Cheap-and-cheerful Taobao $150 knock-off running at 200mm/sec

this is absolutely hilarious, i so was not expecting this to work, to have regretted buying this printer and to be writing about how i wasted crowdfunding backers money. instead, i am astounded to be able to report that a cheap-and-cheerful $150 taobao knock-off can be modified to run with decent quality output at 200mm / sec. that's 200mm/sec PRINTING speed from a 0.4mm nozzle at a layer height of 0.2mm.

a full documentation of the analysis etc. and a log of the (necessary) improvements made is here:
[hands.com]

A video with lots of camera-shake, and a lot of noise, is here: [www.youtube.com]

Edited 2 time(s). Last edit at 05/25/2017 05:01AM by lkcl.

Are you sure it is printing at 200mm/s? Small parts means it will actually spend most of its time accelerating.

Also, I'm not sure I would rate those pieces as high quality myself.

Quote
Origamib
Are you sure it is printing at 200mm/s? Small parts means it will actually spend most of its time accelerating.

i checked an acceleration curve: 4000mm/s^2 means that it takes only about 6mm to get up to 200mm/s. so you're right, it would rouuughly be equivalent to... a 150mm/s static speed.

since writing this initial post i've added printing of a 160mm long part: i did that at 175mm / sec because the amount of shaking of the frame is being subjected to was so high it was actually travelling across the floor

that one came out far better than when i used to run the mendel90 at 200mm/sec.

Quote


Also, I'm not sure I would rate those pieces as high quality myself.

no, and neither would i. if i ran it at only 50mm/sec the quality would be drastically improved. however i am comparing against what i used to run a mendel90 at, and it is definitely a significant improvement.

remember that i have 3 man-YEARS worth of 3D-printing to do (@ 8hrs/day @ 200mm/s) in order to fulfil a crowd-funding campaign, so a little less "high standards" is needed and a bit more "realistic pragmatism" needs to be put in its place

i also have some corrections to make to the (newly designed) x-idler. the X-belt is becoming loose because i haven't supported both ends of the 5mm bolt through the idler. i'll get to that shortly. basically this is a work-in-progress origamib, with the idea being to maximise the "mm / sec / $" metric for batch processing.

How can be speed related to "travelling across the floor"? Travel = force, force = mass * acceleration, speed is nowhere.

Quote
sigxcpu
How can be speed related to "travelling across the floor"? Travel = force, force = mass * acceleration, speed is nowhere.

you are correct, i misspoke: it's down to acceleration not speed. ok so the printer is currently on a stone tiled floor, in an apartment here in taiwan. there is very little in the way of friction between the extruded aluminium of the base and the stone tiles on which it is sitting. the weight of the printbed and the weight of the direct-drive motor in each of y and x independently, in combination with the increased acceleration, is sufficient to cause the printer to break loose from the tiles due to the lack of friction and actually scuttle and jitter about across the floor. it's very funny to watch

ok i have a friend making one of these, here's build pictures (hi A!) blue is replaced, black is original. it's mostly... blue....

all the parts at the back are original. they're okay. the y endstop i used a piece of paper to change the angle at which it sits (it works...) so that the endstop clicks with the printbed bearing holder. extruder is original, carriage is not.



z endstop holder is fine. z rod holder (top), replaced. z motor holder is fine. x-end motor holder, i still just... don't get why people don't get that hanging anything out off-centre on vertical x-rods is anything other than asking for trouble... hence: x-ends are replaced, horizontal rod arrangement.



ditto. z rod holder (off top of picture) replaced. z motor holder fine. x-end idler holder, replaced. carriage, replaced.



y-rod holders ok. y-idler holder: total piece of shit. snapped immediately. z motors fine.



ok underside of printbed, it's VERY IMPORTANT that you create that stand-off for the y-belt. add the bolts first, lock them down with M3 nuts to the plate. add EXTRA NUTS (loose), *THEN* add the plastic part, THEN add a THIRD set of nuts, and use nuts 2 and 3 to make ABSOLUTELY SURE that the belt lines up DEAD HORIZONTAL with the motor. standard trigonometry: angle MUST be zero otherwise belt and motor have a non-linear relationship depending on Y travel distance.

LM8UU bearing holders are fine.

ok this is the new x-end idler holder, takes 3 M3 bolts to hold it together. 2 underneath, but the 3rd one there's a nut trap on the side of the x-end, drop an M3 nut into that then put a horizontal M3 bolt in. you can't do that whilst the M5 idler bolt is in (shown in the picture), it's in the way. i've reorganised this for the motor idler, these parts i had already printed (sorry!)

p.s. these parts are printed very very fast on a work-in-progress printer and haven't been cleaned up in any way.



