep things in range of an average tinkerer.
@@HomoFaciens freezer is always on whether i put plastics or not. if -10c not enough you are right... i like your videos and thinking process. thanks for your videos
But the freezer draws extra power to lower the temperature of the PLA from room temperature to let's say -10°C. That extra energy is what you wave to add to the bill.
I asked Gemini (google AI) and his answer was low speed sharp blades are more efficient than high speed dull blades. Also he mentioned that the most efficient will be to use the large 3D scraps directly in to the 3D printer without any grinding. His claim of energy efficiency was 300W for 5 minutes to get 1kg of of shredded PLA so 25Wh/kg using low speed sharp blades grinder.
...and what CAD files did that super intelligent Gemini deliver? AI is a search engine without mentioning sources and when I asked for a source for the numbers I did not mean a piece of software ;-)
@@HomoFaciens :) Even current LLM based AI's are more than a search engine. They still fail at logical understanding and they are polluted with a lot of incorrect training data. But if used properly it can provide helpful answers like in this case. The AI actualy calculated wrong and first replay was 300W for 5 minutes is 0.25kWh but then I asked for clarification and was able to recognize the mistake and answer correctly as 25Wh not 250Wh. But the low speed sharp blades makes perfect sense. Think about tiring to split a rubber ball in to 10 smaller pieces. You can use a large hammer and try to do that using quite a lot of energy or you can use a very sharp blade to cut the rubber ball requiring way less energy. For a glass ball that will be the opposite and PLA is closer to rubber than glass or stone.
But does the artificial "intelligence" say how a machine must look like? That's why I was asking for CAD files. "Training data" means searching the web and calculating some sort of average answer. Still a search engine to me ;-) Without any link to the "training data" there is: You can believe it, but you don't have to. Where did Gemini get the specific energy for PLA from? That cutting is in general better than random milling is a statement I made in the video, didn't I?
@@HomoFaciens In order to calculate some sort of average answer from a lot of data you need to be able to understand what the data means. They only predict what the next most likely word is but we may be no different from that. The energy required to grind the PLA could be wrong but I think the strategy to use low energy is slow cutting instead of smashing. I could ask Gemini if you made that statement in the video and it will be able to answer saving me about 14 minutes :) I can not currently ask for CAD files but it will not be more than one or two years before this will be a possibility. I asked about the energy needed to melt 1kg of PLA and the answer was Q = 1000 g * 0.93 J/g°C * (175°C - 20°C) = 144750 J but LLM's are still very bad at math as it is 144150J Searching the specific heat for PLA will be a nightmare using google and you get all sort of wild results including the first one that says "1200 J/Ng/K" what is with that Ng ? could be ng but then the unit will be incorrect. Looking in the fist PDF link there are two values 1200 and 1800J/kg/K so the first value is not far off 930 used by Gemini vs 1200 The melting temperature is also fairly accurate at 175°C In any case 144150J is about 40Wh thus 25Wh to grind 1kg of PLA is not unrealistic.
As told before, the core problem of Gemini is: You can believe it or not. There is no way to see on what data the software generated the answer. Gemini delivers a result in seconds, but to prove if that result is correct, you still have to spend hours of research work ;-) There is no "Intelligence" behind the code lines, just parameters, self adjusted by the same piece of software. Shit in, shit out is what all database software suffers from...
Great videos as always!
Very good review!
interesting indeed. I wonder if theres any way to just blow dry ice powder into blend space sometimes to keep temps low? or in least just pipe outgas from it
Of course, cooling to temperatures when PLA gets really brittle is an option, but you must;
(1) Source liquid nitrogen or solid CO2, which isn't available in any hardware store.
(2) add the energy needed to get nitrogen from room temperature to liquid state or CO2 to solid to the energy bill.
Piping out gas or add an intercooler is what I mentioned at the end of the video, didn't I?
@@HomoFaciens Indeed. I wonder if AC output at max cool would work however lol and that is more buyable
did you think about a setup more like flour grinding with rotating rock. the rotation wheel should have a taper and a little oscillation. I think it could be more efficient then swinging plastic balls
You don't want to have rock fragments in your plastic powder. It will end up in a clogged nozzle in the extruder.
@@HomoFaciens good point. the same concept could be transferred to a metal grinding with different taper angle for inner wheel and outside wall though
There's another process used to make fine powders industrially-- melt blowing... something like making 'cotton candy' Might work for low melt plastics like PLA?
