LinkBack URL
About LinkBacks
Why not use the toshiba tb6560 instead? It cost about the same and does not need a micro-controller.
| | |
I'm working on my first cnc machine- it's going to be a fairly simple 3 axis, dremel based, drawer glide and threaded rod drive contraption... and I'm feeling good about the hardware as there are a so many great hardware writeups out there... The software side however, I'm finding a lot less info about. So I'm going to post what I THINK I've figured out... and hope someone smarter than me can tell me what I've got wrong.
From what I see in Mach3, it appears that side of things is reasonably simple - it's going to control the parallel port... specifically 6 pins of it - one direction and one pulse pin for each axis. I'll tell Mach3 how many degrees happen per step, and what linear distance is moved by each revolution...
Then I'm going to feed those six pins into an arduino (I'll be using the analog pins, but configuring them as digital inputs, and eventually I'll remove the arduino and just go to a atmega328p with bare minimum support components... and really nice filter caps...)
The arduino is going to use 12 of the digital output pins to drive 3 L293D driver chips. Each time a pulse is detected on one of the pulse inputs, the arduino will check the state of the direction pin, and either step forward or backward in the stepper position lookup table accordingly.
The 293d chips will be hooked directly to 3 small bipolar steppers, which will be hooked direct drive to the screw drives on the machine.
Will all of that work to interface Mach3 to the steppers? Is there an easier way? Does a plugin or driver exist (or could it exist?) to allow me to skip the arduino and use the parallel port directly to control the 12 pins that feed into the L293D drivers?
And one sort of related question... Mouser carries the L293D for 4 bucksish, and the l293DNE for closer to 2... Is the NE part acceptable, or is there a reason to spend the extra few dollars for the plain D model? I read the datasheet and couldn't figure out what the difference was between the D and the DNE...
Thanks in advance for answering some newbish questions!
Art
Why not use the toshiba tb6560 instead? It cost about the same and does not need a micro-controller.
Because the Toshiba tb6560 has caused a lot of people a lot of problems. Search CNCZone.com for "How I fixed my TB6560" or just "TBG6560" and look at the "Wailing Wall" of blown drivers.
If you like Arduino's and are setup to work with one, then I think you are on track. You can build your own circuit or buy the motor shield using the same chips ready made at
ladyada.net/make/mshield/ and either way just use their Arduino library.
If you want to avoid the Arduino, there are a LOT of open source stepper motor drivers designed to work directly with Mach3 via the parallel port. Basically, anything that accepts step and direction inputs should be fine. For a list of open source options see:
pminmo.com
Having said that, I wonder why you are settled on bipolar motors? bipolar isn't always better, especially for CNC mills where fast rapids are more useful that high starting torque. Take a look at: techref.massmind.org/techref/io/stepper/connections.htm
and avoid 4 wire motors as they do /nothing/ but limit your options.
I'm also not sure why you are tied to the L293D's. There are a LOT of good drivers out there, especially for unipolar drive.
In any case, don't get caught in the trap of buying the motors/drivers first and /then/ figuring out if they will actually run your mill. Estimate or measure the load your mill presents and /then/ figure out what motors / drivers are needed to run that load at the speed you want. See: techref.massmind.org/techref/io/steppers.htm#Estimating
for help with that.
James hosts the single best wiki page about steppers for CNC hobbyists on the net:
http://www.piclist.com/techref/io/steppers.htm Disagree? Tell him what's missing! ,o)
There are many who got the 6560 working also. I suspect that the volume of complaints is simply due to the sheer number of people who bought the relatively poorly designed Ebay drives.
I'm not trying to pick on you, but the link you provided made questionable conclusions. They compared unipolar to bipolar series. Obviously, the latter would suck at high speeds because it has 4 times the inductance. A bipolar parallel connection would result in the highest performance.
Hey H500, no worries I don't mind being picked on if it brings better understanding to light. But I've checked that link again and it compares unipolar, bipolar series, AND bipolar parallel. The specific motor compared is serial vs unipolar, I agree, but that is because parallel is not an option with most motors; it can only be supported by a few 4 wire and 8 wire motors. Yes, parallel would do better than series, but still can't reach the same speeds that a unipolar drive can reach, given the same motor, drive current, and supply power. The video linked just under that comparison is another, better, example.
I'll grant that if you have an 8 wire motor, and if you provide the almost half again extra current from the driver and supply, then the same motor will do better under parallel than unipolar, but that is with a bigger power supply and a driver that can manage the extra current. And those things cost. And if you took that extra cost and used it to buy a bigger motor and ran it unipolar, that advantage would disappear.
If you compare apples to apples, total system cost to total system cost, because unipolar drivers are /always/ going to cost less for the same drive power, (half as many transistors) then there is no advantage to bipolar parallel drive.
Certainly, there is no advantage to bipolar serial drive, which is what all the motors under 8 wires are limited to providing.
I really do think people are hung up on bipolar drives; and they just are not inherently better.
James hosts the single best wiki page about steppers for CNC hobbyists on the net:
http://www.piclist.com/techref/io/steppers.htm Disagree? Tell him what's missing! ,o)
I have not come across anything, practical or theoretical that supports the belief that unipolars can reach higher speeds than bipolars. If you look at typical datasheets, bipolar half coil, parallel and unipolar have the same inductance. In theory, they should all reach the same max speed.
At other speeds, the lower coil resistance of bipolar parallel motors let you pump more current through without heating the motor up. So it will give you up to 40% more torque if the rest of your system is good for it.
My recommendation would be to buy 8 wire motors since they support all configurations and doesn't cost any more.
Totally agree on 8 wire motors. However, they are generally more expensive... not sure why that is, other than the extra cost of the wire.
Totally agree that the inductance is the same. What you are missing is that in bipolar drive, the current must start, reach peak flow in one direction, then slow, stop, and then start in the opposite direction, reach peak flow going backwards, then slow and stop to complete each cycle. In unipolar drive, the current just goes in one direction and back to stop to complete a cycle. So the same inductance is encountered /twice/ for a bipolar drive and only once for a unipolar drive. That disadvantage is /partially/ offset by the greater power of the larger peak to peak cycle, but that offset disappears at higher speeds and a given motor in unipolar mode WILL run faster than in bipolar parallel at the same drive power. If you don't believe me, or the video PMinMO posted, try it yourself.
James hosts the single best wiki page about steppers for CNC hobbyists on the net:
http://www.piclist.com/techref/io/steppers.htm Disagree? Tell him what's missing! ,o)
Yes and no. Lower price bipolar motors are typically 4wire and are bipolar series wound, so they have 4 times the inductance and are slower than the same package in unipolar. That's a big difference at the lower end of the market.Originally Posted by H500
With the larger more expensive motors that can be sourced in 8wire versions, they can be wired as bipolar parallel but often this is not done as it requires a higher operating current that many drives won't do (it usually pushes the current from the 3A/3.5A of most cheaper drivers to >4A etc where you can no longer use cheap drivers and need high end industrial drivers). Mariss from gecko has stated before he prefers to run the motors as bipolar half-coil, giving a good balance between enough low speed torque but good high speed torque, with a reasonable current in the range of most controllers. Which incidentally is exactly the same speed performance and torque from bipolar half-coil as unipolar (which is always half-coil). Wiring the 8wire motor as bipolar series will give the expected increase in low speed bipolar torque or holding torque from the same motor, but reduces high speed torque by a factor of 4 as it has 4 times more inductance. Keep in mind people's CNC stepper motors rarely fail at low speeds or when holding!
So now you have practical and theoretical reasons why in the real world unipolar drivers can and often do give a better result than bipolar.
For the record, the big gain from bipolar motors comes in low-cost throwaway devices like bubblejet printers. These only use low motor speeds so the bipolar low-speed torque gain is worthwhile (and allows a smaller cheaper motor to be used), and there are savings from only using 4 wires, 4pin connectors etc, all good for a low cost disposable device. But in a properly designed high perfromance industrial machine the best choice may be unipolar/bipolar half-coil OR bipolar series/parallel, it depends on a lot of factors.
Interesting assertion, but I think you are misunderstanding how the bipolar drives work. During the pwm ON time, current is driven into the coil. During the pwm OFF time, the coil terminals are either shorted together (fast decay) or connected to the power supply in reverse. (fast decay). The current in the shunt resistor reverses, but not the current in the coil.
I am working on a DSP based drive. Allowing the current to flow backward would retard the rotor and cause resonance issues. I need to verify that it does not inadvertently happen.
Which video are you referring to? The one in your link showed that bipolar half coil has the same speed as unipolar. That is exactly what is expected. Is there a bipolar parallel/ unipolar comparison? Under load, I would expect the parallel one to be better.a given motor in unipolar mode WILL run faster than in bipolar parallel at the same drive power. If you don't believe me, or the video PMinMO posted, try it yourself.
It's easy to get motors with ~2mh inductances. I think the problem is when people get fixated on the holding torques and end up with motors that are totally unsuitable for the drives in their price range.
Valid point. But keep in mind that cheap bipolar and unipolar drives are similar in cost and specification. Even if the drive can only put out 3 amps, motor wired as bipolar parallel will run cooler due to the 50% reduction in coil resistance.With the larger more expensive motors that can be sourced in 8wire versions, they can be wired as bipolar parallel but often this is not done as it requires a higher operating current that many drives won't do (it usually pushes the current from the 3A/3.5A of most cheaper drivers to >4A etc where you can no longer use cheap drivers and need high end industrial drivers).
I do my own designs primarily for learning. I started out preferring the simplicity of unipolar but quickly became dissatisfied with their inferior utilization of the motor. I don't like settling for second best. Bipolar drives are more complex, but not by a huge amount.
I'd love to see some documentation that the current in the coil does not reverse.
James hosts the single best wiki page about steppers for CNC hobbyists on the net:
http://www.piclist.com/techref/io/steppers.htm Disagree? Tell him what's missing! ,o)
You're both right James you seem to be talking about slightly different things. To me it seems you are making the point that the current in a bipolar coil must reverse from step to step, which is completely true. Then H500 is talking about the specifics of the high speed recirculating current in a winding while it is being chopped, which he is correct that does not reverse (until the next full step of course).
But again with many people at the higher performance end of the CNC scale preferring to drive their 8wire motors with bipolar half-coil, there is no performance advantage, that is exactly the performance of unipolar. So there is no "second best". Only in the specific case of people being able to use bipolar parallel is there some gain, if the driver is capable of the increased current requirement.
Unipolar offers benefits in reliability with only 4 larger semiconductors used to switch the motor current rather than 8 smaller semis and the complexity of high side drivers. Also bipolar requires motor current to pass through 2 semis, which causes double the heating of the driver (which can be more than double as high side hbridge driving is known for inefficiency). Both are a significant problem, and I'm also a designer but design exclusively unipolar drivers for the same reason you mention, not wanting to settle for second best! Of course we are both biased in analysis of what is "best" based on what we value. Some people will be sure that servos are "best".
Posting Permissions
| | |