Edited 1 time(s). Last edit at 08/12/2017 04:09AM by lkcl.

so those are light-sprites at the top (of course)... the main carriage (yellow) is PETG after a bad experience with a fan failure ended up melting the extruder... *sigh*... the part-cooling fan is also in PETG (red) as it will be a little bit close to the printbed. the fan ducts are all done with pyopenscad / polysurface, spec'd out with simple one-liner functions to create circles (or squares) as required, then passed to the spline-generating function (another one-liner).

still TBP: the z rod end holders and the y idler. at some point i will replace *all* the parts so that this can be sold as a (verified!!!) kit from HK/Taiwan/China-sourced parts.

lkcl, I seriously doubt it was printing at anything near 200 mm/s. You really cannot print small objects that fast.

One of the main problems when printing small objects is cooling. You cannot print a molten layer of plastic on top of another molten layer of plastic. The layer below has to cool enough so it can fully support the layer above or you will get a really bad surface finish. That is why every slicer has a setting that will slow the print down if the layer takes less than X seconds. That setting is enabled by default.

In Slic3r the setting is under the Cooling tab and is called "Enable Auto Cooling". Slic3r will slow the print speed all the way down to 15 mm/s to try and get a layer to take at least 25 seconds. There are multiple settings that modify the outer perimeter, infill, solid layer and other speeds. Simplify3D has a similar array of settings and I'm sure all other slicers have them too.

It's rather hard to quantify the actual printing speed (especially for small objects) . You should look at the generated g-code. You can see what your slicer used for the printing speed of the different parts of the print.

You mention "maximizing mm/s per dollar", but that seams like a rather meaningless metric. It's relatively easy to make the mechanics move at 200+ mm/s and spit plastic out at that speed, but if the resultant print looks like a blob of goo, the speed means nothing.

As pointed out earlier, the "default print speed" in the slicer cannot be used as a measurement of how fast your printing. It's an input, not an output. The g-code is also not 100% reliable because the mechanics of the machine, controller board, firmware and firmware settings can all influence the effective print speed.

The only way (that I can think of) to get some meaningful metric of how fast a printer can print, would be to come up with a test object, define what an acceptable result would look like and then see how fast you can print the object and still get an acceptable result. The time would be measured in minutes and seconds, not mm/s . Note that the object would have to be large enough that a single layer will take more than 25 seconds at your target speed (or you will have to create some kind of cryogenic cooling system ).

Regards
Josef

Quote
xyze
lkcl, I seriously doubt it was printing at anything near 200 mm/s. You really cannot print small objects that fast.

that is correct: in a previous post we checked an acceleration graph and the maximum speed for the 4000mm/sec^2 setting used the speed it could hit within 20mm was more like 150. it's still a staggering achievement for something that costs so little money.

Quote


One of the main problems when printing small objects is cooling. You cannot print a molten layer of plastic on top of another molten layer of plastic. The layer below has to cool enough so it can fully support the layer above or you will get a really bad surface finish. That is why every slicer has a setting that will slow the print down if the layer takes less than X seconds. That setting is enabled by default.

also correct: the bottom layers (about... 2.5 square centimetrs) were definitely running at full (acceleration-constrained) speed, whilst the top layers which were walls 3mm thick and around 25mm in total were definitely slowed down.

you can check the video for yourself. it's fricking funny.

Quote

In Slic3r

ARGH! slic3r (which is written by a perl programmer who fails to respond to bugreports) makes such a god-awful mess of slicing that i gave up on it in horror a looong time ago. you know that perl is called a "WORN"-out language (Write Once, Read Never) for a very good reason, right? it's so hostile to code readability that not even the programmer who wrote it can understand what they did.

loading up a complex part into repsnapper or pronterface quickly shows just how bad slic3r really is. i *really* cannot recommend it.... anyway... sorry...

Quote

the setting is under the Cooling tab and is called "Enable Auto Cooling". Slic3r will slow the print speed all the way down to 15 mm/s to try and get a layer to take at least 25 seconds. There are multiple settings that modify the outer perimeter, infill, solid layer and other speeds. Simplify3D has a similar array of settings and I'm sure all other slicers have them too.

yes. i tend to set that quite low. 4 seconds or so.

Quote

It's rather hard to quantify the actual printing speed (especially for small objects) . You should look at the generated g-code. You can see what your slicer used for the printing speed of the different parts of the print.

indeed.

Quote

You mention "maximizing mm/s per dollar", but that seams like a rather meaningless metric.

why?

Quote

It's relatively easy to make the mechanics move at 200+ mm/s and spit plastic out at that speed, but if the resultant print looks like a blob of goo, the speed means nothing.

exactly! so if that happens, then you set a hard limit on the speed, and that in turn sets a corresponding (directly proportional) maximum budget that should be set for buying components. it's not that hard, josef.

examples:

* ultimaker-2 costs $1200 and is rated for 250mm/sec but let's say that it turns out "blobs of goo" at that speed, so you have to ramp it down to say... 150mm / sec. the budget for buying a set of 3D printers is $3000, you can only buy 2 ultimaker-2 3D printers (and have some change left over).

maximum build-farm speed achievable without producing "goo": 2 x 150mm / sec = 300 mm / sec.

