MarcoReps ✅ AC Servos ✅ Open Source software and Linux repping ✅ Memes ✅ Teasing us with the probability of using those ac servos to make a 3D printer ✅ Thank you man, you're really doing a great job...
Having worked with EtherCat and Beckhoff equipment I can say that it's impressive how their communication protocols are literally identical to the original 80s implementation, just encapsulated into ethernet packets.
CAN works well, I'm glad they didn't try to roll their own version. And they have done a lot on the RT side to get the peformance as good as they have.
The BACnet and Modbus IP implementations are built similarly; they're thin ethernet wrappers around the original protocols, which while easy to implement and understand for people who spent their whole lives working with BACnet or Modbus, does result in some pretty clunky and redundant protocols with duplicated or unnecessary address fields, checksumming, bad port handling and inconsistent implementation of typical ethernet or higher conventions. If you wanted a well-designed clean implementation, you could strip out 80% of what industrial protocols like this do because ethernet/IP handles most of that functionality already. They leave it in though, because it's what they're used to.
@@siberx4 Yeah I did some work on it for a communication library and like 90% of the packet was communication overhead already handled by TCP/IP. The remaining 10% was "Send me variable XYZ". The cool thing about that is that, with an EtherCAT gateway, I could use the same 2020s Python library to communicate with a PLC from the 80s that was designed for a ring network.
Yay excited to see some more spotlight on EtherCat! I got to work with it for a fun robotics project in my lab, driving some high torque and high precision harmonic drive actuators.
About 6 years ago I made an opensource project ( the DieBieSlave ), an ethercat slave that connects a STM32 to a LAN9252 EtherCAT asic, the amount of attention and use that that project got over that time is amazing, from satellite dish uplink plants to exo skeletons ( Project MARCH ). EtherCAT is gaining more and more traction!
Never thought I’d hear VORON mentioned on one of your videos. I feel like my life is complete now. And yes, you should totally put some of these servos in a VORON and go bananas.
“HevORT” , “RatRig” , and “VZBot” are some other good printer designs along this line of thought to look into as well. I could dig for links, but i have tried to subscribe to like *all of the channels making that content*, so look around and ask if you need help! (Most of these Printers have huge Discord/Reddit Presence too) (Edit: Typo) (Edit 2: Added VZBot)
Yet another excellent release. You’re outdoing yourself this month, Marco. Highly entertaining, enjoyable, and informative, and I’m here for all of it. Thank you for always giving us your best. It’s highly appreciated. Stay safe, and we’ll see you on the next one.
I worked with EtherCAT on device development level using Beckhoff's ET1100 ASIC. It's an interesting protocol as bits are processed and interpreted on the fly as they are coming in, and are retransmitted only like 3 bits later.
Yeah, it's a pity that it isn't used more in the Hobbyist-space. Devices which implement EtherCAT are usually made for industrial use cases, that's at least what Beckhoff targets. That makes them way too expensive for hobby side projects.
@@thevayudev The reason you're seeing them here is because these companies are repurposing copycat precision parts that were mass produced into cheaper packages. I suspect in the next couple of years you're going to see people design home based multipurpose CNC tables that can drop any any tools you want. This isn't the first video I've seen like this, these parts are becoming more obtainable and you could connect them to simple router, plasma cutters, or 3D printer heads and do whatever you want with them. People around me are industrious so there are a few small businesses that are already putting them to work in my neighborhood, which is a very stretch from the traditional machine shops they were in.
@@snarkylive Okay, that does make a lot more sense, that's why none of these are original parts from Beckhoff Automation, those aren't really produced in that high volumes. It does sound like a bright(er) future for the low-budget space if other manufactures can mass produce these components while still being able to communicate with TwinCAT & other EtherCAT stuff. Still waiting for price cuts in the XPlanar/XTS space because those things look like a lot fun to play around with. I mean, couldn't you replace the rails of a CNC machine with XTS rails?
@@thevayudev XPlanar and XTS is definitely a newish thing on the market. Even their competitors (B&R, Bosch Rexroth, and many others) just launched these kinds of products in the last 4-5 years. I doubt we'll see price decrease until the next 5-10 years, especially with current chip issues. As for XTS replacing CNC machine rails? Well, I think it can be done, but your material weight will be quite limited, especially if you have to stack another XTS on top of the other for X-Y plane.
Fantastic stuff. I've recently built my first CNC controller using those "hybrid" stepper/servos and running LinuxCNC. The repetition for each axis did get a bit tedious and makes the innards look way more complicated than they actually are. This type of quality component is a dream for me. Kudos to the LinuxCNC/EtherCAT interface devs for making this available.
What model of controller/motor did you get? And do you use linuxcnc as well, or something like grbl? I'm currently running open loop steppers with grbl, very basic setup but affordable at 200€ for all electronics. However, I'd like to upgrade that system to something more reliable.
6:27 You explained this at the end of the video without explicitly linking the two. The encoder maintains itself and gives a position when requested. The battery allows the encoder to continue to operate even when the system is de-energised. Therefore, the encoder will continue to monitor the rotation of the shaft and keep the current position in its volatile memory. When the system is energised, the servo drive will ask the encoder for current position.
the rules of the speed boat race competition specifically disallow servo drives, but that doesn't mean that you shouldn't build an absolutely _bonkers_ fast printer with these. Vez3d did one with odrive servos at one point, but he changed back to just high power steppers afterwards. I think he's up to printing in the 1200mm/sec range maximum?
