I just want to call out how impressive being able to mill the solder mask off of PCB pads is. With the laser, you might be able to spray paint the board white and make a second pass to make the "silkscreen" layer.
Very impressive. As always, you seem to answer any question I could come up with in addition to making me aware of things that would never occur to me.
Still trying to decide if this is a quality product, but it’s becoming fairly apparent to me that *way* too many engineering hours went into this for it to be half assed. Some of these solutions are really novel and fairly challenging to implement, and yet they’ve done it. I’m liking what I see!
I would change to a wider mill bit (like 1-2 mm) for clearing those large surfaces and leave the delicate structures to the 0.2 mm bit. If working with FR4, the glass will consume those expensive mill bits quickly.
This looks really nice. And the price on the Kickstarter page is way less than what I was expecting (I figured at least $5k ~ $8k). Hopefully that price is fairly similar after launch, might look into grabbing one. Although I am curious to see how well it works on aluminium and brass (eg. for small parts and aluminum front panels, etc.)
With Kickstarter you accept full risk of nondelivery in addition to extended lead time. It sounds like a $5k machine with a $2k discount for accepting terms that make you an investor as much as a customer.
Excellent review. The lines on that filter look very clean, much cleaner than I was expecting. I will definitely be looking out for the filter design tutorial especially the simulation and mechanical design aspects. I hope it doesn't require access to super expensive unobtanium software which is not practical for the average hobbyist/maker. Having Pooch engraved on your phone cover put a smile on my face, he's a lucky boy.
Oh i love to see more reviews in the maker and tooling space! I have the exact same 3DWOX 1 which was visible in the background. It works great for me.
Thank you so much for this detailed review. I am one of the funding members and feel much better about my potential new machine. I hope it is as good as the prototype. And I will subscibe to your channel in hope of seeing your future product improvements on this machine!
Congratulations on the 200th broadcast of TSP! Thank you very much for all the effort you always put in and for the very interesting contributions. I look forward to the upcoming 'signals'. Best regards.
I dunno, I've seen a few tours of Chinese production shops (the ones that crank out the world's PCBs) and they seem to use these 4mm (?) shanks as their workhorses.
I wonder witch end mil is used for the PCB engraving, does not look like a standard v shaped spiral drill or engrave bit which is often used, its more cylindrical, and what is the diameter (I assume 0.2mm and maybe a 1mm long shaft ?) -- whats the outer diameter of this ?
Would love to find out how it also works for panels (for enclosures) and custom RF enclosures. If it can do all that competently... I want one! But the kickstarter ends in 12 days! After that it'll be a lot pricier.
I'm disappointed to see it won't be able to handle steel milling. I'm still looking at a tiag micro mill, it would be great to have auto tool changing.
@@testuser564 Yeah a lot of people asked about that on the kickstarter campaign. I doubt you could do thick pieces of steel without some kind of cooling/lubricant, but they posted this the other day: ruclips.net/video/VTWt6elB6eg/видео.html
14:00 No D for tools. Wonder if it just defaults to D1 or if it only supports one offset for each tool? Also no G54. Does it not support multiple origo. Or is it just assuming G54 if it's not included in the g-code?
and of course this shows my ignorance about how laser engraving works, but I don't understand why the phone case not being completely level would affect the fine details of the resulting image / engraving? I mean isn't this is "single" beam of laser perpendicular to the bed? is there a lense that focuses the picture at a particular depth?
Check out 27:06 ish and look at the laser path in the smoke. It's a cone, not a straight beam. So it has to be refocused to keep up with the variable surface
@@6alecapristrudel yeah, you are correct, it definitely looks like it is focused cone. honestly I am surprised. I would have imagined it would be much more expensive to create a cone of laser and focus it, rather than just have a "single beam" of laser. clearly I need to learn more about this.
@@AliMoeeny All laser beams are cones due to diffraction. If you want to form a high resolution pattern you need a lens with a tight focus. The tighter the focus the shorter the depth of focus and the flatter the object needs to be (or the better the focus tracking needed). Additionally the output of a diode laser is a diverging cone, not a beam, so there has to be a lens in order to form a spot. The choice isn't lens vs no lense, but rather high resolution lens with shallow focus vs low resolution lens with deep focus.
