Some issues might be with 90 Ohm impedance of USB cable (should be 120 Ohm, meaning you will have reflections), and maybe adding ferrite bead to 24V and GND might also help. (I guess capacitors should've already helped to some extend). Btw. I sent you a Jig to cut square metal rods on discord, I extended them to be ~30.5mm long
have you seen the BTT canbus splitter board, it would be a cleaner solution, the manta boards do have their own can bus ports, no need to have a separate USB to can adapter too
Its almost a great solution. It would be perfect if they just gave you 7 power and 7 signal screw terminals. As it is, it takes two to have a clean setup or your gonna have to have different connectors for different heads.
I don’t use Can Bus on my Hydra tool changer. However I need to test it at some point. I really can’t have the wire coming out the back as my printer are all the way to back to the wall. I also just made new clips to hold the back panel that only are 1 mm in thickness my panels from Biltema are 400x600x4 mm so now only about 5mm is needed so I can max out my Y travel. I really whole like to see a 3d printer board that I can connect 4 Can Bus tools on board and don’t need any distribution board. Even if i only have 4 tools my wiring is easy to do, As they all connect directly to the main board. I use Marlin and BIGTREETECH BTT Octopus Max EZ. I need to do some testing with Klipper and RepRap
Is something like the ldo nighthawk usb toolhead board helpful here? I feel like can bus is uniquely useful for multiple toolheads thought (as you have it) As much pain as canbus debugging is.
Exactly - the propper way is to have everything in one chain - I see that people seem to get away with this, but signal quality will be degraded over a correct bus layout with two terminated ends and the more of those branches there are the worse it will be. I’m not set on what I will do with my printers (I am building two with tapchanger) - either run two set of bus wire to the head to allow the head to be in the bus middle or use multiple CAN buses with the host controller in the middle and just two heads at each end. Another thing, about the MKS CAN bus controller: it is isolated, so this one has no connection between bus ground and USB ground and without a common ground the bus is even more degraded.
Exactly - the propper way is to have everything in one chain - I see that people seem to get away with this, but signal quality will be degraded over a correct bus layout with two terminated ends and the more of those branches there are the worse it will be. I’m not set on what I will do with my printers (I am building two with tapchanger) - either run two set of bus wire to the head to allow the head to be in the bus middle or use multiple CAN buses with the host controller in the middle and just two heads at each end. Another thing, about the MKS CAN bus controller: it is isolated, so this one has no connection between bus ground and USB ground and without a common ground the bus is even more degraded. Do you have a product number for that Igus cable by the way?
@@tructruc00 Right, it is a very common mistake. The signal quality isn’t as good as it should be and the more of such violations you do the worse it is. This is plain physics - at a branch you will get a reflection, the head cables are unterminated, so there will be reflections too. All branchings happen at nearly the same location, so all those branch reflections happen at the same time, the head cables are more or less the same length, so those reflections come back to the branch point at nearly the same time. If you want to branch without reflection you would have have to use a different impedance cable to the head, but the you would have problems with the signals coming from the head to the branch point. You can’t overcome physics. While the signal might still be good enough with a single branch, adding 6 branches is really asking for troubles - that’s 5/6th of the signal energy being reflected at the branch point and only 1/6th being passed, of which only 1/(6*6)th goes to each head. Because the cables are rather short, the reflection times are short too, so time is a bit on your side and chances are that things equal out fast enough after a few reflection iterations. That’s why a few mm branch into a chip on a PCB is usually fine. You normally don’t see the signal quality, since it is digital - it either works or it doesn’t, but you won’t know how marginal it is unless you go and scope an eye pattern signal. If you play by the rules you are safe - if you violate them you should know what you are doing.
Could you provide an amazon link for the USB cable you are using? There is a big variety in cable quality, so some sort of list of tested ones would be very helpful. Thanks!
@@GrizzLeeAdams The way the Klipper canbus on MCU works is that all commands for the Manta go though the canbus protocol as well. Meaning it needs to drive the 6 motors in addition to toolheads. This caused homing timeouts for me a year ago. Homing is def the most demanding operation for pushing a lot of messages in a short bursts. I suspect they just got backed up in the transmission queue. The MCU serial interface is clocked at 250kb/s so theoretically a 500kb/s canbus should beat it. I suspect the frame overhead and frame pacing limitations are playing a significant role.
Nu ļoti grandioza parikte.
Bet sarežģīta gan. Bet patika...
Some issues might be with 90 Ohm impedance of USB cable (should be 120 Ohm, meaning you will have reflections), and maybe adding ferrite bead to 24V and GND might also help. (I guess capacitors should've already helped to some extend).
Btw. I sent you a Jig to cut square metal rods on discord, I extended them to be ~30.5mm long
have you seen the BTT canbus splitter board, it would be a cleaner solution, the manta boards do have their own can bus ports, no need to have a separate USB to can adapter too
It supports only up to 5 toolheads for what I know
Bigtreetech CEB is the name
Its almost a great solution. It would be perfect if they just gave you 7 power and 7 signal screw terminals. As it is, it takes two to have a clean setup or your gonna have to have different connectors for different heads.