@@pmcquay1 This is stupid, should be the other way around - encourage development of servo drives, as steppers are boring and mundane and we need more and readily available servos, besides if you are participating in the speedboat challenge you already spent those thousands, coz your prusa or ender can do shit, its the vorons and the like.
@@VEC7ORlt It would be buy your way to victory with servos, which of course every industrial enterprise does. But for a hobbyist the cost is usually a serious barrier.
I do a lot of security work with industrial control systems (some ISO/IEC 62443 stuff, but mostly technical research/consultancy) and EtherCAT is pretty ubiquitous there, in everything from turnstile gates to cereal manufacturing to ship engines. It's one of the better protocols in my opinion - does its job, does it well, has extremely low latency, and doesn't bog down the base protocol with needless complexity. It's fast and reliable enough to be placed in a safety-critical alarm path, which is something I don't say lightly. In marine environments it's common to find it running alongside UDP-encapsulated NMEA for sensors and navigation comms, which saves a ton of cost and complexity on cabling.
The limiting factor for precision of FDM printers is 100% the extruder and filament. If you think about it, anything involving semi-molten plastic is guaranteed to be way less precise than motion control, even if you're just using belts. Also, at higher extrusion rate the heating necessary goes up substantially. Still, most speed 3d printers go with as light if an end effector as possible. It would be interesting to see what's possible if you go heavy (big servo E axis, big hotend) but with sturdy motion control... edit: on second thought, I'm not sure how much a high powered E axis gets you, as your limiting factor is probably still the torque you can apply to the filament.
MirageC and Vez3D have been doing some really good quantitative looks into all this, definitely worth a look if you want some answers to those kinds of questions. But yeah, *for the most part* (not including accelerations, movements etc) the main limit now is how fast you can heat up the filament and push it out. Volcano/Supervolcano Hotends, novel Heatbreaks/Nozzles/Coatings for Things (DLC, PCD Nozzles, etc) are helping, but next generation stuff will be even more interesting, the Goliath Hotend, and to a certain degree more Open Source Pellet Extruders (although iirc those inherently have less precision?!
I’ve sold etherCAT motion products in the industrial world, and as a fan of FOSS, it’s great to see it gaining traction and support from the FOSS community. Exciting to see what it will open up when more FOSS software will be able to support the hardware…
A lot of people in the comments are missing the point Marco is making in the intro, he's an expert, he gets it, the state of comms for industrial equipment is an expensive mess of unnecessarily anachronistic technology due to laziness and complacency by the folks that still pay money for it and don't demand better.
I've actually used Leadshine in an industrial setting, and was quite surprised to see them here. They're excellent for cost, compactness, and overall strength. The only downside I could say is that back when I last checked a little over a year ago their website and ordering process wasn't very hobbyist friendly
I was at the Beckhoff production factory and I am amazed at thier technology... been using Beckhoff for my entire career along with other automation technologies. EtherCat is the way for sure...
This is neat, as a software engineer, it feels like the Right Thing for CNC machines. Doing a 3d printer with those space age controllers would be super fun, and a good tutorial.
The tuning is why I like and use Yaskawa or Mitsubishi. They have real time auto tune and you dont hardly have to do anything to use them other than setting the stiffness setting. Once thats done they are ready to go, they figure out the optimal tuning as you use them.
I did a CNC conversion earlier this year using DMM Tech DYN4 servos and drives. They use those same IEEE-1394 connectors for the encoder, which are a pain in the ass to find parts for (the special locking style housing only seems to be available from AliExpress or similar), but I was able to terminate them with parts from Mouser (minus the housing). I think my next CNC build/conversion, will be looking to use EtherCAT or similar. The Mesa cards and support from Mesa themselves is great (I use LinuxCNC), but EtherCAT would be nice.
I've been working with some Nikon encoders that look to have a similar interface just judging by the scope shot at 24:00, they run at 2.5Mbps (switchable up to 4Mbps) and have 16bit data frames with 1 stop bit, no parity. Probably could implement it on a microcontroller that supported 16-bit UART (along with RS422/485 xcvrs), but I was able to do it on an STM32 using DMA to bit bang the query command and read back the data, then post-process it into actual position. Agreed that the resolution and shaft sensitivity is insane!!!
This is why you always design an RJ45 jack into pretty much any product, at least the engineering revs. You can plumb pretty much anything to it, except maybe RF/HV and with the right parts you can do that too. Inexpensive, robust, mature, and basically anything you want a thing to be when you want it. If you are using "scientific" cameras (IDK how else to describe them, see Edmunds Optics or similar vendor site) you can sync and run your camera networks with test/monitoring events (over several bus types, one of which uses PoE over RJ45 on copper that run from 1-10Gbps), which is a bit of learning curve but makes collecting good data pretty plug and play once you have all the bits in your tool chest, and you can incrementally upgrade chunks, which is nice for you and vendors.
Because of your video i contacted Leadshine - always looking for different form factors and at the moment, stock. Leadshine is a Chinese manufacturer with distributors around the world. The negative big for me is that they don't have STO (safe torque off) - or at least they don't have it certified yet. Safety is paramount! - I'm not touching any powerful servo without STO.
