Selecting PWM Freq for Speed DC Motor Control

[Fish4Fun]

What I have is a 120Vdc brushed, 16-field "surplus" motor in the ~1.5hp range @~18krpm with a date of mfg of 1978. What I am working on is a belt-drive spindle for a CNC router where the spindle speed needs to be variable from ~1krpm to ~6krpm, the belt reduction is 3:1. This is a one-off project so cost of components is trivial. I do not need reversing capabilities, so simple, low-side PWM is my plan....I do plan to have a Hall-Effect shaft sensor with 4 to 12 pulses per revolution for closed loop speed control. I plan to use an Arduino Mega 2560 as a dedicated controller. I would love a bit of input on a "starting" PWM frequency....and the pros and cons of increasing/decreasing the PWM frequency strictly from a motor performance point-of-view....it makes little difference if the switch(es) cost $1ea or $20ea if the expected motor performance is improved. I am also considering a fairly large capacitor bank both in the rectifier portion of the circuit and perhaps between the switch and the motor to reduce brush noise feeding back into the mains supply and to help provide a stable DC operating voltage. Again, thoughts/suggestions appreciated.

There are 8 armature slots and 16 poles, so the brushes should commutate @ 16 * 18,000 / 60 = 4.8khz @ 120Vdc no-load rated rpm....Intuitively I am thinking anything above 20kHz should be sufficient and was considering 25kHz to 30khz to keep the switching frequency out of the audible range....I thought I would **start** with IRF640s (I was thinking 4 in parallel)....I was thinking ~2000uF consisting of 6 * ECEC2DA331DL 330uF 200V capacitors + 2 * 4.5uF polypropylene 400V capacitors for both the post rectifier capacitors and the capacitors between the switch and the motor...mostly because I have all of these items on hand....but I am absolutely looking for suggestions in "general design", PWM frequency/duty cycle, and components. The above laundry list is just some items I have on hand that I think might be suitable as a starting place.

Any thoughts/advice/suggested reading on speed control algorithms would also be appreciated...

Thanks in Advance!

Fish

 

[Arouse1973]

Hi Fish
Do you have an initial schematic yet, we can see?
Adam

 

[Fish4Fun]

Arouse1973,

I don't....but I can....lol. Just a simple low-side chopper is all I am thinking about......never mind :-0 Here ya go...



Not sure if i should add a small inductor to limit the in-rush current to the Capacitor Bank/motor...or if I did add one what might be a reasonable value ...I would likely want to add and an EMI filter to the AC side of the rectifier, but I haven't thought it through enough to add it to the schematic. There is nothing exciting about the above circuit, it is just a simple chopper....mostly I want to make sure that there isn't some "resonant frequency rule" with brushed DC motors and choppers that I am unaware of,...or, for that matter, any other concerns/pitfalls that I should make an effort to avoid....

Thanks in advance!

Fish

 

[Arouse1973]

I think an inductor in series would be a good idea if the motors inductance is quite low. You will have to play with the value a bit not knowing the inrush current it would be difficult to work out. Do you need that much capacitance? Also why the 4 FETs and not one, are these ones you have already?
Adam

 

[Fish4Fun]

Ok, so after thinking a bunch, writing several replies that required me to think more...deleting them..thinking more....finally, I just bread-boarded up a similar circuit (that is, it is pretty much the same except for the components, voltages and motor, lol)....anyway, Using a single IRLZ34, a smaller motor, a 30V // 5A CC/CV bench power supply, various single capacitors in place of the banks across the motor and supply voltage....The supply capacitance proved to be fairly trivial for the case of the bench supply, (but it SHOULD be fairly trivial with a CC/CV bench supply)....The various capacitors across the motor, on the other hand, made a lot of difference in the wave forms on the scope, "the noise" and the speed of the motor.....

After lots of "fiddling" around, I realized that to get any speed control at all I had to drop the PWM base frequency WAY DOWN or DECREASE the amount of capacitance across the motor terminals. In the case of a 10,000uF Capacitor across the motor, a base frequency of ~250hz worked pretty well with duty cycles up to ~10%...beyond that the capacitor was simply "maintaining" the voltage during the "off period" and ramping back up very close to Vcc during the "on cycle"....obviously defeating the point of a PWM speed control....I am not sure exactly where to go from here....as a "chopper" my design is completely unsuitable for spindle speed control because I can't see clear to think of a way to make it respond fast enough to a dynamic load.....In one test run I dropped the capacitance across the motor to 1.5uF and increased the PWM frequency to ~350kHz but the underlying problems still exist.....(I tried several intermediate capacitance values and PWM frequencies all with lack-lustre results). So, I guess the next step is to simply design (or just BUY, lol) a constant voltage supply capable of handling the requisite power requirements of the particular motor in question.... (remember, all of these tests were done with a much lower voltage/power motor...50Vdc//450W Max, the motor I am interested in building a speed control for is 120Vdc // ~1.5kW....just didn't want to "test" with that much voltage/current).

Of course there is another option, lol, I could put all of this to rest for now and just order the 2.2kW BLDC spindle I have been eye-balling since I started building the machine I now need a variable speed spindle for, lol.....likely the best answer

Notes: Without a heat sink the IRLZ34 got very hot but it was an errant wire that let the smoke out of it.....once it was replaced with a new one and a large chunk of aluminum L-Bar was clamped to it with a good smathering of heat-sink compound (and the errant wire was carefully secured) it ran cool as a cucumber.... Moral of that story is "Remember to use a heat sink with power switches!"

Fish

 

[KrisBlueNZ]

You shouldn't need much if any capacitance across the motor. If you want to convert your PWM to an actual DC voltage, use a design based on a buck switching regulator - that is, a series inductor and a catch diode, with some output capacitance.

To reduce power dissipation in the MOSFET, make sure you switch it quickly - use a gate driver IC if necessary, and reduce the capacitance across the motor.

 

[Fish4Fun]

Hey Chris!

Thanks.....I was foolishly attempting to avoid a buck converter topology....**hoping** the motor's inductance would somehow "magic" a simple chopper into a speed control....because of the high current under load, I was thinking a buck converter might not be a great choice....Before I spend any more time fiddling with this, do you really think it is plausible to use a buck configuration in the 1kW to 1.5kW range? The "problem" is that the no-load current draw is relatively low and the "instantaneous" current demand can spike relatively high very quickly....I don't have the 120V no-load current on the actual motor I want to use (mostly because I don't have a 120Vdc source with galvanic isolation, lol), but @ 30Vdc the no-load current is ~450mA; it jumps quickly as a load is applied. I do not have specific data on the designed max current draw, but I suspect it could peak @ 15A or more....the internal resistance is ~8.5 ohms (+/-0.5 ohms depending on rotor position)....While I KNOW Low-Cost speed controls exist for similar motors (I have several routers with "built in speed controls" and they are less than $250 for the entire router, so it would seem a speed control in this power range can be designed and built for less than $250....) but I am not convinced I am going to be able to design/build a speed control in this power-range in a reasonable period of time regardless of how much money I throw @ it unless a fairly "simple" buck converter configuration will do the job (it is patently obvious a simple "chopper" is not going to work...).....I have a plethora of inductors on hand, and plenty of mosfets and mosfet drivers..... and don't mind ordering more of anything.....but I am more than a little concerned about the hazards of a 1kW+ line-powered supply....bit out of my "comfort zone" as it were.

I am assuming that routers with a built-in speed control use a thyristor based design, and perhaps this is a viable option, but I haven't really looked into the details....Or maybe I am just making this too hard? Is there an "easy button"? LOL, if there is it would be a FIRST for one of my projects ;-)

Fish

 

[KrisBlueNZ]

Thanks.....I was foolishly attempting to avoid a buck converter topology....**hoping** the motor's inductance would somehow "magic" a simple chopper into a speed control....

Well, it would, if you didn't put a whole lot of capacitance across the load. Look at a diagram of a standard PWM circuit. There isn't much, if any, capacitance connected across the load. You do need a diode across the load though.

because of the high current under load, I was thinking a buck converter might not be a great choice....Before I spend any more time fiddling with this, do you really think it is plausible to use a buck configuration in the 1kW to 1.5kW range?