* cheap and cheerful taobao knock-off which costs $150 and is nominally rated for 40mm / sec but with a bit of tweaking can do 100mm / sec. the budget is still $3000 so you can buy TWENTY cheap and cheerful taobao knock-offs. 100 mm / sec times twenty equals *2000* mm / sec.

maximum build-farm speed using the knock-off: 2000 mm / sec (a whopping SIX TIMES improvement).

can you see anything which would indicate that that's a meaningless metric? (see below last paragraphs before responding)

Quote

As pointed out earlier, the "default print speed" in the slicer cannot be used as a measurement of how fast your printing. It's an input, not an output. The g-code is also not 100% reliable because the mechanics of the machine, controller board, firmware and firmware settings can all influence the effective print speed.

i'm also finding that i can set the edge output speed to 80mm / sec to reduce corner-ringing whilst keeping the infill speed high. so it's even more complex. with having to experiment with the extruder max-acceleration in the firmware right now (on another printer using a flex3drive) yes you're absolutely right: it's *even more* complex.

Quote

The only way (that I can think of) to get some meaningful metric of how fast a printer can print, would be to come up with a test object, define what an acceptable result would look like and then see how fast you can print the object and still get an acceptable result. The time would be measured in minutes and seconds, not mm/s . Note that the object would have to be large enough that a single layer will take more than 25 seconds at your target speed (or you will have to create some kind of cryogenic cooling system ).

indeed

the target project is to produce 150 sets of 35 parts, total plastic around 240 g per object, for which a rough finger-in-the air calculation for a single 3D printer would be three YEARS of 24x7 3D printing.

reducing that time is easy if you have $USD 20,000 to spend on a batch of the fastest and greatest high-quality USA/EU-sourced 3D printers, but given that i am answerable to some eco-conscious backers of a crowd-funding campaign i have to be a leeetle bit more financially responsible and substitute "creative thinking" for "brute force economic muscle" (unlike what google did with Project ARA... *sigh*....)

so
as this is a serious operation not a "nice to do in spare time" operation i had indeed worked out that the final figure which really matters is "object print time", alongside "acceptable quality" of course. so yes, i'm aware that the metric really should be "minimise print time per dollar" or better "maximise objects printed per time unit per dollar": i just hadn't fully verbalised that, and used the mm / sec / $ metric insteaad, as a non-linear simplistic near-approximation.

realistically, even things like startup time, cool-down time (with or without fans... which again are an extra cost), whether the objects are batchable or not, whether the objects can be *reliably* printed batched or not - all these factors need to be taken into consideration. i left them out of the discussion here as it's primarily about this cheap 3D printer. most of the budget-optimisation discussions have taken place on the arm-netbooks mailing list for the (larger) project.

all that having been said, your input and clarification is really appreciated: i'll definitely stop referring to mm/sec/$ and think of something else more useful (units/hr/$ for example) instead.

ok just some assembly photos and instructions (hello A) - you want to attach the hobbed gear to the NEMA17 before attaching the NEMA17 to the extruder: you can't get at it afterwards without disassembly. don't overtighten the (tiny) grub screws: strip those or the allen hex key and you've had it. push the allen key all the way in then BEND it slightly to the side (off-centre) with quite a bit of force, as you tighten it down. the side-loading pushes the allen key into the grooves of the grub screw hex hole and helps prevent it from stripping both the tool and the grub screw. use common sense though

the small bearing fits into the extruder arm "slot" and there should be an M3 hex nut indentation on the other side. use the correct length bolt so that it doesn't scrape on the NEMA17 or the extruder when the arm is fitted place.

bolts for the NEMA17: make sure you don't put ones in that are too long, check the depth of the NEMA17 holes (i think they're 5 or 6mm). the one in the bottom left corner goes through the filament arm as well as the main extruder body. there *is* no screw in the top *left* corner - that's ok.

small springs and their bolts are held in place by two "indented" hex nuts, you can see the slots from the top (blurry... sorry....) photo.

it's fairly obvious stuff. don't overtighten the arm-holder bolt, the arm won't move. two very long bolts go through the bottom and hold the hot-end in place: don't put those on quite just yet because you have to put it through the carriage and *then* lock the bolts that hold the hotend in place. bit of a pain but hey.





Edited 1 time(s). Last edit at 08/19/2017 01:15PM by lkcl.

ok when attaching the extruder to the carriage it's important to identify the correct orientation. this is a picture from the FRONT. the belt holders must be at the front of the carriage: this then identifies that as "front of carriage". extruder NEMA17 must then be to the RIGHT, not the left.

Quote
lkcl

Quote
xyze
lkcl, I seriously doubt it was printing at anything near 200 mm/s. You really cannot print small objects that fast.

that is correct: in a previous post we checked an acceleration graph and the maximum speed for the 4000mm/sec^2 setting used the speed it could hit within 20mm was more like 150. it's still a staggering achievement for something that costs so little money.