As far as hobby use, I think the issue with ethercat is typically the slave devices need a licence fee, this is usually tied to the IC / Asic that implements the Ethercat slave side of the communication, this can make the slave IC's expensive. What I have seen with the duet boards is the use of Can-FD, but it's not as well developed by far and no-where near as advanced as what we're seeing with the above.
Well, the 10-20 bucks per IC/MCU usually doesn't break the bank of most projects so to speak. But it's one of the downsides of ethercat. For hobby use the whole Ethercat eco system including the usual go-to (twinCat) can be rather overwhelming if you don't have experience with PLCs. Also you better try to stay on the full open source side, especially with the master and if you can any further I/O, because getting actual beckhoff stuff will probably cost you an arm and a leg, unless you are a major industrial partner with >>50% discount. You do get a proper RT field bus (thanks to the DC architecture) with it though, so in applications where that matters it's a great system. I also liked what they showed this year with their modular robotics concept that uses EtherCAT, though I have yet to see the prices. But knowing them it will be pretty much impossible to use them in any price concious project, even in a low volume industrial setting.
@@ignispurgatorius5297 There's the Trinamic TMC8462A-BA which is a BGA part It seems to be around only £20 or so it would be interesting if this could be integrated into the Moteus (especially now that ODrive has gone closed source)
I really like this solution. It's just elegant and makes building a cnc like building Legos and being smart, it's going to be much more forgiving on tolerances after some adjusting.
I hadn't seen a video in a while so I dropped by out of curiosity. I'm subbed with all notifications and I've not seen any of your new content in months. Double checked. Just thought I would pass that along... have a great weekend.
MIKROE sells an "EtherCAT" click which implements an LAN9252 ethercat ASIC configurable over SPI. MIKROE also sells a feather click development board which makes for a very easy custom ethercat node. I recommend starting with an adafruit feather, ethercat click, feather click shield, and the "EasyCAT" Arduino library. The library is technically for a deprecated ethercat Arduino shield but since they used the LAN9252 is basically plug and play with the MIKROE ethercat click. I'm using it now to trigger some rs232 commands from ethercat using the rs232 click.
for those who could not be bothered with writing hal files, the beckhoff twincat 3 engineering software package contains cnc stuff where you can just click on the device to assign it to an axis. its free, kinda. Fill in a capcha every 7 days and you are good to go.
@@st3althyone I found it not that painful, IMO. You don't need username/password or open a website and fill some form to renew license. The engineering software automatically detects if your license is expired when you upload the project to the controller, ask you if you want to generate a 7-day trial license, and shows you a pop up to type your captcha in. I think you don't even need internet connection to do so (correct me if I'm wrong).
We need that 3d printer! Though, the limits on hotend thermal performance also need to be addressed to have any chance of being able to make use of such ludicrous speed.
Nice video! (as always, my wife looks at me when I'm laughing in tears behind the PC.. 'it's that funny German guy', so that's your work title around the family here).. Wouldn't these absolute motors also be aware of motion if the machine is powered off due to the UPS in the encoder cable?
Probably EnDat ,it looks like th once yaskawa servos is using, I have been angling to get the data sheet of those encoders with no success, the pinout is not found anywhere.
Its interesting that these things look almost identical to other drives from other Asian servo drives like Mitubishi or Yaskawa all the way down to the connectors and labeling. The CN2 label and connector type is the same too. I am bet that all these motors get made at the same place and get a different label stuck on them and just the controllers are made by different companies. Or maybe its just an industry standards thing idk.
No, it's a chinesium thing. Leadshine or whatever it's called is not top of the line name in the industry, so rather than reinvent the bicycle on their own, they copy successful products and try to do it cheaper. Most companies out there think the same way and that's how you end up with products looking like copies of each other - because they are.
If you look at the different servo drives out there, there are definitely rebranded/licenced drives and sometimes with a slight cosmetic change too. Eg. look at LS and Beijer servo drives - they're identical. I think Yaskawa and Panasonic are the same drives? Omron and another brand that i can't remember right now use the same drives? I've used several brands of servo drives in my time. For me, there are two big differences between them all 1) is the PC software you use to configure the drives - some look and feel like they haven't changed from the 90s, and others are really modern where they attempt to make configuration idiot proof. And 2) the ability to use the digital IO and monitor certain statuses via PDOs/SDOs. These differences can be seen on identical drives but from different manufacturers too. They tend to all use the same/very similar connectors too, but i think this is an industry standard formally/informally. But this type of thing is seen in other parts of the industry, like EtherCAT I/O modules where they all use spring loaded terminals.
That CN2 connector can also be found on sanyo denki motors, one of the more affordable japanese brands imho and that has been around for years. The copying was done probably one or two decades ago in that case, when they all still mostly used analog inputs for their set values. I agree, that is pretty much just the standard now apparently in that corner of the world. Not sure about western manufacturers.
@@aleksandersuur9475 Spring loaded terminals are considered maintenance free. They're supposed to be superior than a screw down terminal block. In the sparky world, the regulations for some nations state that they can put spring loaded terminal blocks in places a regular screw down version cannot.