Respectfully, an unloaded repeatability test on Z is probably going to give you misleading results because gravity is removing the backlash you want to measure. Some more valuable tests might be to look at Z deflection under cutting loads and to try to test for collet pull-out over time. You also have the equipment to be able to carry out a vibration frequency analysis, which would be informative. Most machinists will be wincing whilst watching this because of the high levels of audible chatter whilst cutting what are relatively soft materials compared to the claims made for the machine. This chatter will greatly reduce tool life and increase the likelihood of scrapped parts. A reasonable analysis would show whether this is due to insufficient work holding, tool deflection issues, backlash in the ways, or more general stiffness problems. A happy machine should cut with a musical “hmmmmmmmmmmm” note. If the machine is yelling “aaarrrrrrrrrrrrrrggggghhhhh!” then something is awry and you will experience surface finish problems or tool breakage.
Thank you for your comments. Indeed the unloaded Z-axis test is not optimal, at the same time PCBs don't pose a major load. The test did show the unit outperforms the spec by a reasonable margin (room for it to get worse under load). The noise suggestion is a great test. I will be sure to try it out especially on the final production unit.
If you plan to cover this in more depth, I might suggest looking at the implications of deflection when using tiny carbide PCB milling bits. If you have backlash, or if there is flex in the machine, then what is most relevant is not the absolute play so much as the amount of excessive cut per tooth proportional to the diameter of the cutter. Carbide is very brittle and so oscillating chatter forces can shatter small bits even when the play in the system is small in absolute terms. Might be amusing to show how this is akin to RF resonance problems but just at much lower frequencies. Sometimes it is useful to have a nice, slow, physical system that you can directly sense to help you understand the behaviour of complex RF systems that you have to observe indirectly.
Is there any support for fiducials or other multi-layer alignment features? I would be KEENLY interested if this could create double-sided laminates that could be sandwiched, and I know I'm not the only one. Many designs are almost required to be 4-layer at this point, or more with those pesky BGAs.
@@TimLF that's not that bad for such a machine. You pat already 1500 for the "normal" Chinese variant, that has half of the stuff and there you are also on your own.
Does the slow, low power 200 W spindle motor cause any problems? When even cheap $150 models come with 300 W motors this one seems woefully undersized! Also some users claim that speeds up to 24,000 RPM gives you faster cutting with less chatter while this one runs at only 15,000.
bit misleading to claim servos are more "accurate", this simply isnt true, servos, like closed loop steppers, are able to track their position, leading to no or less positional errors compared with traditional steppers.
THIS IS MESSY: Makera should make an upgraded version of this video and point to it prominently, by referencing it everywhere, ESPECIALLY on the Kickstarter main page at the top, so one can get the real structure of how the Carvera is driven!!! After wasting my time on dozens of stOOpid HYPE videos, I finally luck up on the only video I could have spent time on!!! so, I now think I may have more of my questions answered: This is confusing since the explanation is not precise... left to interpretation. Hard to decide in what order these assumption/questions should be stated... as some of them become moot once answer to another is given: computer = tablet (hardware) controller = Carvera's controller hardware Carvera Interface = the still unnamed Makera/Carvera interface software runs on the computer. !? = a question unless the assumption listed is correct. UGS/G-code Sender ? = where is that step performed, by what module? is the Carvera Interface module the sender? ~ confusion you speak of GRBL files as if different than G-code ??? on other sites GRBL is shown as the Post-processor(driver?) that is set in the CAM software required to save the CAM format as G-code!? is GRBL(a parser) embedded into the G-code file? ...with the G-code as its data? ...thus the G-code is G-code + GRBL parser code?(firmware) ...since the only file used as input by the sender is the G-code file once it is produced at the CAM!? ...and does that dataset no already include all the parameters, like tools paths, etc... or are some of the parameters still inserted in the set by the Sender? (Carevera's Interface module???) so no need for UGS/G-code Sender!? - Can not the G-code be loaded to the Carvera's controller via a mem stick? no need for Sender? ...then press a button on the Carvera's controller and the job starts! - The computer software(Carvera's Interface) accesses the Carvera controller's memory since we read the file from the internal controller's storage !? files/jobs(G-code) from the computer storage are passed that to the controller's internal memory storage!? how does the G-code land in the controller's memory!? -what software module provides the toolpaths setup, and other needed parameters to be inserted in the G-code? - Must a computer be always plugged in(wi-fi) to the controller!? ...so the trigger to start the job is a software one only(on the computer)!? Once the file is loaded, all the parameters can be adjusted then a trigger actuated to start the job!? by what? how/where?