Have you looked at the gcode file ? Did you confirm that the slicer actually wrote F12000 for the speeds? Your calculations should be based on the values in the gcode file and not the values in your slicer, since those are the values that the firmware will be reading.

Quote
xyze

Quote
lkcl

Quote
xyze
lkcl, I seriously doubt it was printing at anything near 200 mm/s. You really cannot print small objects that fast.

that is correct: in a previous post we checked an acceleration graph and the maximum speed for the 4000mm/sec^2 setting used the speed it could hit within 20mm was more like 150. it's still a staggering achievement for something that costs so little money.

Have you looked at the gcode file ? Did you confirm that the slicer actually wrote F12000 for the speeds? Your calculations should be based on the values in the gcode file and not the values in your slicer, since those are the values that the firmware will be reading.

someone asked a similar question when i first started running a mendel90 at 200mm/sec about 3 or so years ago. i took a different approach: i simply timed how long it took to get from one side to the other as i was printing 230mm long parts across the diagonal: 1 second. the mendel90 extruder at those speeds would sometimes break the thread, which is why, after a lot of thought, it occurred to me to set skeinforge's "outer edge" speed setting much lower (80 to 100 mm/sec).

anyway yes here's some of the code for a part i did today (not on this cheap-and-cheerful printer, that's in bits at the moment) - the travel speed is set on this one to 200mm/sec, inner fill to 180 and the edge is.. i think it's 80mm/sec. there's a F3000 in there which is 50mm/sec, i can't exactly remember all the settings in skeinforge.

acceleration (in the duet 0.8.5 config settings) i think i ramped down to 3500mm/s^2 because even at 80mm/s edge speed i'm getting quite a lot of ringing on corners, even with the bug-fixes for arcs @ corners i've added. i also set quite a large "smoothing" on the internal fill (3mm) which causes the extruder to not be thrown around quite so violently. shakes the desk and there's marks on it from the aluminium where it skitters about.... anyway, for "functional" parts i'm not so concerned about the ringing, i just want to get parts done and out the door.

[edit: oh! i remember why there's such a quick extruder move: it's a retract. i set retract speed to 250mm/sec, and with the acceleration on the extruder set in the duet firmware to 75 mm/s - remember there's a 40:1 wormdrive on a Flex3Drive so that seems very low but it actually isn't) i let the duet take care of deciding what actual speed the retract should happen at, which is very small anyway. ]

G1 E4.56432 F15000.0
G1 X107.9439 Y83.1029 Z0.2 F12000.0
G1 X107.7725 Y83.3408 Z0.2 F12000.0
G1 X107.6551 Y83.6094 Z0.2 F12000.0
G1 X103.1837 Y97.7763 Z0.2 F12000.0
G1 X102.8712 Y98.2436 Z0.2 F12000.0
G1 X102.3565 Y98.4696 Z0.2 F12000.0
G1 X80.8252 Y101.2853 Z0.2 F3000.0
G1 X80.6186 Y101.1712 Z0.2 F3000.0
G1 X80.6662 Y100.94 Z0.2 F3000.0
G1 E4.66432 F15000.0
G1 X81.3644 Y100.3113 Z0.2 F900.0 E4.67381
G1 X81.7274 Y100.0174 Z0.2 F900.0 E4.67853

Edited 2 time(s). Last edit at 08/20/2017 03:55PM by lkcl.

Do you still have the gcode file that was used to print the 200 mm/s part in the video ? Should be easy to look at that file, no ?

All of the G1 ... F12000 lines listed above are not extruding any filament. Only the last 2 lines are extruding filament (while moving) and those are at 15 mm/s.

Quote
xyze
Do you still have the gcode file that was used to print the 200 mm/s part in the video ? Should be easy to look at that file, no ?

no of course not: it was weeks ago, and i really don't keep track of auto-generated file output (edit: attached typescript. it *might* be one of the 70 files in the v3.0 folder. but now that i think about it it might be in the v2.0 folder.)

Quote


All of the G1 ... F12000 lines listed above are not extruding any filament. Only the last 2 lines are extruding filament (while moving) and those are at 15 mm/s.

probably because it was at the beginning (first layer) where the first layer speed is set very low - here's an arbitrary section from somewhere in the middle of the file. layer height... 0.4 so second layer.