Worked in flight simulation industries. Used to calibrate this servo drives. Mostly Yaskawa brand for secondary Control loading system. For primary Control loading system there's much more expensive hardwares: MOOG
batteries for encoders are extremely important for robotic applications because you cannot home the robot if it is inside a fixture for example. The battery allows the robot to be controllable even after a power failure without moving the robot. All higher end industrial robots have batteries for the encoders
oooh, that's a pricey system - I have total servo envy! I'd love to upgrade my CNC mill to EtherCAT, but at the ghetto end of the price spectrum (somehow). :)
Do you have any idea on the price range and delivery times of these leadshine servo/drive systems? Are they shipped from China? Is most of their range in stock or built to order? Thanks!
If the manufacturer can't provide a datasheet for those encoders, expect pretty terrible linearity on those encoders. They may spit out 23 bits of counts, but at essentially random increments!
25:27 We can see MCU has builtin EtherCAT controller (so its can be AX58100, AX58400, LAN9255 or XMC4800) but it has external PHY (it rules out ASIX chips). Chip looks like BGA package and IIRC these Leadshine servodrives were available before 2020 - my guestimate is that Infineon XMC4800 is running the show
I use 23 bit servo motors at work. They're far too precise for anything we do. I scale it down to below 10,000 pulses/rev. I'd like to know if there is a real life situation that requires ~8.4 million pulses/rev?
Depends on how precise you want to controll something and if you can use additional gear boxes.Though usually when it comes to anything above 16 bit you have to ask stable that readout is. Looking at the numbers it outputs is cute and all, but plotting them is usually way more interesting. Most systems with these high ADCs have multiple bits that pretty much just show constant noise. At that point you may as well just use a 10/12 bit ADC. Not to mention that most 24 bit ADCs use delta-sigma which in turn isn't as fast usually.
@@ignispurgatorius5297 Do you have an application example that requires a 23 bit encoder? ADCs are notoriously unstable - take a sample and dispose of the first couple of LSBs.
The top people in the speedboat leaderboard already use servos instead of steppers. Yours may be nicer, but it's not all about the motors. Those people are also using custom hotends and cooling solutions. It's up to you if you want to dive down that rabbit hole.
Very, very cool stuff! Way over my head but I like it ;-) Seeing as you're very into motors, drives, and all that, maybe you could help answer something. Is there any sort of servo controller + motor combination which can do both continuous rotation, and absolute positioning? Thinking of a CNC lathe (way off in the future, mind you) and whether it's possible to do both spinny spinny and positioning of the spindle usign just the one drive motor.
Building a super Voron me be a great challenge. I use mine with 48V on x and y and you can go up to 1m/s. I thing the most difficulty thing will be to squeez out enough plastic the same time. But Marc you like Lasers and Servos. Try out to build a super fast lasercutter/engraver.
MarcoReps ✅
AC Servos ✅
Open Source software and Linux repping ✅
Memes ✅
Teasing us with the probability of using those ac servos to make a 3D printer ✅
Thank you man, you're really doing a great job...
Having worked with EtherCat and Beckhoff equipment I can say that it's impressive how their communication protocols are literally identical to the original 80s implementation, just encapsulated into ethernet packets.
CAN works well, I'm glad they didn't try to roll their own version.
And they have done a lot on the RT side to get the peformance as good as they have.
Nothing wrong with that🤷🏻♂
The BACnet and Modbus IP implementations are built similarly; they're thin ethernet wrappers around the original protocols, which while easy to implement and understand for people who spent their whole lives working with BACnet or Modbus, does result in some pretty clunky and redundant protocols with duplicated or unnecessary address fields, checksumming, bad port handling and inconsistent implementation of typical ethernet or higher conventions.
If you wanted a well-designed clean implementation, you could strip out 80% of what industrial protocols like this do because ethernet/IP handles most of that functionality already. They leave it in though, because it's what they're used to.
@@siberx4 Yeah I did some work on it for a communication library and like 90% of the packet was communication overhead already handled by TCP/IP. The remaining 10% was "Send me variable XYZ".
The cool thing about that is that, with an EtherCAT gateway, I could use the same 2020s Python library to communicate with a PLC from the 80s that was designed for a ring network.
Yay excited to see some more spotlight on EtherCat! I got to work with it for a fun robotics project in my lab, driving some high torque and high precision harmonic drive actuators.
About 6 years ago I made an opensource project ( the DieBieSlave ), an ethercat slave that connects a STM32 to a LAN9252 EtherCAT asic, the amount of attention and use that that project got over that time is amazing, from satellite dish uplink plants to exo skeletons ( Project MARCH ). EtherCAT is gaining more and more traction!
More EtherCAT Stuff *AND* it’s *Open Source*?!? Sounds amazing! I’ll definitely need to look into your work.
Is people like you we appreciate
Never thought I’d hear VORON mentioned on one of your videos. I feel like my life is complete now. And yes, you should totally put some of these servos in a VORON and go bananas.
“HevORT” , “RatRig” , and “VZBot” are some other good printer designs along this line of thought to look into as well.
I could dig for links, but i have tried to subscribe to like *all of the channels making that content*, so look around and ask if you need help!
(Most of these Printers have huge Discord/Reddit Presence too)
(Edit: Typo) (Edit 2: Added VZBot)
Oh yes please
Yet another excellent release. You’re outdoing yourself this month, Marco. Highly entertaining, enjoyable, and informative, and I’m here for all of it. Thank you for always giving us your best. It’s highly appreciated. Stay safe, and we’ll see you on the next one.
