Post by thread:Stepper drive methods....

Chris Loiacono wrote: {Quote hidden}> First of all this is a personal project that I took on partly because > I don't have much experience controlling motors and want to broaden my > horizons a bit. With that said....... > > I'm driving 12V unipolar steppers with a 12V supply and am a little > disappointed at the performance - namely speed wise. I can get the > torque needed but only with low step rates. After putting together > and playijng with my own FET driver, I tried the Allegro 5804 chips, > which is what I am using now. They made the cotroller code and > hardware simple for pennies. An Allegro engineer then talked me into > checking out their bipolar microstepping chopper chips, which I will > do - but not before I get more out of the unipolars & 5804's. (or is > this a waste of time?) > > I understand that the voltage across the motor can be increased up to > a published limit as long as I add ballast to limit the current. I > did a simple test with some resistors and a supply I had on hand, but > it seemed that with the current at like levels, the performance was > approximately the same. > > Thomson/Airpax recommends L/4R - This sounds like 4X the voltage to > me. What about a 2R setup? > > Allegro suggests using zeners at a V a bit less than the supply level > to help dump the BEMF quickly. That would mean the diodes would need > to be rated for the full motor current, I would think.... > > I guess my question is if I should bother with any/all of this, or > scrap the uipolar design and move on to the bipolar chopper chips? If > I only knew how much improvement to expect by raising the voltage and > adding the zeners......Also, I don't think I can beat the under $10 > cost of the new closed loop chips with my own chopper design..... > Anyone feel like sharing any clues to help keep me from bad > decisions? - I really do not enjoy working with motors (yet) and > obviously have much to learn. Perhaps with more experience they will > grow on me. > > TIA, > Chris Hi Chris, it is a common belief that stepper motors work easily as a lamp. The fact that applying current to the phases make the motor move, doesn't mean everything is done. A stepper motor has a complex drunk and emotional personality. It requires care, study, learning and above all, understanding. You can read all night long about stepper motors, and find out less than 5% of what is possible, and ending up knowing less than you started. This is why there are so many different kinds, models and type os stepper motors, everyone was produced or developped trying to solve one or another problem or necessity in the market. Few things I can resume about them; Forget about voltage, pay attention in the current on the coils. It WILL require to use a power supply at least with double voltage then the stated to the motor. You can use current limiting resistors, or better, current limiting electronics. As anything else, each stepper motor is designed for a particular task, some have more torque, others speed, other both. It is common to see stepper motors getting warm or very hot under low speed, since the coils will drain pure DC current. Under medium speed, good torque, the motor can run not so hot, as a matter of fact, they could run without even getting warm. At high speed, the motor start to fail, the mechanics simply CAN'T react to the fast change in the magnetic field, it start to loose steps and make crazy things. A 1.8° step, 200 steps per revolution motor, can't give you speed higher than 80 to 100 RPM without starting to fail, or with a torque so low that you can't drive anything with. We should always remember that a stepper motor, is a precision mechanic motor, something developped to produce safe and accurate positioning in lab and industrial equipment or machinery. They are not intended to drive a fan blade or a drill. For those, there are other kind of motors. What kind or motor are you using? brand? model? voltage? current? what speed you got actually? Also, it is good to remember, that low voltage and high current motor gives you more torque and faster response. High voltage and low current motors are to be used where low torque and low speed is required, so low current drivers can be used, small transistors, etc. It will be difficult to drain high torque and speed from a 12V motor, you can start to hope this when using 5V and below motors. Best results you can get with motors around 2V @ 4A. I have some projects using Allegro chips, they work pretty well, but they require good heatsink and they don't drive higher currents. You really need to implement discrete high current transistors. 