This machine is *incredibly* poor value for money. You can get a proper 6040 router(or 3040 if you really have not enough space) for ~$1500, and another $2000 for a really powerful, properly built ATC spindle with a real ISO20 taper, not toy collets. Hell, if you have space, you can get a pre-built, pre-configured 6090 cast iron floor-mounted router with the aforementioned ATC spindle for _less_ money. This thing has what I'd call "all the features" but if they were built as an educational demo rather than a practical tool. It has an ATC, emulating proper tool holders - but it's crippled. It has a wireless probe, emulating Renishaw style probes - but it's crippled by way lower accuracy(microswitch instead of 3-point contact method). It has servos and ballscrews - that's a plus, but crippled by thin aluminium extrusion(50kg total machine weight is feather light) - 10 microns of play on a completely unloaded tiny machine is garbage. It has a nice UI - but the python UI is responsible for motion control, which is an *incredibly* stupid thing to do, and others have already reported failed jobs due to crashes. A bare minimum for good machines is LinuxCNC(which is also far more extensible). They try to have a laser - but there's *no safety* whatsoever. They provide *no good goggles* - insane! Not to mention straight up illegal, since you're in the US. A few watts of blue will absolutely destroy your eyes immediately with a single 4% reflection. I wouldn't buy this machine for more than $1000, and I don't recommend anyone does either. Spend a few hours digging into alternatives.
I don't know how familiar you are with CNCs, we have 6040,3040 hobby level CNCs, we have lots of manual iron-cast vertical mills that were transferred into CNCs, we have homemade 5-axis CNC machines, we have 1325 wood carving machines, we have access and have years of VMC machine center experience, and industrial control software user experiences. But we still use Carvera every day right in our office, for every small project, esp PCB prototyping. I don't know why you compare the Carvera and the parts with others which are nonsense to compare. Why you don't compare the machine to something similar on the market if you do some investigation in advance? The different machines have different targeting audiences.
Youre thinking like a machinist, not an electrical engineer who wants to cut PCBs out. It's also nice for tooling not to cost a large percentage of the machine's price. Also, don't fool yourself into believing that a more powerful spindle will make those Chinese machines any more capable. They are limited by machine rigidity just as much as this one. For what it does, it's fine
I think you're comparing apples to oranges. A lot of people just aren't in the market for a big lump of cast iron. A lot of people don't have the skills to deal with no-name Chinese machinery or the time to figure it out for themselves. This machine is remarkably good value when compared to something like the Roland MDX-50.
This repeatability is not so good. The machine would be more precise if they upgraded some things but clearly it is lacking. I have worked on very large CNC machines with 10 times better precision than this.
You try and make a point that is not valid in this case. This machine fits on your desktop and doesn't require a forklift or warehouse to move or operate. Try comparing a similar sized machine to its repeatability.
Holy crap, I want one. This is the first time I've seen worthwhile PCB performance from a desktop CNC.
I just want to call out how impressive being able to mill the solder mask off of PCB pads is. With the laser, you might be able to spray paint the board white and make a second pass to make the "silkscreen" layer.
Great idea!
Very impressive. As always, you seem to answer any question I could come up with in addition to making me aware of things that would never occur to me.
@11:05 I like how they use one of the off the shelf dust collector from a robot vacuum cleaner.
Still trying to decide if this is a quality product, but it’s becoming fairly apparent to me that *way* too many engineering hours went into this for it to be half assed. Some of these solutions are really novel and fairly challenging to implement, and yet they’ve done it. I’m liking what I see!
I would change to a wider mill bit (like 1-2 mm) for clearing those large surfaces and leave the delicate structures to the 0.2 mm bit. If working with FR4, the glass will consume those expensive mill bits quickly.
He said it was FR1.
@@jaro6985 Yes. I wanted to point out that cutting glass based FR4 is tougher for the mill bits than paper based FR1.
Dear Santa,
I know I've been naughty - again - but...
Awesome machine! TSP logo engraved on the cover is very satisfying to see.
90% of this went way over my head but I loved every second of it. What an impressive machine
This looks really nice. And the price on the Kickstarter page is way less than what I was expecting (I figured at least $5k ~ $8k). Hopefully that price is fairly similar after launch, might look into grabbing one. Although I am curious to see how well it works on aluminium and brass (eg. for small parts and aluminum front panels, etc.)
With Kickstarter you accept full risk of nondelivery in addition to extended lead time. It sounds like a $5k machine with a $2k discount for accepting terms that make you an investor as much as a customer.
