Electronics > Projects, Designs, and Technical Stuff
Right motors, PSU for CNC mill
langwadt:
--- Quote from: 3dgeo on September 21, 2019, 05:58:01 pm ---
--- Quote from: langwadt on September 21, 2019, 10:34:02 am ---nothing special about the pcb, main objective was that I wanted four drivers with connectors fit in a 100mm wide box.
if you want modular just buy some X-NUCLEO-IHM03A1
--- End quote ---
Did you use regular copper thickness or double? Do you have an idea what max current they can drive with your PCB design with acceptable temperature?
After making it how do you feel – is it worth the time? I know there is X-NUCLEO-IHM03A1, but according to ST they are only for testing only...
You mind sharing machine you using it on?
--- Quote from: langwadt on September 21, 2019, 10:34:02 am ---they aren't particularly smooth or silent, if you want that tmc5160/tmc2160 with external fets is a better choice and they have standard step/direction drive so will work with standard motion controllers
--- End quote ---
Why they are not smooth or silent? They have 128 microstepping, even tho it isn't 256, but still, they should be silent enough, what gives?
tmc5160/tmc2160 adds an extra step – figuring out right mosfets, and I'm fine with SPI, I actually prefer SPI instead of step/direction.
--- End quote ---
just standard copper. I don't know what current they can handle but they are fine for my ~4A need, above that the small current sense resistors I used are probably at the limit.
as with most diy electronics, if you have to ask if it is worth the time you shouldn't be doing it
ST says it is for testing only because they haven't bother with all the approval testing and they sell them at give away prices
The machine is cobbled together from aluminium extrusions, mgn12 rails, 1204 ball screws and an 800W spindle
they are smooth and when milling you can't hear them, but at stand still they hiss
there is no advantage in needing spi for direction and it means it won't work with any standard motion controller code
3dgeo:
--- Quote from: langwadt on September 21, 2019, 06:29:53 pm ---just standard copper. I don't know what current they can handle but they are fine for my ~4A need, above that the small current sense resistors I used are probably at the limit.
--- End quote ---
Yeah, I don't think my motor will use more than 4A, so I can do 10x10cm as well, tho I might make it from thicker copper. I looked at external mosfets – suitable ones cost a LOT, 36 fets in total – not worth it in this case, so I will stick with all in one chip.
--- Quote from: langwadt on September 21, 2019, 06:29:53 pm ---as with most diy electronics, if you have to ask if it is worth the time you shouldn't be doing it
--- End quote ---
I rephrase it – does it works as you expected? Any regrets, design mistakes? ;D
--- Quote from: langwadt on September 21, 2019, 06:29:53 pm ---The machine is cobbled together from aluminium extrusions, mgn12 rails, 1204 ball screws and an 800W spindle
they are smooth and when milling you can't hear them, but at stand still they hiss
--- End quote ---
I'm considering mgn12 and ball screws, but I'm not sure they worth that extra cost compared to regular 8mm lead screw and C beams with plastic wheels. I've seen videos that mgn12 on aliexpress are junk, that's why I'm hesitating, how do they work for you and where did you buy them?
--- Quote from: langwadt on September 21, 2019, 06:29:53 pm ---there is no advantage in needing spi for direction and it means it won't work with any standard motion controller code
--- End quote ---
Well, I didn't dig in to powerSTEP01 SPI features, but I'm guessing that stalling detection and other features are communicated via SPI and interrupt pin. Plus I need to learn Gcode anyway, so writing my own controller code to work with SPI is not an issue. Also I just like fancyness of SPI ^-^
langwadt:
--- Quote from: 3dgeo on September 21, 2019, 10:06:29 pm ---
--- Quote from: langwadt on September 21, 2019, 06:29:53 pm ---just standard copper. I don't know what current they can handle but they are fine for my ~4A need, above that the small current sense resistors I used are probably at the limit.
--- End quote ---
Yeah, I don't think my motor will use more than 4A, so I can do 10x10cm as well, tho I might make it from thicker copper. I looked at external mosfets – suitable ones cost a LOT, 36 fets in total – not worth it in this case, so I will stick with all in one chip.
--- Quote from: langwadt on September 21, 2019, 06:29:53 pm ---as with most diy electronics, if you have to ask if it is worth the time you shouldn't be doing it
--- End quote ---
I rephrase it – does it works as you expected? Any regrets, design mistakes? ;D
--- Quote from: langwadt on September 21, 2019, 06:29:53 pm ---The machine is cobbled together from aluminium extrusions, mgn12 rails, 1204 ball screws and an 800W spindle
they are smooth and when milling you can't hear them, but at stand still they hiss
--- End quote ---
I'm considering mgn12 and ball screws, but I'm not sure they worth that extra cost compared to regular 8mm lead screw and C beams with plastic wheels. I've seen videos that mgn12 on aliexpress are junk, that's why I'm hesitating, how do they work for you and where did you buy them?
--- Quote from: langwadt on September 21, 2019, 06:29:53 pm ---there is no advantage in needing spi for direction and it means it won't work with any standard motion controller code
--- End quote ---
Well, I didn't dig in to powerSTEP01 SPI features, but I'm guessing that stalling detection and other features are communicated via SPI and interrupt pin. Plus I need to learn Gcode anyway, so writing my own controller code to work with SPI is not an issue. Also I just like fancyness of SPI ^-^
--- End quote ---
you can get suitable FETs $0.20, the dual dual fet used on the TMC5160-stepStick is ~$0.50
it works as expected, only mistake was two swapped pins on a uart connector. I would have like more of the unused pins on the CPU to be accessible and I should have kept the option to run the motors at higher voltage
even bad mgn12/ballscrews are miles better than plastic wheels on c-beam and lead screws, I bought them from robotdigg
I also bought some reasonable nice hgh15 rails on ebay, 2*1500mm + 4 carriages for ~100gbp
SPI is nice for setting currents and reading errors, but for direction it is a hassle, and it has nothing to do with gcode, it is the
firmware that needs modifications
I never got the stall detection to work on the powerstep01, the datasheet it limited on how should be done (to say the least) and
I'm not even sure it works when not using the internal motiongenerator
3dgeo:
--- Quote from: langwadt on September 21, 2019, 11:16:51 pm ---you can get suitable FETs $0.20, the dual dual fet used on the TMC5160-stepStick is ~$0.50
--- End quote ---
I think it defeats the purpose of using external fets if fets are that puny. If go with external fets than go all out and use something like IPT015N10N5 tho I bet TMC5160 would choke trying to drive 4 of those beasts with 16000pF capasitance each >:D Even if it can drive those puppies – total mosfet cost would be 130 usd. Ok, I know those are overkill, but I don't see any point of using small external fets with high resistance.
I guess all come to programming...
--- Quote from: langwadt on September 21, 2019, 11:16:51 pm ---it works as expected, only mistake was two swapped pins on a uart connector. I would have like more of the unused pins on the CPU to be accessible and I should have kept the option to run the motors at higher voltage
--- End quote ---
Swapping uart pins – so common mistake... :D Why didn't you break all pins out? I always do, even if I'm not using them... Whats is limiting the voltage? This is one of the features I like about powerSTEP01 drivers, their max voltage is 85V. I want to have option to drive control board from 60 to 80V.
--- Quote from: langwadt on September 21, 2019, 11:16:51 pm ---even bad mgn12/ballscrews are miles better than plastic wheels on c-beam and lead screws, I bought them from robotdigg
--- End quote ---
Are you sure about this? Cos I think it is way easier to make round plastic wheels and aluminum profiles with high precision than to make rails and keep them straight while shipping. And even if their precision are the same, price difference is huge. I really should go to dedicated CNC forum to find this out to be honest...
--- Quote from: langwadt on September 21, 2019, 11:16:51 pm ---
I also bought some reasonable nice hgh15 rails on ebay, 2*1500mm + 4 carriages for ~100gbp
--- End quote ---
This video is why I'm so skeptic about rails:
--- Quote from: langwadt on September 21, 2019, 11:16:51 pm ---
SPI is nice for setting currents and reading errors, but for direction it is a hassle, and it has nothing to do with gcode, it is the
firmware that needs modifications
I never got the stall detection to work on the powerstep01, the datasheet it limited on how should be done (to say the least) and
I'm not even sure it works when not using the internal motiongenerator
--- End quote ---
Are you saying that powerSTEP01 has issues what comes to programming? If so than it's a deal breaker, bin there (and partially still am) with other chip whom documentation sucks and even manufacturer support can't help with the matter...
Do TMC5160 have issues/are pain to program as well?
langwadt:
--- Quote from: 3dgeo on September 22, 2019, 03:20:32 am ---
--- Quote from: langwadt on September 21, 2019, 11:16:51 pm ---you can get suitable FETs $0.20, the dual dual fet used on the TMC5160-stepStick is ~$0.50
--- End quote ---
I think it defeats the purpose of using external fets if fets are that puny. If go with external fets than go all out and use something like IPT015N10N5 tho I bet TMC5160 would choke trying to drive 4 of those beasts with 16000pF capasitance each >:D Even if it can drive those puppies – total mosfet cost would be 130 usd. Ok, I know those are overkill, but I don't see any point of using small external fets with high resistance.
I guess all come to programming...
--- End quote ---
the fets on the stepstick are 12A and ~6mOhm apart from the 40V rating much more than that would be pointless
--- Quote from: 3dgeo on September 22, 2019, 03:20:32 am ---
--- Quote from: langwadt on September 21, 2019, 11:16:51 pm ---it works as expected, only mistake was two swapped pins on a uart connector. I would have like more of the unused pins on the CPU to be accessible and I should have kept the option to run the motors at higher voltage
--- End quote ---
Swapping uart pins – so common mistake... :D Why didn't you break all pins out? I always do, even if I'm not using them... Whats is limiting the voltage? This is one of the features I like about powerSTEP01 drivers, their max voltage is 85V. I want to have option to drive control board from 60 to 80V.
--- End quote ---
breaking out the pins take space and even more if they have to go to a reasonable connector
the powersupplies are limited to ~30V input, I had intended to split the input in two, so the motor voltage was separate
--- Quote from: 3dgeo on September 22, 2019, 03:20:32 am ---
--- Quote from: langwadt on September 21, 2019, 11:16:51 pm ---even bad mgn12/ballscrews are miles better than plastic wheels on c-beam and lead screws, I bought them from robotdigg
--- End quote ---
Are you sure about this? Cos I think it is way easier to make round plastic wheels and aluminum profiles with high precision than to make rails and keep them straight while shipping. And even if their precision are the same, price difference is huge. I really should go to dedicated CNC forum to find this out to be honest...
--- End quote ---
the choice between rails and aluminium with wheels is the choice between cheap and good
--- Quote from: 3dgeo on September 22, 2019, 03:20:32 am ---
--- Quote from: langwadt on September 21, 2019, 11:16:51 pm ---
I also bought some reasonable nice hgh15 rails on ebay, 2*1500mm + 4 carriages for ~100gbp
--- End quote ---
This video is why I'm so skeptic about rails:
--- End quote ---
and if you look at his later videos he is using mgn12 an ballscrews in his latest design
--- Quote from: 3dgeo on September 22, 2019, 03:20:32 am ---
--- Quote from: langwadt on September 21, 2019, 11:16:51 pm ---
SPI is nice for setting currents and reading errors, but for direction it is a hassle, and it has nothing to do with gcode, it is the
firmware that needs modifications
I never got the stall detection to work on the powerstep01, the datasheet it limited on how should be done (to say the least) and
I'm not even sure it works when not using the internal motiongenerator
--- End quote ---
Are you saying that powerSTEP01 has issues what comes to programming? If so than it's a deal breaker, bin there (and partially still am) with other chip whom documentation sucks and even manufacturer support can't help with the matter...
Do TMC5160 have issues/are pain to program as well?
--- End quote ---
everything complicated is going to be a pain to program, but it the little I've looked at it it seems the documentation on the TMC is better
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version