8 x IRF530 is a good solution for a 4 to 6A motor, but of course, it requires a current management, using 0.05 ohm resistors, a LM339 and a reference voltage, you can produce a current limiter or chopper and keep the motor working at its limit. Whenever playing with stepper motors, keep your mind open to strange results, write down every test for later comparison and consult, you WILL get lost during all the experiments of pulse time, voltage, current, speed, torque, etc. For the speed vs torque tests, the best is attaching a force drag to the motor, something like a belt driving a fan underwater or if you are good in mechanics, a felt type break system, a rotating aluminum disk sandwiched in between two stationary felt break pads, like a car's disk brake system, but using a felt or cloth to create the drag. Using springs and some adjustment screws you can produce and calibrate a precise drag to the motor. Then, you can play with pulse timming and current and find out the best torque and speed. One of the tests you can do for "missing steps", is just programming your microcontroller to do several back and forth steps, 20 FW, 135 BKW, 183 FW and so on, but ending up in the same exactly point where started, I mean, FW = BKW. Then you glue a pointer to the shaft (toothpick?) and expect it to stop exactly where started, if not, the motor lost steps. You WILL play for hours with this system, until you find out the best numbers for this particular motor. When you find out the motor limits for torque and speed, never make it work more than 85% toward those limits. Wagner. -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email spam_OUTlistservTakeThisOuTmitvma.mit.edu with SET PICList DIGEST in the body

> What kind or motor are you using? brand? model? voltage? > current? what > speed you got actually? I have been working with small 55mm 7.5 degree 12V , 30 ohm unipolar motors, 20 ohm 12V unipolar, and also some 5V bipolars @ 5.2 ohms. I have several bench supplies - up to 40V. Ihave the actual load assembled, and have been using it for motor testing over the past few weeks. I have run the 12V 20 ohm motors OK at between 60 - 100 RPM. I was able to get approx 4X the speed out of 12V 30 ohm motors by running them over the rated current, but, of course they don't like that very much. > > Also, it is good to remember, that low voltage and high > current motor gives > you more torque and faster response. That's why I went to 5V motors for next tests. Also Bipolar. {Quote hidden}>High voltage and low > current motors > are to be used where low torque and low speed is required, so > low current > drivers can be used, small transistors, etc. It will be > difficult to drain > high torque and speed from a 12V motor, you can start to hope > this when > using 5V and below motors. Best results you can get with > motors around 2V > @ 4A. I have a feeling that this may be more than I need for this application - the torque isn't very much. I don't have a small enough torque wrence to measure it, so I don't know exactly how light it is, but consider that all of the small unipolar motors drive it fairly well already. I spent a fair amount of time reading, listening, and slowly trying thigs to get this far already. > > I have some projects using Allegro chips, they work pretty > well, but they > require good heatsink and they don't drive higher currents. > You really > need to implement discrete high current transistors. Will I really need this for < 1A? 8 x > IRF530 is a good > solution for a 4 to 6A motor, but of course, it requires a current > management, using 0.05 ohm resistors, a LM339 and a > reference voltage, you > can produce a current limiter or chopper and keep the motor > working at its > limit. I have a good assortment of IRF devices handy - can you think of where I might find a simple circuit to try? At < 1A, would it be worth the effort? - The Allegro choppers are <$10 per axis..... > > One of the tests you can do for "missing steps", is just > programming your > microcontroller to do several back and forth steps, I have done exactly thi, except with much larger numbers of steps. I can lose steps if I try to step too quickly now, but slow rates as I mentioned never lose steps in two bytes worth of back and forth... > > When you find out the motor limits for torque and speed, > never make it work > more than 85% toward those limits. Perhaps this is why I'mnot happy with my results yet... Thanks.... -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email .....listservKILLspam@spam@mitvma.mit.edu with SET PICList DIGEST in the body

part 1 1943 bytes content-type:text/plain; (decoded 7bit) Roman Black wrote: > Wagner Lipnharski wrote: > >> It is common to see stepper motors getting warm or very hot under low >> speed, since the coils will drain pure DC current. Under medium >> speed, good torque, the motor can run not so hot... > > Sorry Wagner, that's wrong, at low speeds and stopped > the motor runs at min heat. At higher speeds the motor > inductance means that MORE average voltage is applied > per second for constant current, and the motor heat > *increases* at speed. > -Roman Not on my motors. I needed to increase belt reduction to speed up motors, so they cool down. At +-7 rpm the 5V @ 1.4A they almost boil water, the metalic frame where motors are bolted gets hot. Changing this to +-30 rpm they run smoothly warm. It was not caused by any ring or signal bouce. Power logger simply shows the current limit is reached after motor stops after each step, DC current, so power average goes to sky. Note that I don't use constant current, I use 12V and current limit. During many months tested different combinations, including the native Allegro PWM control, external current chopper, etc. You can see on the top waveform, there is a wasted full holding current period, just generating heat. I optimized software to cut out or reduce power (chopper) during the holding current, but then torque is reduced drastically. The bottom wavform uses the same control, but 4x faster, with 1/3 of the top power consumed. I can understand your explanation above if you apply constant current all the time, since during pre-mechanical motion the coils suck more power, then you will have a power peak during pre-motion and motion. This should not happens when limiting current, or just using a higher VCC with ballast resistors. -- http://www.piclist.com hint: To leave the PICList piclist-unsubscribe-requestspam_OUTmitvma.mit.edu part 2 9939 bytes content-type:application/octet-stream; (decode) part 3 2 bytes -