@@jjoonathan7178 They advertise a 38% discount, so yes, indeed a $5K+ machine
Mine finally shipped! Can't wait to use it
Excellent review. The lines on that filter look very clean, much cleaner than I was expecting. I will definitely be looking out for the filter design tutorial especially the simulation and mechanical design aspects. I hope it doesn't require access to super expensive unobtanium software which is not practical for the average hobbyist/maker. Having Pooch engraved on your phone cover put a smile on my face, he's a lucky boy.
CopperCAM for PCBs is about $80, not that bad.
@@jaro6985 I was talking about software for designing the microwave components filters, splitters, circulators, etc, not the cam part.
On kickstarter there is now a cheaper version: Carvera Air. Hope this will be almost accurate as the these Carvera
Would love to replace my table top CNC machine with this auto level/tool change unit. Thanks for the detailed review.
Thanks!
How noisy is it , practical for homelab usage?
With the door closed it is not bad, certainly useable for my case.
Really nice!
Oh i love to see more reviews in the maker and tooling space! I have the exact same 3DWOX 1 which was visible in the background. It works great for me.
I bought the 3DWOX1 based on his previous review, and also have the 2X. So helpful to have real-deal reviews.
Thank you so much for this detailed review. I am one of the funding members and feel much better about my potential new machine. I hope it is as good as the prototype. And I will subscibe to your channel in hope of seeing your future product improvements on this machine!
Congratulations on the 200th broadcast of TSP! Thank you very much for all the effort you always put in and for the very interesting contributions. I look forward to the upcoming 'signals'. Best regards.
I have watched several of these reviews and I have yet to see any of them show the controller board or motion system.
Superb machine. Really nice product
I wish this had a better spindle with a more common "industrial taper" like iso-20. Probably was not possible in the budget.
Yeah, those collets look like something from nail saloon.... toy machine. Not meant to do real metalwork.
I dunno, I've seen a few tours of Chinese production shops (the ones that crank out the world's PCBs) and they seem to use these 4mm (?) shanks as their workhorses.
Nice review ! What software do you use for design (CAD) the filter PCB and what software for the CAM part?
What a review Shahriar ! Thanks 1000 times for your work ! I can't wait to see the video of the CNC made RF filters !
What did your friend use to design the hairpin filter?
This is surprisingly awesome little machine
I wonder witch end mil is used for the PCB engraving, does not look like a standard v shaped spiral drill or engrave bit which is often used, its more cylindrical, and what is the diameter (I assume 0.2mm and maybe a 1mm long shaft ?) -- whats the outer diameter of this ?
Would love to find out how it also works for panels (for enclosures) and custom RF enclosures. If it can do all that competently... I want one! But the kickstarter ends in 12 days! After that it'll be a lot pricier.
I'm disappointed to see it won't be able to handle steel milling. I'm still looking at a tiag micro mill, it would be great to have auto tool changing.
It can mill steel, you just have to be slow and careful with it. They posted a video to their youtube demonstrating that the other day.
@@oolong2 that's good to know I did not see any thing talking about that
@@testuser564 Yeah a lot of people asked about that on the kickstarter campaign. I doubt you could do thick pieces of steel without some kind of cooling/lubricant, but they posted this the other day:
ruclips.net/video/VTWt6elB6eg/видео.html
Does ATC use air whilst changing tools, ie. what method does it use for the ATC
14:00 No D for tools. Wonder if it just defaults to D1 or if it only supports one offset for each tool?
Also no G54. Does it not support multiple origo. Or is it just assuming G54 if it's not included in the g-code?
Ah, its using G54 in another file.
Looks like a good build for people who want to mill soft aluminum and stay sane
have you built any other PCBs with this machine yet?
I have. It shows up in some of my other videos.
This is a nice machine for the money. Did you use a backer material under the PCB when cutting it out from the panel?
Yes, I used another PCB under the top one.
Cool tool :) Leather is a bit of a worry as most modern Tanning processes use Chromium so play carefully with the Laser.
"this is of course a picture of Audrey Hepburn" :)
and of course this shows my ignorance about how laser engraving works, but I don't understand why the phone case not being completely level would affect the fine details of the resulting image / engraving? I mean isn't this is "single" beam of laser perpendicular to the bed? is there a lense that focuses the picture at a particular depth?
Check out 27:06 ish and look at the laser path in the smoke. It's a cone, not a straight beam. So it has to be refocused to keep up with the variable surface
@@6alecapristrudel yeah, you are correct, it definitely looks like it is focused cone. honestly I am surprised. I would have imagined it would be much more expensive to create a cone of laser and focus it, rather than just have a "single beam" of laser. clearly I need to learn more about this.