I don't know if it would be practical or not. Those power levels are outside my experience.

The "problem" is that the no-load current draw is relatively low and the "instantaneous" current demand can spike relatively high very quickly....I don't have the 120V no-load current on the actual motor I want to use (mostly because I don't have a 120Vdc source with galvanic isolation, lol), but @ 30Vdc the no-load current is ~450mA; it jumps quickly as a load is applied.

I understand what you're saying, but I couldn't say whether it's workable or not.

I do not have specific data on the designed max current draw, but I suspect it could peak @ 15A or more....the internal resistance is ~8.5 ohms (+/-0.5 ohms depending on rotor position)....While I KNOW Low-Cost speed controls exist for similar motors (I have several routers with "built in speed controls" and they are less than $250 for the entire router, so it would seem a speed control in this power range can be designed and built for less than $250....) but I am not convinced I am going to be able to design/build a speed control in this power-range in a reasonable period of time regardless of how much money I throw @ it unless a fairly "simple" buck converter configuration will do the job (it is patently obvious a simple "chopper" is not going to work...).....I have a plethora of inductors on hand, and plenty of mosfets and mosfet drivers..... and don't mind ordering more of anything.....but I am more than a little concerned about the hazards of a 1kW+ line-powered supply....bit out of my "comfort zone" as it were.

Me too!

I am assuming that routers with a built-in speed control use a thyristor based design, and perhaps this is a viable option, but I haven't really looked into the details....

I think that depends on how fast the response needs to be. If you use the phase control method with a thyristor, you can't turn the motor OFF until the end of the mains half-cycle. Your design rectifies and smoothes the mains, then uses PWM at some other frequency that you choose. I assume you did that because you want immediate control of the output power.

This is really outside my experience - motor control, fast response, etc. If you post more details about the application, someone may be able to give you some specific advice.

Or maybe I am just making this too hard? Is there an "easy button"? LOL, if there is it would be a FIRST for one of my projects ;-)

Have you looked for projects, teardowns, or schematics of existing units online? What about application notes from companies that make devices that are used in this kind of thing?

 

[Fish4Fun]

Thanks Kris!

Your points are, as always, salient and tend to "tidy up" many of my conflicting thoughts. I am 100% confident that a slow and methodical approach with a recursive reading/searching/reading loop would eventually yield a solution....but this project is an attempt to merge two hobby projects....one electronic AND one CNC....I have not done the proper due diligence on the electronic side of CNC spindle control, but I am nearing the completion of a rather large CNC project, and it needs a spindle, lol. I am fortunate in that I don't really have any budget mandates that preclude me simply buying exactly what I want, so I am inclined to "shelf" the electronics project (for now at least) and order a spindle/speed control that is ready to bolt onto my CNC project and connect seamlessly with my existing CNC controller.... Using a DC motor was already a compromise (I wanted to build the spindle around a BLDC motor, but KNEW that design would require a much larger time investment....so I decided late-bound that I would "settle for a 'simple' DC motor/speed control" and re-vamp the spindle to a BLDC design at some later date....)....but it seems I have underestimated the complexity of a 1kW+ DC speed control; so, rather than "rush" in to inevitable short-term failure I think my "easy button is hidden in plain sight"....'Buy it Now'..... I will return to this project, perhaps this Winter, because I would like to add line-powered DC motor control to my tool box, but rushing into it right now is just silly....perhaps even delusional.

In the unlikely event anyone else is looking @ high-power, line supplied, DC motor speed control: I suspect the best "starting place" will be in "tread mill" related threads/schematics as there is a large "surplus market" in cheap DC tread mill motors...This page outlines two circuits, one Transistor based, one Thyristor based..... http://homemadecircuitsandschematic...treadmill-motor-speed-controller-circuit.html . There are many, many other examples of both approaches, generally with a manual/analog based control mechanism that could be replaced relatively easily with a digital interface with or without rotor position feedback.

I will update this thread when I have done due diligence on the topic and have time to do proper experiments and pay appropriate attention to safety concerns. I suspect that the thyristor based approach with proper line filtering and perhaps some PFC correction will ultimately be the best "general solution".

Fish