Hello all, If you really want a look at the top of the line in stepper motor and servo motor designs take a look at http://www.compumotor.com when only the best will do I use Compumotor. Having tried many commercial designs, and finding that most performed poorly or not at all. Seeing Roman's design, he has worked out most of the bugs I have found in many high end (so called by their Mfgs.) motor drive applications. The only commercial people willing to come on site and work with us, were the people from Compumotor. These motors are for the high end of things where precision counts, and most applications won't warrant the expense, but I find Roman's prices very reasonable for the product he has designed. So thumbs up to Roman. Vern Jones, Sound Research htp://http://www.foothill.net/~soundres Spehro Pefhany wrote: {Quote hidden}> > At 02:52 PM 2/10/2003 -0500, you wrote: > >I've looked at the lini-stepper, and might use it if there was profit to be > >had in this project. 3 Axes worth of linisteppers gets expensive though, > >especially for such small motors. What I'm doing now is pretty close and > >only costs about $10 per axis total. It may be that in taking the next step, > >the linistepper may be a bargain - but I just haven't gotten there yet. I'll > >look harder at using Roman's design... > > Chris:- > > If you really need high performance, you might want to look at DC servos > rather than steppers. With a built-in optical encoder (and some drive > circuitry) they can "look" just like a microstepping PWM stepper controller > to the driving software. Steppers have issues with resonances and such like, > as well as the obvious one of getting enough drive current due to winding > inductance. > > Best regards, > > Spehro Pefhany --"it's the network..." "The Journey is the reward" > spamBeGonespeffspamBeGoneinterlog.com Info for manufacturers: http://www.trexon.com > Embedded software/hardware/analog Info for designers: http://www.speff.com > > -- > http://www.piclist.com hint: To leave the PICList > TakeThisOuTpiclist-unsubscribe-requestEraseMEspam_OUTmitvma.mit.edu -- http://www.piclist.com hint: To leave the PICList RemoveMEpiclist-unsubscribe-requestTakeThisOuTmitvma.mit.edu>

Hi all, It was interesting to see the scope photo's of Wagner's Stepper pulses. They demonstrate the problems associated with driving steppers really well. However, much of the discussion seems to have missed some of the basic theory of why they work the way they do and why there are different methods of driving them. If you go back to basic electronics remember that applying a voltage through a switch and a resistor to a discharged capacitor will show the voltage at the capacitor starting at 0 and gradually increasing until it reaches the supply voltage. The current starts high and tapers off to zero amps. (Discounting leakage here). An inductor (like the winding of a stepper motor) works the opposite way. Apply a voltage across the windings and you get 0 current initially and the full supply voltage. Over time, the voltage drops to almost zero and the current to maximum based on the internal DC resistance of the wire in the inductor. Time and the DC resistance is the issue here. Like any electromagnet, the strength of the field is related to the current through the windings and to get to that maximum you want the current to max as quickly as possible. The current rating is based on the internal resistance of the windings and the saturation point of the magnets and core. More current doesn't infinitely translate into more torque. Plus, because the stepper moves the armature, the inductance changes as the rotor turns. You can see that in Wagner's photo. Once the system is stable again the current once again builds up to some maximum. Thing is, for max torque you want that current as high as allowed as soon as possible. The only way to get the current into the winding faster is to increase the pressure on those electrons (increase the voltage). However, problem is that once the maximum current is reached, the extra voltage tries to increase it way beyond the coil rating; not a good thing. So there are 3 ways to do this all. 1. The L method. This is where the coil voltage and internal resistance of the stepper motor winding match the applied voltage. Simple to use but not good for high speeds because it takes time to get to maximum current and if that isn't reached before the next step pulse, the torque is lower than maximum and eventually the load of turning the rotor exceeds the torque built up during the short pulse. 