@@AliMoeeny All laser beams are cones due to diffraction. If you want to form a high resolution pattern you need a lens with a tight focus. The tighter the focus the shorter the depth of focus and the flatter the object needs to be (or the better the focus tracking needed).
Additionally the output of a diode laser is a diverging cone, not a beam, so there has to be a lens in order to form a spot. The choice isn't lens vs no lense, but rather high resolution lens with shallow focus vs low resolution lens with deep focus.
keren mesin CNC nya
Will you share the files, please?
I want!!!
What software does this machine work with? Mach3/4?
The dust collection looks very similar to something from a Xiaomi robot cleaner.
The software probably has Reversal Error Compensation; maybe even Leadscrew Error Compensation.
Respectfully, an unloaded repeatability test on Z is probably going to give you misleading results because gravity is removing the backlash you want to measure. Some more valuable tests might be to look at Z deflection under cutting loads and to try to test for collet pull-out over time.
You also have the equipment to be able to carry out a vibration frequency analysis, which would be informative. Most machinists will be wincing whilst watching this because of the high levels of audible chatter whilst cutting what are relatively soft materials compared to the claims made for the machine. This chatter will greatly reduce tool life and increase the likelihood of scrapped parts. A reasonable analysis would show whether this is due to insufficient work holding, tool deflection issues, backlash in the ways, or more general stiffness problems.
A happy machine should cut with a musical “hmmmmmmmmmmm” note. If the machine is yelling “aaarrrrrrrrrrrrrrggggghhhhh!” then something is awry and you will experience surface finish problems or tool breakage.
Thank you for your comments. Indeed the unloaded Z-axis test is not optimal, at the same time PCBs don't pose a major load. The test did show the unit outperforms the spec by a reasonable margin (room for it to get worse under load).
The noise suggestion is a great test. I will be sure to try it out especially on the final production unit.
If you plan to cover this in more depth, I might suggest looking at the implications of deflection when using tiny carbide PCB milling bits. If you have backlash, or if there is flex in the machine, then what is most relevant is not the absolute play so much as the amount of excessive cut per tooth proportional to the diameter of the cutter. Carbide is very brittle and so oscillating chatter forces can shatter small bits even when the play in the system is small in absolute terms. Might be amusing to show how this is akin to RF resonance problems but just at much lower frequencies. Sometimes it is useful to have a nice, slow, physical system that you can directly sense to help you understand the behaviour of complex RF systems that you have to observe indirectly.
Is there any support for fiducials or other multi-layer alignment features? I would be KEENLY interested if this could create double-sided laminates that could be sandwiched, and I know I'm not the only one. Many designs are almost required to be 4-layer at this point, or more with those pesky BGAs.
Is the Spindle reversible? And is the Spindle variable speed?
Not sure about reversible, but speed can be controlled.
There is no reason to reverse a high speed spindle like this that I can think of.
Hello Shahriar. Do you have any price point for this machine?
Please check the Kickstarter page. It has all the information.
Darn nice machine, but probably if I ask, I can't afford it :-p
US $3,250
@@TimLF that's not that bad for such a machine. You pat already 1500 for the "normal" Chinese variant, that has half of the stuff and there you are also on your own.
👍👍
Does the slow, low power 200 W spindle motor cause any problems? When even cheap $150 models come with 300 W motors this one seems woefully undersized! Also some users claim that speeds up to 24,000 RPM gives you faster cutting with less chatter while this one runs at only 15,000.
It is fine for my needs, mostly PCB and Al work.
16:00 1/1000 of a mm is.... 1 micrometer! But I am sure you knew that :) I do understand you are in the land of silly fractions.
Dear Santa... xD
👍👍💖🙏💖👍👍
bit misleading to claim servos are more "accurate", this simply isnt true, servos, like closed loop steppers, are able to track their position, leading to no or less positional errors compared with traditional steppers.
THIS IS MESSY: Makera should make an upgraded version of this video and point to it prominently, by referencing it everywhere, ESPECIALLY on the Kickstarter main page at the top, so one can get the real structure of how the Carvera is driven!!! After wasting my time on dozens of stOOpid HYPE videos, I finally luck up on the only video I could have spent time on!!! so, I now think I may have more of my questions answered:
This is confusing since the explanation is not precise... left to interpretation.
Hard to decide in what order these assumption/questions should be stated... as some of them become moot once answer to another is given:
computer = tablet (hardware)
controller = Carvera's controller hardware
Carvera Interface = the still unnamed Makera/Carvera interface software runs on the computer.
!? = a question unless the assumption listed is correct.
UGS/G-code Sender ? = where is that step performed, by what module? is the Carvera Interface module the sender?
~ confusion you speak of GRBL files as if different than G-code ???
on other sites GRBL is shown as the Post-processor(driver?) that is set in the CAM software required to save the CAM format as G-code!? is GRBL(a parser) embedded into the G-code file? ...with the G-code as its data? ...thus the G-code is G-code + GRBL parser code?(firmware) ...since the only file used as input by the sender is the G-code file once it is produced at the CAM!?
...and does that dataset no already include all the parameters, like tools paths, etc... or are some of the parameters still inserted in the set by the Sender? (Carevera's Interface module???) so no need for UGS/G-code Sender!?
- Can not the G-code be loaded to the Carvera's controller via a mem stick? no need for Sender?
...then press a button on the Carvera's controller and the job starts!
- The computer software(Carvera's Interface) accesses the Carvera controller's memory since we read the file from the internal controller's storage !?
files/jobs(G-code) from the computer storage are passed that to the controller's internal memory storage!?
how does the G-code land in the controller's memory!?
-what software module provides the toolpaths setup, and other needed parameters to be inserted in the G-code?
- Must a computer be always plugged in(wi-fi) to the controller!?
...so the trigger to start the job is a software one only(on the computer)!?
Once the file is loaded, all the parameters can be adjusted then a trigger actuated to start the job!? by what? how/where?
Copper tape, Xacto knife versus CNC machine versus cheap PCB from China.
0.001 is one micron
0.001mm is 1um.
This machine is *incredibly* poor value for money. You can get a proper 6040 router(or 3040 if you really have not enough space) for ~$1500, and another $2000 for a really powerful, properly built ATC spindle with a real ISO20 taper, not toy collets.
Hell, if you have space, you can get a pre-built, pre-configured 6090 cast iron floor-mounted router with the aforementioned ATC spindle for _less_ money.
This thing has what I'd call "all the features" but if they were built as an educational demo rather than a practical tool. It has an ATC, emulating proper tool holders - but it's crippled. It has a wireless probe, emulating Renishaw style probes - but it's crippled by way lower accuracy(microswitch instead of 3-point contact method). It has servos and ballscrews - that's a plus, but crippled by thin aluminium extrusion(50kg total machine weight is feather light) - 10 microns of play on a completely unloaded tiny machine is garbage. It has a nice UI - but the python UI is responsible for motion control, which is an *incredibly* stupid thing to do, and others have already reported failed jobs due to crashes. A bare minimum for good machines is LinuxCNC(which is also far more extensible).
They try to have a laser - but there's *no safety* whatsoever. They provide *no good goggles* - insane! Not to mention straight up illegal, since you're in the US. A few watts of blue will absolutely destroy your eyes immediately with a single 4% reflection.
I wouldn't buy this machine for more than $1000, and I don't recommend anyone does either. Spend a few hours digging into alternatives.
An NXP 1769 running Smoothieware is "responsible for motion control", the UI basically just sends the gcode and commands to it.
@@techydiy which is horrible. The UI shouldn't be responsible for gcode.
I don't know how familiar you are with CNCs, we have 6040,3040 hobby level CNCs, we have lots of manual iron-cast vertical mills that were transferred into CNCs, we have homemade 5-axis CNC machines, we have 1325 wood carving machines, we have access and have years of VMC machine center experience, and industrial control software user experiences.
But we still use Carvera every day right in our office, for every small project, esp PCB prototyping.
I don't know why you compare the Carvera and the parts with others which are nonsense to compare. Why you don't compare the machine to something similar on the market if you do some investigation in advance?
The different machines have different targeting audiences.
Youre thinking like a machinist, not an electrical engineer who wants to cut PCBs out. It's also nice for tooling not to cost a large percentage of the machine's price. Also, don't fool yourself into believing that a more powerful spindle will make those Chinese machines any more capable. They are limited by machine rigidity just as much as this one. For what it does, it's fine
I think you're comparing apples to oranges. A lot of people just aren't in the market for a big lump of cast iron. A lot of people don't have the skills to deal with no-name Chinese machinery or the time to figure it out for themselves. This machine is remarkably good value when compared to something like the Roland MDX-50.
This repeatability is not so good. The machine would be more precise if they upgraded some things but clearly it is lacking. I have worked on very large CNC machines with 10 times better precision than this.
You try and make a point that is not valid in this case. This machine fits on your desktop and doesn't require a forklift or warehouse to move or operate. Try comparing a similar sized machine to its repeatability.