I don’t use Can Bus on my Hydra tool changer. However I need to test it at some point. I really can’t have the wire coming out the back as my printer are all the way to back to the wall. I also just made new clips to hold the back panel that only are 1 mm in thickness my panels from Biltema are 400x600x4 mm so now only about 5mm is needed so I can max out my Y travel. I really whole like to see a 3d printer board that I can connect 4 Can Bus tools on board and don’t need any distribution board. Even if i only have 4 tools my wiring is easy to do, As they all connect directly to the main board. I use Marlin and BIGTREETECH BTT Octopus Max EZ. I need to do some testing with Klipper and RepRap
Why is there a resistor on the distribution board? Doesn't the toolboard already have one built in?
Is something like the ldo nighthawk usb toolhead board helpful here?
I feel like can bus is uniquely useful for multiple toolheads thought (as you have it) As much pain as canbus debugging is.
So is the distribution board just a splitter for canbus?
did you use TMC5160T Pro for all your drivers?
so I bought a 100W USB-C cable that has 8 wires in it. so I guess I use two pairs for each phase or CAN signal then?
You need to daisy chain it. spurs will be flakey. Easy with igus cable, 2 twisted pair and 2 power in one cable
Exactly - the propper way is to have everything in one chain - I see that people seem to get away with this, but signal quality will be degraded over a correct bus layout with two terminated ends and the more of those branches there are the worse it will be.
I’m not set on what I will do with my printers (I am building two with tapchanger) - either run two set of bus wire to the head to allow the head to be in the bus middle or use multiple CAN buses with the host controller in the middle and just two heads at each end.
Another thing, about the MKS CAN bus controller: it is isolated, so this one has no connection between bus ground and USB ground and without a common ground the bus is even more degraded.
Exactly - the propper way is to have everything in one chain - I see that people seem to get away with this, but signal quality will be degraded over a correct bus layout with two terminated ends and the more of those branches there are the worse it will be.
I’m not set on what I will do with my printers (I am building two with tapchanger) - either run two set of bus wire to the head to allow the head to be in the bus middle or use multiple CAN buses with the host controller in the middle and just two heads at each end.
Another thing, about the MKS CAN bus controller: it is isolated, so this one has no connection between bus ground and USB ground and without a common ground the bus is even more degraded.
Do you have a product number for that Igus cable by the way?
He is just using a star topology which is really common
@@tructruc00 Right, it is a very common mistake. The signal quality isn’t as good as it should be and the more of such violations you do the worse it is.
This is plain physics - at a branch you will get a reflection, the head cables are unterminated, so there will be reflections too. All branchings happen at nearly the same location, so all those branch reflections happen at the same time, the head cables are more or less the same length, so those reflections come back to the branch point at nearly the same time. If you want to branch without reflection you would have have to use a different impedance cable to the head, but the you would have problems with the signals coming from the head to the branch point.
You can’t overcome physics.
While the signal might still be good enough with a single branch, adding 6 branches is really asking for troubles - that’s 5/6th of the signal energy being reflected at the branch point and only 1/6th being passed, of which only 1/(6*6)th goes to each head. Because the cables are rather short, the reflection times are short too, so time is a bit on your side and chances are that things equal out fast enough after a few reflection iterations. That’s why a few mm branch into a chip on a PCB is usually fine.
You normally don’t see the signal quality, since it is digital - it either works or it doesn’t, but you won’t know how marginal it is unless you go and scope an eye pattern signal.
If you play by the rules you are safe - if you violate them you should know what you are doing.
@@tructruc00 Which is the wrong way to do CAN bus unless you are going to have repeater at hub and separate networks.
Could you provide an amazon link for the USB cable you are using? There is a big variety in cable quality, so some sort of list of tested ones would be very helpful. Thanks!
would the BIGTREETECH CEB work inplace of the custom board you have there?
Mellow Has a canbus splitter board
Why not use the onboard canbus of the manta?
@@GrizzLeeAdams The way the Klipper canbus on MCU works is that all commands for the Manta go though the canbus protocol as well.
Meaning it needs to drive the 6 motors in addition to toolheads. This caused homing timeouts for me a year ago.
Homing is def the most demanding operation for pushing a lot of messages in a short bursts. I suspect they just got backed up in the transmission queue.
The MCU serial interface is clocked at 250kb/s so theoretically a 500kb/s canbus should beat it. I suspect the frame overhead and frame pacing limitations are playing a significant role.
great stuff, but do something about the video orientation.
what is this rats nest?
Nice Video! where did you get the connectors shown in ua-cam.com/video/QkXT-72ibj8/v-deo.html here? Tried to find them both
"Better safe than sorry..." then proceeds to show some really unsafe wiring.
Oh hush you. 😁