2. The LR method. Very popular and again very simple but also has some drawbacks. The power supply must be capable of a fairly high voltage an current and once the current is reached the extra power is dissipated through the current limiting resistors. A 12V, 1A stepper with 72 volts applied ultimately needs to drop 60V at 1A (60W) across the power resistor; and there can be 4 of those resistors. The advantage is that during the initial application of power, the higher drive voltage causes the motor current to ramp up much faster. This means that maximum torque is reached much faster and the motor can turn faster. One down side is that as the current increases, the voltage is dropped across the resistor and the drive voltage is reduced slowing down the rise time. The other down side is high power resistors are usually wire wound and have their own inductance that adds to the problems; it must be much much lower than the motor inductance or it restricts the current build up. 3. The Chopper method. This is similar to the LR method but uses a transistor to apply the high voltage drive to the coil. A low resistance (low inductance) sensing resistor is in the circuit and the voltage developed across this is compared to some sort of limit which shuts down the current. So now you have the full supply voltage (72V) applied to this 12V coil and once 1A of current is produced, the current is chopped off. The magnetic field developed in the winding now begins to decay and as the current drops of the power is again applied effectively holding the current at 1A. How this is done depends on the design. It can be clocked or done at the natural resonance of the inductor and sensing filter capacitance. The advantage is the full drive voltage is there during the entire time up to maximum current resulting in maximum torque as quickly as possible. The disadvantage is that the circuitry is more complex. Now back to Wagner's photo. The dip occurs because the inductance changes during the rotor motion and the reduced current will limit the step rate because the torque starts to fall off. So imagine that the step rate makes pulses that are have the width of the bottom trace. You can see that there wouldn't be as much torque and the motor would start to skip steps. A higher power driver that chops will (usually up to 25kHz) maintain the current at maximum during the entire step up to the point where the steps happen faster than the current can build up with the particular applied voltage. In such a case, the motor will get pretty well just as hot running at full speed as it will sitting idle. Hope that helps. John Dammeyer Wireless CAN with the CANRF module now available. http://www.autoartisans.com/products Automation Artisans Inc. Ph. 1 250 544 4950 > {Original Message removed}

part 1 6164 bytes content-type:text/plain; (decoded 7bit) John Dammeyer wrote: > Hi all, > > It was interesting to see the scope photo's of Wagner's Stepper > pulses. They demonstrate the problems associated with driving > steppers really well. We took hundreds of pictures at that time, probably I can find one with the current, mechanical dragging and mechanical response of the final structure (vibration, echoes, delays, bending, etc). As far as we found out, different machines will make the same motors and drivers behave differently. A final fine tunning is always required. Sometimes this tunning would demand hardware changes. We have the same machine in two different models, one has long axis structure, so, mechanical vectors are different than the short one. We needed to adjust the software for this one, just to compensate the long mechanical echo response. Few microseconds more in the pulses avoid confront the echo mechanical load, it increased performance and reduced motor load / temperature. Of course, everyone here knows, using bigger motors would solve all the problems, but we can't use motors bigger than NEMA34 in this machines. The shown picture is a project documented image file. [snip] {Quote hidden}> 3. The Chopper method. This is similar to the LR method but uses a > transistor to apply the high voltage drive to the coil. A low > resistance (low inductance) sensing resistor is in the circuit and the > voltage developed across this is compared to some sort of limit which > shuts down the current. So now you have the full supply voltage (72V) > applied to this 12V coil and once 1A of current is produced, the > current is chopped off. The magnetic field developed in the winding > now begins to decay and as the current drops of the power is again > applied effectively holding the current at 1A. How this is done > depends on the design. It can be clocked or done at the natural > resonance of the inductor and sensing filter capacitance. > > The advantage is the full drive voltage is there during the entire > time up to maximum current resulting in maximum torque as quickly as > possible. > > The disadvantage is that the circuitry is more complex. One of the solutions we implemented was using the Allegro 2998, a two full bridge driver, one IC per motor. It has a current shut off input pin per bridge, it can be used to provide current chopping. I did it using a common emitter shunt resistor (0.5 ohm) and comparing such voltage with a fixed voltage from a resistor divider from 5V, around 0.545V, so, the chop starts