Yep I second that, I find everything you do videos on so interesting plus Marko's sense of humour is the best! 😅👏
@@TheZombieSaints Indeed!
Considering how little I actually understand of the electronics side it’s crazy how captivating and entertaining your uploads are..
I worked with EtherCAT on device development level using Beckhoff's ET1100 ASIC. It's an interesting protocol as bits are processed and interpreted on the fly as they are coming in, and are retransmitted only like 3 bits later.
Yeah, it's a pity that it isn't used more in the Hobbyist-space. Devices which implement EtherCAT are usually made for industrial use cases, that's at least what Beckhoff targets. That makes them way too expensive for hobby side projects.
@@thevayudev The reason you're seeing them here is because these companies are repurposing copycat precision parts that were mass produced into cheaper packages. I suspect in the next couple of years you're going to see people design home based multipurpose CNC tables that can drop any any tools you want. This isn't the first video I've seen like this, these parts are becoming more obtainable and you could connect them to simple router, plasma cutters, or 3D printer heads and do whatever you want with them. People around me are industrious so there are a few small businesses that are already putting them to work in my neighborhood, which is a very stretch from the traditional machine shops they were in.
@@snarkylive Okay, that does make a lot more sense, that's why none of these are original parts from Beckhoff Automation, those aren't really produced in that high volumes. It does sound like a bright(er) future for the low-budget space if other manufactures can mass produce these components while still being able to communicate with TwinCAT & other EtherCAT stuff. Still waiting for price cuts in the XPlanar/XTS space because those things look like a lot fun to play around with. I mean, couldn't you replace the rails of a CNC machine with XTS rails?
@@thevayudev XPlanar and XTS is definitely a newish thing on the market. Even their competitors (B&R, Bosch Rexroth, and many others) just launched these kinds of products in the last 4-5 years. I doubt we'll see price decrease until the next 5-10 years, especially with current chip issues.
As for XTS replacing CNC machine rails? Well, I think it can be done, but your material weight will be quite limited, especially if you have to stack another XTS on top of the other for X-Y plane.
Fantastic stuff. I've recently built my first CNC controller using those "hybrid" stepper/servos and running LinuxCNC. The repetition for each axis did get a bit tedious and makes the innards look way more complicated than they actually are. This type of quality component is a dream for me. Kudos to the LinuxCNC/EtherCAT interface devs for making this available.
What model of controller/motor did you get? And do you use linuxcnc as well, or something like grbl? I'm currently running open loop steppers with grbl, very basic setup but affordable at 200€ for all electronics. However, I'd like to upgrade that system to something more reliable.
6:27 You explained this at the end of the video without explicitly linking the two. The encoder maintains itself and gives a position when requested. The battery allows the encoder to continue to operate even when the system is de-energised. Therefore, the encoder will continue to monitor the rotation of the shaft and keep the current position in its volatile memory. When the system is energised, the servo drive will ask the encoder for current position.
I remember seeing ThisOldTony using ethercat when he did his last CNC conversion, definitely long overdue for such a technical channel.
Yeah, too bad he didn't convert all axis when he did it. I think he left the Z-axis as is, if memory serves me right.
the rules of the speed boat race competition specifically disallow servo drives, but that doesn't mean that you shouldn't build an absolutely _bonkers_ fast printer with these. Vez3d did one with odrive servos at one point, but he changed back to just high power steppers afterwards. I think he's up to printing in the 1200mm/sec range maximum?
currently at 2000mm/s, not the speedboat though.
Disallow? Jeez, why, as if there are many printers with servo drives out there.
@@VEC7ORlt I think its to encourage stepper driver development and prevent things like dropping thousands of dollars on industrial drives...
@@pmcquay1 This is stupid, should be the other way around - encourage development of servo drives, as steppers are boring and mundane and we need more and readily available servos, besides if you are participating in the speedboat challenge you already spent those thousands, coz your prusa or ender can do shit, its the vorons and the like.
@@VEC7ORlt It would be buy your way to victory with servos, which of course every industrial enterprise does. But for a hobbyist the cost is usually a serious barrier.
If you keep releasing this many videos in quick succession, we're going to start getting use to it. 😁
Yeah, we're becoming spoiled brats.🤣🤣
OMG, this is the most german explanation of etherCAT ever! I love it!
mr. reps so many videos recently love u thank u
I do a lot of security work with industrial control systems (some ISO/IEC 62443 stuff, but mostly technical research/consultancy) and EtherCAT is pretty ubiquitous there, in everything from turnstile gates to cereal manufacturing to ship engines. It's one of the better protocols in my opinion - does its job, does it well, has extremely low latency, and doesn't bog down the base protocol with needless complexity. It's fast and reliable enough to be placed in a safety-critical alarm path, which is something I don't say lightly. In marine environments it's common to find it running alongside UDP-encapsulated NMEA for sensors and navigation comms, which saves a ton of cost and complexity on cabling.
The limiting factor for precision of FDM printers is 100% the extruder and filament. If you think about it, anything involving semi-molten plastic is guaranteed to be way less precise than motion control, even if you're just using belts. Also, at higher extrusion rate the heating necessary goes up substantially.
Still, most speed 3d printers go with as light if an end effector as possible. It would be interesting to see what's possible if you go heavy (big servo E axis, big hotend) but with sturdy motion control...
edit: on second thought, I'm not sure how much a high powered E axis gets you, as your limiting factor is probably still the torque you can apply to the filament.
MirageC and Vez3D have been doing some really good quantitative looks into all this, definitely worth a look if you want some answers to those kinds of questions.
But yeah, *for the most part* (not including accelerations, movements etc) the main limit now is how fast you can heat up the filament and push it out.
Volcano/Supervolcano Hotends, novel Heatbreaks/Nozzles/Coatings for Things (DLC, PCD Nozzles, etc) are helping, but next generation stuff will be even more interesting, the Goliath Hotend, and to a certain degree more Open Source Pellet Extruders (although iirc those inherently have less precision?!
I’ve sold etherCAT motion products in the industrial world, and as a fan of FOSS, it’s great to see it gaining traction and support from the FOSS community. Exciting to see what it will open up when more FOSS software will be able to support the hardware…
CNC for some reason never got the memo that it's 2022, thanks for bringing this to everyone's attention!!!
A lot of people in the comments are missing the point Marco is making in the intro, he's an expert, he gets it, the state of comms for industrial equipment is an expensive mess of unnecessarily anachronistic technology due to laziness and complacency by the folks that still pay money for it and don't demand better.
So many videos recently! Thanks Marco. EtherCAT definitely lands on the "if I ever build my own CNC" list.
I've actually used Leadshine in an industrial setting, and was quite surprised to see them here. They're excellent for cost, compactness, and overall strength. The only downside I could say is that back when I last checked a little over a year ago their website and ordering process wasn't very hobbyist friendly
Most OEMs like that are likely wanting you to go through a VAR. Though that also means higher costs which in turn isn't hobbyist friendly.
I was at the Beckhoff production factory and I am amazed at thier technology... been using Beckhoff for my entire career along with other automation technologies. EtherCat is the way for sure...
This is neat, as a software engineer, it feels like the Right Thing for CNC machines. Doing a 3d printer with those space age controllers would be super fun, and a good tutorial.
The tuning is why I like and use Yaskawa or Mitsubishi. They have real time auto tune and you dont hardly have to do anything to use them other than setting the stiffness setting. Once thats done they are ready to go, they figure out the optimal tuning as you use them.
Being blessed with so many videos recently, feels great!
I know your channel isn’t a cnc machine channel, but yours are by far are my favorite.
I did a CNC conversion earlier this year using DMM Tech DYN4 servos and drives. They use those same IEEE-1394 connectors for the encoder, which are a pain in the ass to find parts for (the special locking style housing only seems to be available from AliExpress or similar), but I was able to terminate them with parts from Mouser (minus the housing). I think my next CNC build/conversion, will be looking to use EtherCAT or similar. The Mesa cards and support from Mesa themselves is great (I use LinuxCNC), but EtherCAT would be nice.
I've been working with some Nikon encoders that look to have a similar interface just judging by the scope shot at 24:00, they run at 2.5Mbps (switchable up to 4Mbps) and have 16bit data frames with 1 stop bit, no parity. Probably could implement it on a microcontroller that supported 16-bit UART (along with RS422/485 xcvrs), but I was able to do it on an STM32 using DMA to bit bang the query command and read back the data, then post-process it into actual position.
Agreed that the resolution and shaft sensitivity is insane!!!
This is why you always design an RJ45 jack into pretty much any product, at least the engineering revs.
You can plumb pretty much anything to it, except maybe RF/HV and with the right parts you can do that too.
Inexpensive, robust, mature, and basically anything you want a thing to be when you want it.
If you are using "scientific" cameras (IDK how else to describe them, see Edmunds Optics or similar vendor site) you can sync and run your camera networks with test/monitoring events (over several bus types, one of which uses PoE over RJ45 on copper that run from 1-10Gbps), which is a bit of learning curve but makes collecting good data pretty plug and play once you have all the bits in your tool chest, and you can incrementally upgrade chunks, which is nice for you and vendors.
Finally! A new LinuxCNC video!
Because of your video i contacted Leadshine - always looking for different form factors and at the moment, stock. Leadshine is a Chinese manufacturer with distributors around the world.
The negative big for me is that they don't have STO (safe torque off) - or at least they don't have it certified yet.
Safety is paramount! - I'm not touching any powerful servo without STO.
As far as hobby use, I think the issue with ethercat is typically the slave devices need a licence fee, this is usually tied to the IC / Asic that implements the Ethercat slave side of the communication, this can make the slave IC's expensive.
What I have seen with the duet boards is the use of Can-FD, but it's not as well developed by far and no-where near as advanced as what we're seeing with the above.
Well, the 10-20 bucks per IC/MCU usually doesn't break the bank of most projects so to speak. But it's one of the downsides of ethercat. For hobby use the whole Ethercat eco system including the usual go-to (twinCat) can be rather overwhelming if you don't have experience with PLCs. Also you better try to stay on the full open source side, especially with the master and if you can any further I/O, because getting actual beckhoff stuff will probably cost you an arm and a leg, unless you are a major industrial partner with >>50% discount. You do get a proper RT field bus (thanks to the DC architecture) with it though, so in applications where that matters it's a great system. I also liked what they showed this year with their modular robotics concept that uses EtherCAT, though I have yet to see the prices. But knowing them it will be pretty much impossible to use them in any price concious project, even in a low volume industrial setting.