Surely no easy task. You will very likely end with some sort of "cotton PLA" rather than with a powder. Melt blowing (atomizing) works fine with metals, but long plastic molecules have different properties.
Hello, I make this comment regarding the software to control the CNC with DC motors and optical encoders, which other program could be used? I only have Windows
...but this video definitely isn't about a CNC machine!?!?
@@HomoFaciens Excuse me, what happens is that I wrote to you in the homemade CNC videos but you never responded to me
What about having a look at my pages? There you can even find a column named "contact".
@@HomoFaciens thanks👍🙂
@@HomoFaciens I already wrote to the email
what if we put failed plastics in the freezer?
They become a bit more brittle. If your freezer is really cold, the PLA becomes really brittle. But:
(1) you have to add that freezing to the energy bill
(2) really low temperatures are not that common in a household and I'd like to keep things in range of an average tinkerer.
@@HomoFaciens freezer is always on whether i put plastics or not. if -10c not enough you are right... i like your videos and thinking process. thanks for your videos
But the freezer draws extra power to lower the temperature of the PLA from room temperature to let's say -10°C. That extra energy is what you wave to add to the bill.
I asked Gemini (google AI) and his answer was low speed sharp blades are more efficient than high speed dull blades. Also he mentioned that the most efficient will be to use the large 3D scraps directly in to the 3D printer without any grinding.
His claim of energy efficiency was 300W for 5 minutes to get 1kg of of shredded PLA so 25Wh/kg using low speed sharp blades grinder.
...and what CAD files did that super intelligent Gemini deliver?
AI is a search engine without mentioning sources and when I asked for a source for the numbers I did not mean a piece of software ;-)
@@HomoFaciens :) Even current LLM based AI's are more than a search engine. They still fail at logical understanding and they are polluted with a lot of incorrect training data.
But if used properly it can provide helpful answers like in this case.
The AI actualy calculated wrong and first replay was 300W for 5 minutes is 0.25kWh but then I asked for clarification and was able to recognize the mistake and answer correctly as 25Wh not 250Wh.
But the low speed sharp blades makes perfect sense. Think about tiring to split a rubber ball in to 10 smaller pieces. You can use a large hammer and try to do that using quite a lot of energy or you can use a very sharp blade to cut the rubber ball requiring way less energy. For a glass ball that will be the opposite and PLA is closer to rubber than glass or stone.
But does the artificial "intelligence" say how a machine must look like? That's why I was asking for CAD files.
"Training data" means searching the web and calculating some sort of average answer. Still a search engine to me ;-)
Without any link to the "training data" there is: You can believe it, but you don't have to.
Where did Gemini get the specific energy for PLA from?
That cutting is in general better than random milling is a statement I made in the video, didn't I?
@@HomoFaciens In order to calculate some sort of average answer from a lot of data you need to be able to understand what the data means. They only predict what the next most likely word is but we may be no different from that.
The energy required to grind the PLA could be wrong but I think the strategy to use low energy is slow cutting instead of smashing.
I could ask Gemini if you made that statement in the video and it will be able to answer saving me about 14 minutes :)
I can not currently ask for CAD files but it will not be more than one or two years before this will be a possibility.
I asked about the energy needed to melt 1kg of PLA and the answer was Q = 1000 g * 0.93 J/g°C * (175°C - 20°C) = 144750 J but LLM's are still very bad at math as it is 144150J
Searching the specific heat for PLA will be a nightmare using google and you get all sort of wild results including the first one that says "1200 J/Ng/K" what is with that Ng ? could be ng but then the unit will be incorrect. Looking in the fist PDF link there are two values 1200 and 1800J/kg/K so the first value is not far off 930 used by Gemini vs 1200
The melting temperature is also fairly accurate at 175°C
In any case 144150J is about 40Wh thus 25Wh to grind 1kg of PLA is not unrealistic.
As told before, the core problem of Gemini is: You can believe it or not. There is no way to see on what data the software generated the answer. Gemini delivers a result in seconds, but to prove if that result is correct, you still have to spend hours of research work ;-)
There is no "Intelligence" behind the code lines, just parameters, self adjusted by the same piece of software. Shit in, shit out is what all database software suffers from...
type in title (pastic)
Thanks for pointing to that typo! I have corrected that.
Typo in comment (type) 😂