around 1.1A. Later on we included a mini trimpot to better adjust the chop current. See pictured attached. I made it smaller than 5k, so no complains about the bandwidth. The picture is a cutout from a bigger diagram, but allow us to see the Allegro 2998 and the current chopper circuit. The circuit board track capacitance was enough to generate a good chop frequency. A single LM339 SOIC SMD IC was enough to provide the chopping of two motors (two Allegro 2998 chips). Two other resistors for each LM339 were used to "artificially lift" the shunt resistors voltage, so the LM339 cut off the 2998 via microcontroller pin when motor "power off" is required. Sometimes I want a motor to be fully released, not holding torque with any power applied to any phase, so this trick is very welcome. Yes, the circuit gets a little bit complex, but worth the effort. With a current chopper system, you increase torque a little bit, with drastically lower thermal. Not all motors work well with this solution, some require to drop down the chopping frequency, other requires you to delay when to start the chopping pulses, if not, they simply don't move (should not chop while starting to move). The 2998 works up to 50V and 2A if I remember well, what gives you good chance to play openwide, but be aware of power dissipation. The bipolar transistors inside the 2998 are a thermal pig, they require a nice heatsink if you play with currents above 1A constantly. Working in such limits, the 2998 package heatsink is NOT enough to transfer such thermal energy to the other heatsink, even using a fanned larger feathered heatsink and silicon compound. I am planning to produce a small PCB with SMT discrete components and 8x IRF530 (or better RDSON) to simulate the 2998, incorporating the chopper and resistors, so it can replace the 2998 directly in the actual board, and at the new ones too. > Now back to Wagner's photo. The dip occurs because the inductance > changes during the rotor motion and the reduced current will limit the > step rate because the torque starts to fall off. So imagine that the > step rate makes pulses that are have the width of the bottom trace. > You can see that there wouldn't be as much torque and the motor would > start to skip steps. Even with the chopper solution we have a similar power waveform from the other email picture, since there is no much to help. When the motor starts to move its rotor, the inductance goes to hell, the current is unpredictable, it will always depend on the mechanical motor load, and that, only depends on a thousand things. If the motor is fully stopped for several milliseconds, then start to move several steps at once, the waveforms change all the time, due the load echoes in the mechanics involved. If the load mechanics represent a long vector, then you are doomed, since sometimes dozens of milliseconds after the motor is stopped you see back current on the drivers, by mechanic delays and echoes. There is a situation when this mechanical echo is so strong that it generates back current into the Allegro 2998 and it goes back to the 5V logic, causing several problems and control loss. We found out two special situations, signature of the long mechanical driven, and we simply cut out the motor power when those two occasions would happen. It is ridiculous, but no other solution was at hand at the time. Wagner. -- http://www.piclist.com hint: To leave the PICList RemoveMEpiclist-unsubscribe-requestEraseMEEraseMEmitvma.mit.edu part 2 4887 bytes content-type:application/octet-stream; (decode) part 3 2 bytes -

part 1 6163 bytes content-type:text/plain; (decoded 7bit) John Dammeyer wrote: > Hi all, > > It was interesting to see the scope photo's of Wagner's Stepper > pulses. They demonstrate the problems associated with driving > steppers really well. We took hundreds of pictures at that time, probably I can find one with the current, mechanical dragging and mechanical response of the final structure (vibration, echoes, delays, bending, etc). As far as we found out, different machines will make the same motors and drivers behave differently. A final fine tunning is always required. Sometimes this tunning would demand hardware changes. We have the same machine in two different models, one has long axis structure, so, mechanical vectors are different than the short one. We needed to adjust the software for this one, just to compensate the long mechanical echo response. Few microseconds more in the pulses avoid confront the echo mechanical load, it increased performance and reduced motor load / temperature. Of course, everyone here knows, using bigger motors would solve all the problems, but we can't use motors bigger than NEMA34 in this machines. The shown picture is a project documented image file. [snip] {Quote hidden}> 3. The Chopper method. This is similar to the LR method but uses a > transistor to apply the high voltage drive to the coil. A low > resistance (low inductance) sensing resistor is in the circuit and the > voltage developed across this is compared to some sort of limit which > shuts down the current. So now you have the full supply voltage (72V) > applied to this 12V coil and once 1A of current is produced, the > current is chopped off. The magnetic field developed in the winding > now