G1 X120.9622 Y110.1103 Z0.4 F10800.0 E55.1281
G1 X121.0621 Y110.0696 Z0.4 F10800.0 E55.12919
G1 X121.1479 Y110.0041 Z0.4 F10800.0 E55.13028
G1 X121.2526 Y109.8993 Z0.4 F10800.0 E55.13178
G1 X121.2693 Y109.8513 Z0.4 F10800.0 E55.13229
G1 X121.2418 Y109.8084 Z0.4 F10800.0 E55.13281
G1 E55.03281 F15000.0
G1 X103.6861 Y99.0566 Z0.4 F12000.0
G1 X103.4888 Y98.8317 Z0.4 F12000.0
G1 X103.4536 Y98.5346 Z0.4 F12000.0
G1 X103.5929 Y98.2698 Z0.4 F12000.0
G1 E55.13281 F15000.0
G1 X117.6511 Y84.2116 Z0.4 F10800.0 E55.33367
G1 X117.6662 Y84.1357 Z0.4 F10800.0 E55.33446
G1 X117.6018 Y84.0927 Z0.4 F10800.0 E55.33524
G1 X117.3771 Y84.0927 Z0.4 F10800.0 E55.33751
G1 X117.2735 Y84.1133 Z0.4 F10800.0 E55.33857
G1 X117.1857 Y84.172 Z0.4 F10800.0 E55.33964
G1 X103.2592 Y98.0985 Z0.4 F10800.0 E55.53863
G1 X103.1833 Y98.1136 Z0.4 F10800.0 E55.53941
G1 X103.1403 Y98.0492 Z0.4 F10800.0 E55.54019
G1 X103.1403 Y97.8246 Z0.4 F10800.0 E55.54246

Edited 1 time(s). Last edit at 08/20/2017 07:31PM by lkcl.

Quote
lkcl

Quote
xyze
Do you still have the gcode file that was used to print the 200 mm/s part in the video ? Should be easy to look at that file, no ?

no of course not: it was weeks ago, and i really don't keep track of auto-generated file output (edit: attached typescript. it *might* be one of the 70 files in the v3.0 folder. but now that i think about it it might be in the v2.0 folder.)

Then I guess we will never know what the actual speed was for the object in your video. Posting gcode for a different object does not help. As pointed out in my original post, the speeds in the gcode file can be completely different from what you set in the slicer, depending on the size of the object and other settings.

Quote
lkcl
* cheap and cheerful taobao knock-off which costs $150 and is nominally rated for 40mm / sec but with a bit of tweaking can do 100mm / sec. the budget is still $3000 so you can buy TWENTY cheap and cheerful taobao knock-offs. 100 mm / sec times twenty equals *2000* mm / sec.

maximum build-farm speed using the knock-off: 2000 mm / sec (a whopping SIX TIMES improvement).

I think there are a few things missing in your calculations. The base cost of the printer is $190. $150 + $40 for shipping. Then there's the replacement parts like a new Melzi board, Mk3 printbed, etc. How much labour goes into upgrading the printer. How do you account for that ? How often will the cheap printers break down when run 24/7 ?

Did you measure the time it took to print the part at 50 mm/s and the time it took to print the part at 150 mm/s? It's actual printing time that matters, not the slicer setting. How can you base all of your calculations on a slicer setting ? Father Time does not care what you tell the slicer you would like to print at, it only cares about the actual printing time

Did you print the part before you made all of your improvements so you could compare before and after to confirm that the money and time you spent was worth it ?

I'm sorry to say, but the end result of your changes is certainly not "astounding". If I gave you money and you send me that part, I would send it back and ask for a refund. Even the part printed at "50 mm/s" is not of acceptable quality, so unless you can improve the quality to an acceptable level, all the money you "save" on cheap printers will ultimately be wasted. Quality should be your #1 priority, and then go from there. But that's just me. Others may be completely happy with those parts.

Quote
xyze

Then I guess we will never know what the actual speed was for the object in your video. Posting gcode for a different object does not help. As pointed out in my original post, the speeds in the gcode file can be completely different from what you set in the slicer, depending on the size of the object and other settings.

in this case, after dealing with 3d printng for almost 4 years now you get a "feel" for the sound at certain speeds, and i've also looked at other files and found that what is set is what is outputted, with the exception of the shorter layers. it was definitely running "quick" and you can confirm that from the video alone.

Quote

Quote
lkcl
* cheap and cheerful taobao knock-off which costs $150 and is nominally rated for 40mm / sec but with a bit of tweaking can do 100mm / sec. the budget is still $3000 so you can buy TWENTY cheap and cheerful taobao knock-offs. 100 mm / sec times twenty equals *2000* mm / sec.

maximum build-farm speed using the knock-off: 2000 mm / sec (a whopping SIX TIMES improvement).

I think there are a few things missing in your calculations.

.

yes, of course. i did not outline the full scenario: someone else's full scenario will be different from the very particular full scenario that i am targetting, which is specifically to fulfil a crowdfunding campaign's series of printed parts (over 5,000 in total).

Quote

The base cost of the printer is $190. $150 + $40 for shipping. Then there's the replacement parts like a new Melzi board, Mk3 printbed, etc

indeed. the melzi is... well.. i haven't thrown it in the bin, but it might as well be there. the MK3 was borrowed from another printer. all of this is entirely missing the point: i bought this printer because it was a fast way to get a complete set of parts at a reasonable price.

*when* it comes to buying as-many-3d-printers-as-will-fit-into-a-$USD3000-budget, i'll optimise *that* specifically for the purposes of maximising the units-per-hour-per-dollar. taking into account bulk part shipping cost, whether people in the new crowdfunding campaign are willing to pay for shipping because they're individuals with some spare cash to help out so it's not so critical for them to minimise or even include shupping costs.

it's quite complex and you're focussing on the wrong things, at the wrong time, in effect, josef. the point of the post was to have a laugh at how ridiculously low-cost the pritner is (they included a $40 spool of rubbish 1.75mm PLA in that $150 - i didn't know they were going to do that and i would have asked them to take that off the price if i had known, for example), compared to how stupidly fast it can be run without falling to bits.

Quote

. How much labour goes into upgrading the printer.

it doesn't matter because time is not an important factor, nor am i paying anyone else to do it.

Quote

How do you account for that ?

in a libre-run open community-backed crowdfunded project people are volunteering their time and expertise... so we don't account for that.

Quote

How often will the cheap printers break down when run 24/7 ?

we'll find out, won't we? luckily that's why you get 20 of them for $USD 3,000 so that one breaking doesn't matter so much.

Quote

Did you measure the time it took to print the part at 50 mm/s and the time it took to print the part at 150 mm/s?

no, because i was too busy laughing at the thing skittering across the stone tiles of the apartment.

Quote

It's actual printing time that matters, not the slicer setting.

indeed.

Quote

How can you base all of your calculations on a slicer setting ?

as i said already: i'm not.

Quote

Father Time does not care what you tell the slicer you would like to print at, it only cares about the actual printing time

indeed.

Quote

Did you print the part before you made all of your improvements so you could compare before and after to confirm that the money and time you spent was worth it ?

no i did not. i compared against the results of a mendel90 which i previously owned and used to run at 200mm/sec. which had a really good part cooling system (where this one had absolutely none at all at the time of the tests).

Quote

I'm sorry to say, but the end result of your changes is certainly not "astounding". If I gave you money and you send me that part, I would send it back and ask for a refund. Even the part printed at "50 mm/s" is not of acceptable quality, so unless you can improve the quality to an acceptable level, all the money you "save" on cheap printers will ultimately be wasted. Quality should be your #1 priority, and then go from there. But that's just me. Others may be completely happy with those parts.

none of the prints that i did at the time had a part cooling fan: the original didn't come with one and i hadn't designed one to fit into the (very hastily-replaced) horizontal carriage arrangement. i also hadn't - and still haven't - finished stiffening up the frame. so to make a quality-based judgement - based on an unfinished design adjustment on something that's still being assessed and improved - really isn't useful.

the purpose of this exercise is to maximise units printed per hour per dollar, with this _particular_ design being focussed on minimising the cost of components (whereas the other printer i'm designing is focussed on higher quality parts - as opposed to lower priced parts - as a way to increase the target). if we're *right at the beginning* of that exercise, it's not really helpful to make judgements as a way to say that the entire exercise must be invalid, is it?

overall, josef, you're drilling down to a level of detail which really isn't that important at this stage of the exercise, to reach a premature conclusion at best, on something that's still being analysed. can i ask: what is the purpose of doing that?

Edited 1 time(s). Last edit at 08/21/2017 05:53PM by lkcl.

Quote
lkcl
the purpose of this exercise is to maximise units printed per hour per dollar, with this _particular_ design being focussed on minimising the cost of components (whereas the other printer i'm designing is focussed on higher quality parts - as opposed to lower priced parts - as a way to increase the target). if we're *right at the beginning* of that exercise, it's not really helpful to make judgements as a way to say that the entire exercise must be invalid, is it?

If you can't print an acceptable part in X amount of time, then what is the point of trying to print it in half or a third of the time and then basing your calculations on times that do not produce acceptable parts? If you are really at the beginning of the exercise, then you should not be making claims like "(a whopping SIX TIMES improvement)".

Quote
lkcl
overall, josef, you're drilling down to a level of detail which really isn't that important at this stage of the exercise, to reach a premature conclusion at best, on something that's still being analysed. can i ask: what is the purpose of doing that?

Luke, I would not consider any of the things I mentioned as "detail", they are all very high level basic things to consider. I do not understand how you can view the most basic metrics that your calculations are using as "a level of detail which really isn't that important at this stage".

The only purpose of my original post was to point out that you are not printing at the speed that you think you are and that you are looking at the wrong metric. I thought you might be interested in getting that right.

For determining the speed it need some more informations:

* Acceleration (Default and max !), Yerk, max speeds of the firmware (upload configuration .h)
* Accelerations / speeds entered at the slicer software (best to upload the settings file)
* complete Gcode file - upload it
* Time the whole print took.

Chri

---

[chrisu02.wordpress.com] Quadmax Intel Delid Tools

Quote
xyze

If you can't print an acceptable part in X amount of time, then what is the point of trying to print it in half or a third of the time and then basing your calculations on times that do not produce acceptable parts? If you are really at the beginning of the exercise, then you should not be making claims like "(a whopping SIX TIMES improvement)".

josef, can you appreciate that it's not very useful to compare a preliminary (near-)first print against a planned hypothetical proposed target? even if this printer *after improvements have been made* turns out to only be capable of... 100mm/sec or even less that's still a huge improvement in the units/hour/budget figure.

give it time ok!

Quote


The only purpose of my original post was to point out that you are not printing at the speed that you think you are and that you are looking at the wrong metric. I thought you might be interested in getting that right.

thank you for pointing that out (several times), i do appreciate that you've made the effort to communicate that, and yes i'm changing the wording i'll use in the future to "units/hr/budget" (or similar, still TBD). i apologise we seem to be getting into detail rather than acknowledging that, yes i am interested in moving *towards* getting it right, on an appropriate schedule. i have a lot to do, and a lot going on: my current focus isn't on improving this particular printer on an accelerated schedule right this minute: it's on the other one. and getting the next revision PCB out the door for the crowdfunding campaign.

so i have a suggestion: i actually really appreciate the focus that you have on detail, which i would otherwise miss (because i have a different kind of mind and focus). it is however a little early: i have a friend assembling one of these (as you can see from the build posts), he's a mechanical engineer just starting on the learning process about 3D printing. could you help him out, when he has questions? i'm attempting to get him to join here, however he's a little reticent after his first experience joining a local community 3D printing forum turned into a bit of a... umm... i want to use colloquial words here that might not be politically correct, whoops

Quote
Chri
For determining the speed it need some more informations:

* Acceleration (Default and max !), Yerk, max speeds of the firmware (upload configuration .h)
* Accelerations / speeds entered at the slicer software (best to upload the settings file)
* complete Gcode file - upload it
* Time the whole print took.

Chri

hi Chri, thanks for this - when my friend has assembled his version (first ever 3D printer) or when i have a new controller board and have time to redo some of the improvements i made for my friend (i.e. print them out), and then reassemble it, i'll be able to provide that. right now mine is partially disassembled as i needed parts from it to put the other printer together. i only have the one Duet 0.8.5 (and it would be too costly to add to this china clone anyway, the Duet being more costly than the entire set of parts for this printer!!)

*briefly* - as this is not my primary focus and i haven't operated the machine for at least two weeks now - i can make the marlin firmware section available, it looks like i was using absolutely mad acceleration rates and max X/Y speed rates, to let the slicer have "free rein". i can't recall if i was still using Cura 21.04 at the time or if i had already switched back to skeinforge - it's been too long.

i'll be able to focus on this printer again in good time: give it a couple weeks ok?

// default settings 
//Anycubic
//#define DEFAULT_AXIS_STEPS_PER_UNIT   {80,80,392.6,E_STEPS_PER_MM}  
// Taobao thing
#define DEFAULT_AXIS_STEPS_PER_UNIT   {100,100,397.6,E_STEPS_PER_MM}  

#define DEFAULT_MAX_FEEDRATE          {400, 400, 6, 30}    // (mm/sec)
#define DEFAULT_MAX_ACCELERATION      {5000,5000,150,8000}    // X, Y, Z, E maximum start speed for accelerated moves. E default values are good for skeinforge 40+, for older versions raise them a lot.

#define DEFAULT_ACCELERATION          5000    // X, Y, Z and E max acceleration in mm/s^2 for printing moves 
#define DEFAULT_RETRACT_ACCELERATION  5000   // X, Y, Z and E max acceleration in mm/s^2 for r retracts

Quote
lkcl

so i have a suggestion: i actually really appreciate the focus that you have on detail, which i would otherwise miss (because i have a different kind of mind and focus). it is however a little early: i have a friend assembling one of these (as you can see from the build posts), he's a mechanical engineer just starting on the learning process about 3D printing. could you help him out, when he has questions? i'm attempting to get him to join here, however he's a little reticent after his first experience joining a local community 3D printing forum turned into a bit of a... umm... i want to use colloquial words here that might not be politically correct, whoops

That is exactly why I do not participate in online forums much either. They too often degenerate into pointless arguments (or much worse). This is definitely one of the better behaved forums. I'm glad you are taking my input as constructive criticism, which is exactly how it is meant.

I can certainly try to help if your friend has questions, but Google already has the answer, and yes, it really is "42" .

Cheers and good luck with your project.

Josef

Quote
xyze

That is exactly why I do not participate in online forums much either. They too often degenerate into pointless arguments (or much worse). This is definitely one of the better behaved forums. I'm glad you are taking my input as constructive criticism, which is exactly how it is meant.

*nods*... been looking around, someone referred me to the 3d printing subreddit, and i found it surprisingly active and informative.

Quote

I can certainly try to help if your friend has questions, but Google already has the answer, and yes, it really is "42" .

Cheers and good luck with your project.

thxman. can't but help my friend from making glaring google-info-mistakes: he's not quite sure yet what he's looking at... or for... if you know what i mean: keeps sending me this random stuff...

Edited 1 time(s). Last edit at 08/23/2017 03:51PM by lkcl.

ok so my friend's started the build, here's the basic frame. he's bought extra 100mm base uprights so as to be able to use 40x40mm triangles in all corners on the base. every single corner possible has three 40x40mm triangles, except where that interferes with parts that are going on, next. this still leaves the z-uprights in need of some form of U-shaped outer bracing (dibond, lexan, similar U-shape bracing to the mendel90) or cross-bracing at the back, in order to achieve full rigidity. otherwise the z-uprights can still parallelogram.

with the base having *twelve* 40x40mm triangles that forms a complete rigid cube: it's not going anywhere, so unlike many mendel style repraps does not rely on gravity for its print quality.




Edited 2 time(s). Last edit at 08/27/2017 03:28AM by lkcl.

I love to overbuild things, but that's just ridiculous. Are the ends of the frame members milled square and bolted directly to each other, or is he using all those brackets to hold all the pieces in square alignment?

Maybe he had a huge, free supply of t-slot hardware. I can't think of any other reason someone would create that mess over just buying larger t-slot extrusions. Simply using larger t-slot would have been much cheaper and more rigid, and would have looked a lot nicer. If that's 4040 t-slot, for that small of a frame it is more than rigid enough without any of the corner braces. I hope he's not planning to use end support guide rails and plastic motor/pulley mounts.

That reminds me a bit of this monstrosity:



I can't imagine the thought process that led to that, unless it was something like "I want to make, the ugliest, nastiest looking, most difficult to service and operate coreXY printer I can possibly think of." In that case, goal achieved!

Edited 7 time(s). Last edit at 08/27/2017 11:30AM by the_digital_dentist.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
the_digital_dentist
I love to overbuild things, but that's just ridiculous. Are the ends of the frame members milled square and bolted directly to each other, or is he using all those brackets to hold all the pieces in square alignment?

haha no they're not bolted to each other, it's just to hold everything together. first 3D printer he's ever built but he's a mechanical engineer. and in HK.

Quote

Maybe he had a huge, free supply of t-slot hardware.

huge yes free no but he _did_ order 1,000 40x40x20 cast triangles wholesale from a factory in shenzhen, and the t-nuts from the market in the Futian District of Huaqiang Road... and a bulk job-lot of 25mm and 40mm fans... and another batch of 50mm blowers...

the 40x40x20 cast triangles turned out to be ten times cheaper than by the time they get to Europe or the US. however it was a leetle excessive even for him getting the MOQ of 1,000, so if ever you want some i can put you in touch and he can sell them to you ok? cheap @ 1/2 the price, my frieeeend

Quote

I can't think of any other reason someone would create that mess over just buying larger t-slot extrusions. Simply using larger t-slot would have been much cheaper and more rigid, and would have looked a lot nicer. If that's 4040 t-slot, for that small of a frame it is more than rigid enough without any of the corner braces. I hope he's not planning to use end support guide rails and plastic motor/pulley mounts.

it's 2020... and this *is* the "cheap-and-cheerful" thread where we're attempting to maximise the "units per hour per dollar" metric (thanks for the encouragement on that josef). so unless something can be found that's easy and cheap to make in metal here in TW, HK or Shenzhen, it'll be staying as plastic.

2020 is dirt cheap here, because not just of 3D printing but also cheap taobao-sold CNCs, and also those low-cost BGA Infra-red Soldering Stations (T962 if i remember correctly), and many many more random bits of equipmentt, they all use 2020 extrusion.

Quote

That reminds me a bit of this monstrosity:

20150809_211147_preview_featured.jpg

oooo i liiike it, it's so prettyyyyyy

Quote

I can't imagine the thought process that led to that, unless it was something like "I want to make, the ugliest, nastiest looking, most difficult to service and operate coreXY printer I can possibly think of." In that case, goal achieved!

ironically, it's not very stiff either. the triangle braces down to the middle of the uprights simply transfer any parallelogramming or rhombus-ing to the *centre* of those uprights, causing them to bend and flex. it's *really* important to have cross-bracing that's put under tension, which then creates compression on the uprights, taking up any "slack" in the material properties of the 2020 extrusion.

ok so at 200 mm / sec i am still getting "skipping". this is with 0.9 deg steppers. i will be putting in the 84 oz-in 48mm long 2A beasts to see how those fare: they're 1.8 degree / step.

at 150mm / sec and 1000 mm / min min-jerk speeds this is a successful print. the artefacts (stringing) is from skeinforge. still configuring there...

the top view is where the "corner bulging" has been eliminated, thanks i believe to using the high min-jerk speeds (effectively disabling acceleration protection).

i really want to track down what's going on with this massive skipping, see whether it's firmware or just running the NEMA17s too fast

[www.youtube.com]

okay! photos to follow, my friend has his fully assembled and we're doing commissioning. last main thing to do which will have to wait a couple of days is the extruder. getting the tension right to avoid slipping or overcurrent is proving challenging.

also we had the printbed MOSFET go up in smoke so will be going to the markets to get a replacement TO-220 N-channel MOSFET or five. all good stuff, not bad ffor a very first 3D printer ever

Does this printer model have a name?