Power source for a network of powerful MX motors
Power source for a network of powerful MX motors
by petercohen
Tue Nov 20, 2012 3:46 pm
Hello, what is the best way to power a network of powerful MX motors such as the MX-108? As they require lots of current (over 6 amp?), do we have to provide a power source for each individual motors? Can they all share from a single power source? Any user experience and recommendation for power supply appreciated. Thanks.
by MarcoP
Wed Nov 21, 2012 11:52 am
Hi
We have the same exact problem for the humanoid robot we are designing (http://robosavvy.com/forum/viewforum.php?f=26)
You can power everything from one power source.
The problem with the daisy chain setup in the servo cables is that the current for all the servos has to pass trough the initial servo. This means you will have increased heat (maybe even melting of wires depending on current) and voltage drop. This will cause the servos toward the end to loose some torque.
Our planned solution is to use connection schema seen in the image:
We are going to use thicker copper cables (top of image) for power distribution. The ones we are using are rated for 100A. We will then cut the 2 power wires in the servo cable and solder them to the main cable. That way, the current to power each servo is split between two paths into the servo, which reduces losses and heating.
A single lipo is used to power everything. We also use some electronics to enable switching between batteries or a dc power supply as seen here: http://robosavvy.com/forum/viewtopic.php?t=8902
The current for all the servos is still shared in the bus wire, but since it's a much thicker wire that is no longer a concern.
A setup were you would run one pair of wires from the power supply to each servo is also possible, but the advantages would be minimal.
Regards
Marco
We have the same exact problem for the humanoid robot we are designing (http://robosavvy.com/forum/viewforum.php?f=26)
You can power everything from one power source.
The problem with the daisy chain setup in the servo cables is that the current for all the servos has to pass trough the initial servo. This means you will have increased heat (maybe even melting of wires depending on current) and voltage drop. This will cause the servos toward the end to loose some torque.
Our planned solution is to use connection schema seen in the image:
We are going to use thicker copper cables (top of image) for power distribution. The ones we are using are rated for 100A. We will then cut the 2 power wires in the servo cable and solder them to the main cable. That way, the current to power each servo is split between two paths into the servo, which reduces losses and heating.
A single lipo is used to power everything. We also use some electronics to enable switching between batteries or a dc power supply as seen here: http://robosavvy.com/forum/viewtopic.php?t=8902
The current for all the servos is still shared in the bus wire, but since it's a much thicker wire that is no longer a concern.
A setup were you would run one pair of wires from the power supply to each servo is also possible, but the advantages would be minimal.
Regards
Marco
by PaulL
Wed Nov 21, 2012 3:59 pm
A bigger pair of power wires is a lot better than "home run" power wiring (separate power back to source from each servo). The problem one will find is that the bulk of the insulation alone is a problem. Fatter wires, high flexibility (high strand count), silicone jacket, branching off for each servo, that's the way to go. As always, length of run and gauge of wire is the key to preventing heating / power loss / voltage drop.
by petercohen
Wed Nov 21, 2012 4:34 pm
RN1AsOf091407 wrote:A bigger pair of power wires is a lot better than "home run" power wiring (separate power back to source from each servo). The problem one will find is that the bulk of the insulation alone is a problem. Fatter wires, high flexibility (high strand count), silicone jacket, branching off for each servo, that's the way to go. As always, length of run and gauge of wire is the key to preventing heating / power loss / voltage drop.
Thanks for the suggestion. Could you please provide more detailed specifications on how fat, what values of strand count and gauge of wire would be ideal? Am I correct that the wires entering and leaving each dynamixel can be those thin 3P wires from Robotis?
by PaulL
Wed Nov 21, 2012 5:01 pm
... It depends on how far and how many servos you're going to run off a given tap...
If we're talking about a Biped, you need four power runs - arms, legs. The shorter the run, the smaller the gauge you can get away with. For my RN-1, my plan is to use 20AWG, but then I'm using servos with much less current draw.
You'll likely be somewhere between 16 AWG and 10 AWG depending on the size of your bot and servo count (amp max for each power run).
The kind of high flex wire used in RC hobbies should be just fine as far as strand count goes.
Btw, soldering these fat wires isn't easy. It takes a LOT of heat. I have used a Weller soldering gun at 140 watts on 12 AWG wire. For taps like this, you'll want to skin the wire by using a hobby razor knife to slit around the wire in two places, then cut across the middle section to free the silicone. Bear in mind, once you start feeding solder into fatter stranded wire, it will try to flow along the wire, making it stiffer under the insulation for some distance from where you apply the solder. At this point, it will be a bit of trial and error. It would be good to experiment before you get too far into it.
As for the Robotis servos, I would guess that they sized the wire appropriately at the servos...
If we're talking about a Biped, you need four power runs - arms, legs. The shorter the run, the smaller the gauge you can get away with. For my RN-1, my plan is to use 20AWG, but then I'm using servos with much less current draw.
You'll likely be somewhere between 16 AWG and 10 AWG depending on the size of your bot and servo count (amp max for each power run).
The kind of high flex wire used in RC hobbies should be just fine as far as strand count goes.
Btw, soldering these fat wires isn't easy. It takes a LOT of heat. I have used a Weller soldering gun at 140 watts on 12 AWG wire. For taps like this, you'll want to skin the wire by using a hobby razor knife to slit around the wire in two places, then cut across the middle section to free the silicone. Bear in mind, once you start feeding solder into fatter stranded wire, it will try to flow along the wire, making it stiffer under the insulation for some distance from where you apply the solder. At this point, it will be a bit of trial and error. It would be good to experiment before you get too far into it.
As for the Robotis servos, I would guess that they sized the wire appropriately at the servos...
by PaulL
Wed Nov 21, 2012 5:41 pm
A quick comment...
As I'm using PWM-controlled servos, each servo needs a signal back to the controller. For that, I got this:
http://www.litchfieldstation.com/xcart/product.php?productid=410120&cat=171&page=1
The servo signal wires are always the same gauge on the 3 wire servo connections, but it doesn't need to be. It's just a PWM signal, and doesn't need to be so big (and inflexible!).
Btw, if anyone doesn't know, DCC in Model Railroading stands for Digital Command Control, a means whereby signals AND power are sent through the same pair of wires. I've wondered for some time if that method, along with the tiny controllers they use, could be modified for our needs... I'm also into model railroading.. There is also "transponding", sending signals back through those same pairs of wires - Digitrax does some neat work there, I have a 4 zone controller for transponding, pretty nifty...
As I'm using PWM-controlled servos, each servo needs a signal back to the controller. For that, I got this:
http://www.litchfieldstation.com/xcart/product.php?productid=410120&cat=171&page=1
The servo signal wires are always the same gauge on the 3 wire servo connections, but it doesn't need to be. It's just a PWM signal, and doesn't need to be so big (and inflexible!).
Btw, if anyone doesn't know, DCC in Model Railroading stands for Digital Command Control, a means whereby signals AND power are sent through the same pair of wires. I've wondered for some time if that method, along with the tiny controllers they use, could be modified for our needs... I'm also into model railroading..
There is also "transponding", sending signals back through those same pairs of wires - Digitrax does some neat work there, I have a 4 zone controller for transponding, pretty nifty...
by petercohen
Thu Nov 22, 2012 2:07 pm
Thanks Marco for the suggestion and nice photo. Why there are four cables entering and leaving the dynamixel? For my MX motor, there is a 3P socket on each side. So, am I right that except for the two thick ground and power cables, the rest of the cables can be the stock 3P cables from Robotis? In other words, I just cut the power and ground of each 3P cable (while keeping the data cable intact). Then, solder them to the corresponding thick cable?
by MarcoP
Thu Nov 22, 2012 4:36 pm
Hi
It's 4 wires because i usually use the version with two 2 data lines (RS485).
What you said seems correct. But the 3 wires cables we have here have the same thickness for all 3 wires.
You should check the pin out on the servo just to make sure.
I would also suggest that when you use it for the first time use just one servo just in case.
Rgds
It's 4 wires because i usually use the version with two 2 data lines (RS485).
What you said seems correct. But the 3 wires cables we have here have the same thickness for all 3 wires.
You should check the pin out on the servo just to make sure.
I would also suggest that when you use it for the first time use just one servo just in case.
Rgds