begins to decay and as the current drops of the power is again > applied effectively holding the current at 1A. How this is done > depends on the design. It can be clocked or done at the natural > resonance of the inductor and sensing filter capacitance. > > The advantage is the full drive voltage is there during the entire > time up to maximum current resulting in maximum torque as quickly as > possible. > > The disadvantage is that the circuitry is more complex. One of the solutions we implemented was using the Allegro 2998, a two full bridge driver, one IC per motor. It has a current shut off input pin per bridge, it can be used to provide current chopping. I did it using a common emitter shunt resistor (0.5 ohm) and comparing such voltage with a fixed voltage from a resistor divider from 5V, around 0.545V, so, the chop starts around 1.1A. Later on we included a mini trimpot to better adjust the chop current. See pictured attached. I made it smaller than 5k, so no complains about the bandwidth. The picture is a cutout from a bigger diagram, but allow us to see the Allegro 2998 and the current chopper circuit. The circuit board track capacitance was enough to generate a good chop frequency. A single LM339 SOIC SMD IC was enough to provide the chopping of two motors (two Allegro 2998 chips). Two other resistors for each LM339 were used to "artificially lift" the shunt resistors voltage, so the LM339 cut off the 2998 via microcontroller pin when motor "power off" is required. Sometimes I want a motor to be fully released, not holding torque with any power applied to any phase, so this trick is very welcome. Yes, the circuit gets a little bit complex, but worth the effort. With a current chopper system, you increase torque a little bit, with drastically lower thermal. Not all motors work well with this solution, some require to drop down the chopping frequency, other requires you to delay when to start the chopping pulses, if not, they simply don't move (should not chop while starting to move). The 2998 works up to 50V and 2A if I remember well, what gives you good chance to play openwide, but be aware of power dissipation. The bipolar transistors inside the 2998 are a thermal pig, they require a nice heatsink if you play with currents above 1A constantly. Working in such limits, the 2998 package heatsink is NOT enough to transfer such thermal energy to the other heatsink, even using a fanned larger feathered heatsink and silicon compound. I am planning to produce a small PCB with SMT discrete components and 8x IRF530 (or better RDSON) to simulate the 2998, incorporating the chopper and resistors, so it can replace the 2998 directly in the actual board, and at the new ones too. > Now back to Wagner's photo. The dip occurs because the inductance > changes during the rotor motion and the reduced current will limit the > step rate because the torque starts to fall off. So imagine that the > step rate makes pulses that are have the width of the bottom trace. > You can see that there wouldn't be as much torque and the motor would > start to skip steps. Even with the chopper solution we have a similar power waveform from the other email picture, since there is no much to help. When the motor starts to move its rotor, the inductance goes to hell, the current is unpredictable, it will always depend on the mechanical motor load, and that, only depends on a thousand things. If the motor is fully stopped for several milliseconds, then start to move several steps at once, the waveforms change all the time, due the load echoes in the mechanics involved. If the load mechanics represent a long vector, then you are doomed, since sometimes dozens of milliseconds after the motor is stopped you see back current on the drivers, by mechanic delays and echoes. There is a situation when this mechanical echo is so strong that it generates back current into the Allegro 2998 and it goes back to the 5V logic, causing several problems and control loss. We found out two special situations, signature of the long mechanical driven, and we simply cut out the motor power when the two occasions would happen. It is ridiculous, but no other solution was at hand at the time. Wagner. -- http://www.piclist.com hint: To leave the PICList RemoveMEpiclist-unsubscribe-requestspam_OUTKILLspammitvma.mit.edu part 2 4887 bytes content-type:application/octet-stream; (decode) part 3 2 bytes -

On Monday 10 February 2003 02:08 pm, you wrote: > Chris:- > > If you really need high performance, you might want to look at DC > servos rather than steppers. With a built-in optical encoder (and > some drive circuitry) they can "look" just like a microstepping PWM > stepper controller to the driving software. Steppers have issues with > resonances and such like, as well as the obvious one of getting > enough drive current due to winding inductance. Spehro, When I first read this, I thought you were talking about R/C type servos. ;-) Now I realize that you must mean something a little more industrial. This sounds like a really good way to do it. To put it mildly, my experiences with using various cheap stepper motors in my little robot project have been less than satisfactory. Much of it being virtually no torque when running at any speed and not much at slow speeds either. Plus the noise and vibration are quite irritating. I've been thinking about switching to gearhead motors and using PWM, but didn't like the idea of losing the dead reckoning ability of stepper motors. I realized that by using an optical encoder or some other feedback mechanism I could get some good acuracy, but didn't think of just adding another layer of processing to get a PWM encoded wheel to look like a stepper. Thanks for the information. michael brown -- http://www.piclist.com hint: To leave the PICList RemoveMEpiclist-unsubscribe-requestTakeThisOuTspammitvma.mit.edu>