@@ignispurgatorius5297 There's the Trinamic TMC8462A-BA which is a BGA part
It seems to be around only £20 or so it would be interesting if this could be integrated into the Moteus
(especially now that ODrive has gone closed source)
I really like this solution. It's just elegant and makes building a cnc like building Legos and being smart, it's going to be much more forgiving on tolerances after some adjusting.
And we want to thank you for entertaining us. You are awesome.
I wrote firmware and software for custom EtherCAT modules at my last employer. It was a pretty interesting protocol to work with.
Ta Ta in Kannada, wow that got me out of no where, man you are a goat of electronics😀
This was a lot of fun to follow. Thanks for this video!
Can’t wait to see the project that you will make with these!
I hadn't seen a video in a while so I dropped by out of curiosity. I'm subbed with all notifications and I've not seen any of your new content in months. Double checked. Just thought I would pass that along... have a great weekend.
MIKROE sells an "EtherCAT" click which implements an LAN9252 ethercat ASIC configurable over SPI. MIKROE also sells a feather click development board which makes for a very easy custom ethercat node. I recommend starting with an adafruit feather, ethercat click, feather click shield, and the "EasyCAT" Arduino library. The library is technically for a deprecated ethercat Arduino shield but since they used the LAN9252 is basically plug and play with the MIKROE ethercat click. I'm using it now to trigger some rs232 commands from ethercat using the rs232 click.
Ethercat is an excellent system. Once you use it, you keep it and don't look back
Can’t wait to see what you are planning to do with these 😮
Can't wait for one of the fastest benchy print on the internet
for those who could not be bothered with writing hal files, the beckhoff twincat 3 engineering software package contains cnc stuff where you can just click on the device to assign it to an axis. its free, kinda. Fill in a capcha every 7 days and you are good to go.
That sounds like an attainable pain in the ass. 😆😆
@@st3althyone I found it not that painful, IMO. You don't need username/password or open a website and fill some form to renew license. The engineering software automatically detects if your license is expired when you upload the project to the controller, ask you if you want to generate a 7-day trial license, and shows you a pop up to type your captcha in. I think you don't even need internet connection to do so (correct me if I'm wrong).
@@FlyingPlastic356 Cool, that doesn’t sound too bad. Thanks.
The upload frequency is making my spidey senses tingle
No nut November indeed. The occasional humor in your videos in between the extra-geeky stuff (which is why I'm here) is great. Love your vids
I love the dry delivery of reps' humor. At 24:18: "...wow check out the sensitivity of the shaft! Hey, it's no-nut-November after all".
I would love to see an over engineered 3d printer poweder by this setup! The Benchy Test has to be done :D
Nice to see fieldbus stuff on here
Thanks for the subtle humor. I love it :D
I made an Ethercat master almost 9 years ago in C++. it was fun but hard :)
congrats on 200k at last marco!
We need that 3d printer! Though, the limits on hotend thermal performance also need to be addressed to have any chance of being able to make use of such ludicrous speed.
Fast Benchy please. There is a lot more to that "sport" than kinematics so I think it would still be an interesting challenge.
I'd love to see those smaller servos put to work in an auto tool-changer
I really had to laugh when you showed off your old setup, and remember the oil pump EMI issue X)
Nice video! (as always, my wife looks at me when I'm laughing in tears behind the PC.. 'it's that funny German guy', so that's your work title around the family here).. Wouldn't these absolute motors also be aware of motion if the machine is powered off due to the UPS in the encoder cable?
Congrats on the 200k!
Beckhoff und EtherCAT on Open Source Level. I like it
The encoder probably uses either the SSI, BiSS, Hiperface, or EnDat interface
Probably EnDat ,it looks like th once yaskawa servos is using, I have been angling to get the data sheet of those encoders with no success, the pinout is not found anywhere.
Its interesting that these things look almost identical to other drives from other Asian servo drives like Mitubishi or Yaskawa all the way down to the connectors and labeling. The CN2 label and connector type is the same too. I am bet that all these motors get made at the same place and get a different label stuck on them and just the controllers are made by different companies. Or maybe its just an industry standards thing idk.
No, it's a chinesium thing. Leadshine or whatever it's called is not top of the line name in the industry, so rather than reinvent the bicycle on their own, they copy successful products and try to do it cheaper. Most companies out there think the same way and that's how you end up with products looking like copies of each other - because they are.
If you look at the different servo drives out there, there are definitely rebranded/licenced drives and sometimes with a slight cosmetic change too. Eg. look at LS and Beijer servo drives - they're identical. I think Yaskawa and Panasonic are the same drives? Omron and another brand that i can't remember right now use the same drives?
I've used several brands of servo drives in my time. For me, there are two big differences between them all 1) is the PC software you use to configure the drives - some look and feel like they haven't changed from the 90s, and others are really modern where they attempt to make configuration idiot proof. And 2) the ability to use the digital IO and monitor certain statuses via PDOs/SDOs. These differences can be seen on identical drives but from different manufacturers too.
They tend to all use the same/very similar connectors too, but i think this is an industry standard formally/informally. But this type of thing is seen in other parts of the industry, like EtherCAT I/O modules where they all use spring loaded terminals.
@@Shocker99 Spring loaded terminals on EtherCAT devices is a pretty funny observation, but likely spot on.
That CN2 connector can also be found on sanyo denki motors, one of the more affordable japanese brands imho and that has been around for years. The copying was done probably one or two decades ago in that case, when they all still mostly used analog inputs for their set values. I agree, that is pretty much just the standard now apparently in that corner of the world. Not sure about western manufacturers.
@@aleksandersuur9475 Spring loaded terminals are considered maintenance free. They're supposed to be superior than a screw down terminal block.
In the sparky world, the regulations for some nations state that they can put spring loaded terminal blocks in places a regular screw down version cannot.
To discover what processor that is, maybe you can use your thermal camera to image the heat pattern, and compare that to some other brands?
Worked in flight simulation industries. Used to calibrate this servo drives. Mostly Yaskawa brand for secondary Control loading system. For primary Control loading system there's much more expensive hardwares: MOOG
Absolute Legende!
batteries for encoders are extremely important for robotic applications because you cannot home the robot if it is inside a fixture for example. The battery allows the robot to be controllable even after a power failure without moving the robot. All higher end industrial robots have batteries for the encoders
oooh, that's a pricey system - I have total servo envy! I'd love to upgrade my CNC mill to EtherCAT, but at the ghetto end of the price spectrum (somehow). :)
Since 2019 until now, have you faced any issues with your current setup in your machine?
This video Is a topic I am excited about!
Not gonna lie Marco, that "swooshing" made me clinch my buttocks. lol
Was not expecting that no nut november joke, I almost fell out of my chair laughing
24:25 I bet the Reps scale is logarithmic as well
time to build the nastiest most precise voron ever built
Do you have any idea on the price range and delivery times of these leadshine servo/drive systems? Are they shipped from China? Is most of their range in stock or built to order? Thanks!
If the manufacturer can't provide a datasheet for those encoders, expect pretty terrible linearity on those encoders. They may spit out 23 bits of counts, but at essentially random increments!
25:27 We can see MCU has builtin EtherCAT controller (so its can be AX58100, AX58400, LAN9255 or XMC4800) but it has external PHY (it rules out ASIX chips). Chip looks like BGA package and IIRC these Leadshine servodrives were available before 2020 - my guestimate is that Infineon XMC4800 is running the show
This is absolutel AWESOME!
Marco, Use those motors for Solar Panel optimizing collection the way things are going your going to need that extra watt or 2. Just say'n.
Oh what should I do with these insanely cool new motors ? The possibilities! Reality - they will collect dust on a shelf :D :D
If it wasn't because of backlash it could be geared and be extremely scary. But also awesomely powerful xD
now where do you actually buy these leadshine servo drives
I use 23 bit servo motors at work. They're far too precise for anything we do. I scale it down to below 10,000 pulses/rev.
I'd like to know if there is a real life situation that requires ~8.4 million pulses/rev?
Depends on how precise you want to controll something and if you can use additional gear boxes.Though usually when it comes to anything above 16 bit you have to ask stable that readout is. Looking at the numbers it outputs is cute and all, but plotting them is usually way more interesting. Most systems with these high ADCs have multiple bits that pretty much just show constant noise. At that point you may as well just use a 10/12 bit ADC. Not to mention that most 24 bit ADCs use delta-sigma which in turn isn't as fast usually.
@@ignispurgatorius5297 Do you have an application example that requires a 23 bit encoder?
ADCs are notoriously unstable - take a sample and dispose of the first couple of LSBs.
Best Monday ever, i show up to two Marco videos!
Perfect for a wire EDM machine 😉
Can't wait to stick some on my ender 3!
So much videos. its like Christmas!!!
Ohh I see, your homelab cloning setup finally yields, so you can put out more videos!
The top people in the speedboat leaderboard already use servos instead of steppers. Yours may be nicer, but it's not all about the motors. Those people are also using custom hotends and cooling solutions. It's up to you if you want to dive down that rabbit hole.
So how do those leadshine components compare to genuine Omron devices?
Twincat 3, thats what i used for bechoff gear when i worked with industrial automation...
Very, very cool stuff! Way over my head but I like it ;-) Seeing as you're very into motors, drives, and all that, maybe you could help answer something. Is there any sort of servo controller + motor combination which can do both continuous rotation, and absolute positioning? Thinking of a CNC lathe (way off in the future, mind you) and whether it's possible to do both spinny spinny and positioning of the spindle usign just the one drive motor.
What a great ad this video is
No Nut November. I see what you did there ;) You are a funny guy! Keep up the great work.
Thanks, that was cool!
At first I was like oh, another Chinese free stuff video, but then... PPMs on a servo???? Count me in!!
22:27 I'd say the first thing you should do with these is playing a little bit of "Thunderstruck" a'la Floppotron 😅
The purpose of the battery is so the encoder can keep its value correct even if the whole system is powered off and someone moves the motor.
Building a super Voron me be a great challenge. I use mine with 48V on x and y and you can go up to 1m/s. I thing the most difficulty thing will be to squeez out enough plastic the same time. But Marc you like Lasers and Servos. Try out to build a super fast lasercutter/engraver.
The First time seeing an Beckhoff Pc running someting other then a real time windows Operating System.
I've never clicked so fast
Stepstepstepstepstep
Loved it