On Monday 10 February 2003 09:36 pm, you wrote: > In my experimentations with building a robot platform I have had a > similar experience. I started out using a pair of Vexta PK245 > stepper motors as drive. These require about 1 amp per phase which > is 2 amps per motor with two windings excited for max torque. So > with 4 amps as a current requirement I had to go with a sealed lead > acid battery. Even a small one weighs a lot! Performance was very > disappointing (it barely moved on a hard surface!). > > I eventually switched to 1/4 scale RC servos. These make for very > inexpensive gearhead motors when modified to rotate freely. I > removed all the electronics and drove it with PWM from a PIC. The > ones I have will put out 150+ oz-in of torque and typically draw > 100mA no load to 500mA loaded heavy. They cost about $30 each. > > Now, I did not have a requirement for precise positioning. So you > still need to tackle that problem. > > Let me know if you want further info. I've been recently playing around with PWM and a modified R/C servo. (I sure like that 628 ;-) They have far more power than the steppers I played with, even though it's really a standard size servo. I've just been a little intimidated by the encoder complication. With Spehro's concept, I can have my cake and eat it to, sort of. I will just make the encoded servo look like an "ideal" stepper motor to the software above it, and then treat it thusly. I'm hoping that this will make adapting to a stronger type motor easier. I appreciate the info on the lower power requirements, as I too have been dragging around a big SLA battery. I also appreciate the offer on the info, but I really need to try and slay this dragon myself. I'm sure I'll have plenty of questions for you and the group. ;-) michael -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads

Chris Loiacono wrote: > > I've looked at the lini-stepper, and might use it if there was profit to be > had in this project. 3 Axes worth of linisteppers gets expensive though, > especially for such small motors. What I'm doing now is pretty close and > only costs about $10 per axis total. It may be that in taking the next step, > the linistepper may be a bargain - but I just haven't gotten there yet. I'll > look harder at using Roman's design... > > (Funny how I'll spend client $ somuch faster...) Ok, so you need good high speed performance for a *very* low cost?? Here's a way i've done it with low power motors. Use a unipolar driver, 2-phase on wave drive. This can be as easy and cheap as 4x cheap NPN transistor and 4x base resistor. If motor current is under 500mA per phase you can use 4x BC337, about 5c each. For a few more pennies you can use a octal darl driver chip and lose a bit of performance. The 4x transistors are configured to turn on hard and have no significant losses. Then drive the motor + (common) winding with a constant current driver. This can be as cheap as a TO-220 PNP darlington, 2 resistors and a zener. And a heatsink. :o) This setup will outperform your R/LR proposal and be very cheap, WELL under your $10 per axis total. The higher the PSU voltage the more heat lost at low speeds via the heatsink but the more motor torque you get at high speeds. -Roman -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads

Wagner Lipnharski wrote: > > Roman Black wrote: > > Sorry Wagner, that's wrong, at low speeds and stopped > > the motor runs at min heat. At higher speeds the motor > > inductance means that MORE average voltage is applied > > per second for constant current, and the motor heat > > *increases* at speed. > Not on my motors. I needed to increase belt reduction to speed up motors, > so they cool down. > At +-7 rpm the 5V @ 1.4A they almost boil water, the metalic frame where > motors are bolted gets hot. Changing this to +-30 rpm they run smoothly > warm. It was not caused by any ring or signal bouce. Power logger simply > shows the current limit is reached after motor stops after each step, DC > current, so power average goes to sky. > > Note that I don't use constant current, I use 12V and current limit. Ok, i'll buy that. The one case where the motor runs cooler at high speed is with a *constant voltage* driver, BUT generally nobody does this as the resulting very low coil current gives very low torque at high speeds making it only suitable for very low power loads. I'd also suggest that instead of changing your gearing to cause less motor current and less heating, you could probably just have lowered the PSU voltage to give less current and I2R losses and caused the same effect; less motor heat and less motor torque. -Roman -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads

Roman Black wrote: {Quote hidden}> Wagner Lipnharski wrote: >> >> Roman Black wrote: >>> Sorry Wagner, that's wrong, at low speeds and stopped >>> the motor runs at min heat. At higher speeds the motor >>> inductance means that MORE average voltage is applied >>> per second for constant current, and the motor heat >>> *increases* at speed. > >> Not on my motors. I needed to increase belt reduction to speed up >> motors, so they cool down. >> At +-7 rpm the 5V @ 1.4A they almost boil water, the metalic frame >> where motors are bolted gets hot. Changing this to +-30 rpm they >> run smoothly warm. It was not caused by any ring or signal bouce. >> Power logger simply shows the current limit is reached after motor >> stops after each step, DC current, so power average goes to sky. >> >> Note that I don't use constant current, I use 12V and current limit. > > > Ok, i'll buy that. The one case where the motor runs > cooler at high speed is with a *constant voltage* driver, > BUT generally nobody does this as the resulting very > low coil current gives very low torque at high speeds > making it only suitable for very low power loads. > > I'd also suggest that instead of changing your gearing > to cause less motor current and less heating, you could > probably just have lowered the PSU voltage to give > less current and I2R losses and caused the same effect; > less motor heat and less motor torque. > -Roman Roman, you are correct. If you observe the picture I post yesterday, the top waveform (low rpm) it has two stages, first is the moving power when the motor advances one step, the second stage is the holding power, when the motor keeps its steady position and holds the shaft while the mechanics attached finishes the delayed motion. It is important to make sure the mechanical step completes. The bottom waveform (rpm x 4) shows practically only the stepping power, no holding power appears there, there is no time for it, even that we apply the same step power pulse time. It means that the system is more susceptible to mechanical failures due its own delay. In some opportunity we were plotting mechanical movement using a small linear voltage generator tied to the motor shaft and other at the end of the mechanical long shafts. You can scope the generator output and visualize perfectly how the mechanics answer to the motor. At certain high rpm, the stepper motor moved almost the complete step, then it came back to the original position. The mechanical load, even somehow light mass and torque, at that acceleration for a single step it was more than what the motor could do, so, to the motor it was almost as if the load was locked down, the motor shake and backed up. Of course, torque went to hell. At that point we decided to increase motor size to NEMA34 with higher power, and definitely never go for a motor with step bigger than 2 degrees, and learn the lesson to never more use direct metal connections (gears or direct shaft drive) - always timming urethane belts, they absorb and store energy. We also implemented a slow start motion by software. The application is a CNC machine with X and Y axis. Vectors size can go from 8 to 65k steps. It means that if when doing a 1000 steps vector, the first and last 10 or so steps will be done in low speed, the middle steps in normal high speed. This ensure a steady mechanical motion start and a good and solid motion stop. In the middle of the motion, the stepper motor just keep pushing the mechanics forward. It is important to find out a nice high speed while the mechanics load delay and echo serve as its own motion filter and ending up as an almost linear motion, without much noise and mechanical ripple. The first and last 10 steps are done in less than 100ms, so, "normally" not accounted as speed error. This acceleration and break speed change is dependent to how steep is the angle from the actual to the new vector to be done. If less than 30° and no stop between vectors, then no speed change is necessary. Wagner. -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads

>In some opportunity we were plotting mechanical movement using >a small linear voltage generator tied to the motor shaft and >other at the end of the mechanical long shafts. You can scope >the generator output and visualize perfectly how the mechanics >answer to the motor. At certain high rpm, the stepper motor moved >almost the complete step, then it came back to the original >position. The mechanical load, even somehow light mass and torque, >at that acceleration for a single step it was more than what the >motor could do, so, to the motor it was almost as if the load was >locked down, the motor shake and backed up. I think you are observing what I have seen described in an article in Hewlett Packard Journal from the 1970's. The article describes the design of an X-Y plotter, and how they used the proposed stepper motor as a generator to see the waveform from the windings (presumably with all the other hardware attached), and then using a ROM based waveform generator to make the drive waveforms, containing harmonics upto about the 3rd or 5th harmonic of the step rate. They found this was the only way to get the motor to step fast enough to achieve the desired pen speed. -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads