PWM and the practical slowest speed

[definitionofis]

I read a lot of PWM links and cannot find the answer
to the question as to what is the lower limit of speed that
can be attained that way.

My power window motors are too fast (100 rpm).
I discovered that gears will be expensive and somewhat rare,
to match the 12 pitch pinion on these motors.

I am going to mount my axle directly to these motors
and use PWM to slow them.

Can I get less than 1 rpm that way ?

At some point I suspect torque will die quickly and
the speed drop to zero from perhaps 30% of full speed
with no gradation in between. My guess is that PWM will
give some drop off to failure, suddenly, at some
known practical limit of slowness.

So what is that limit ?  Is it 10% or 20% or 1% of full speed ?
Some of you likely already know, without me wasting
time experimenting.

[Soeren]

Hi,

At 100 RPM there's gears involved, so the interesting thing is how much.
Without that bit of info, it's hard to say how low you can go (it will depend on other specs of the motor as well).
To get a reasonable torque AND low RPM (without gears) you need feedback, but then you can go pretty low, but 1 RPM (i.e. 6°/s) takes gears.

What are you gonna use the motor for?
Why not just replace the pinion mounted now with something that you can get at a reasonable price?

[definitionofis]

I thought maybe the gearing, which is already involved, would have enough internal
momentum to let me PWM it quite low.  I have no internal motor specs..

I am using it to move a 22 lb. robot using 0.22 m diameter wheels.
I cannot easily pull that pinion. I could afix a 3/4" iron pipe over it and
couple it to another shaft with a different gear set.

I found a sensor and a clear plastic disk with an array of almost invisible hair-lines,
used as a positon/speed sensor in an old broken Lexmark ink-jet printer.
I will use that for feedback.

So give me a percentage low end speed cut off from your experience using PWM.

[definitionofis]

This wiper motor page was helpful:
http://www.scary-terry.com/wipmtr/wipmtr2.htm

Plus I did a test with my 100 rpm motor where I just tagged it
on/off as fast as I could, once.

I got a momentum drift of 1/4 to 1/3 of a turn.
That is 90 to 120 degrees.  

I surely do need to gear it down for my slow moving needs.

PWM is likely to not get it slowed enough.

[Soeren]

Hi,

I thought maybe the gearing, which is already involved, would have enough internal
momentum to let me PWM it quite low.  I have no internal motor specs..

Yes, but "quite low" probably have another meaning for you than for me and circumstances will change it for us both, so in engineering, we prefer metrics
In a case like this, it's allways best to experiment a bit to see what's attainable.

I am using it to move a 22 lb. robot using 0.22 m diameter wheels.

All that weight will help to make it NOT go that far with a single pulse.

I cannot easily pull that pinion. I could afix a 3/4" iron pipe over it and
couple it to another shaft with a different gear set.

Heat (think torch or hot air gun) helps a lot when you need to remove such a part and there's tools for this as well.

I found a sensor and a clear plastic disk with an array of almost invisible hair-lines,
used as a positon/speed sensor in an old broken Lexmark ink-jet printer.
I will use that for feedback.

Yes, I have a number of those myself. They might work, but only if you don't intend to go very fast, as that might lead to information overflow. In that case, a disc with less segments could be made.

So give me a percentage low end speed cut off from your experience using PWM.

Like I said - Nobody can answer that, as there's too little exact data, but I think you might be able to make something useable out of it - experiment - and change what's not working along the way.

I did a test with my 100 rpm motor where I just tagged it on/off as fast as I could could, once.

I got a momentum drift of 1/4 to 1/3 of a turn. That is 90 to 120 degrees. 

I surely do need to gear it down for my slow moving needs.

PWM is likely to not get it slowed enough.

You have no chance making short pulses that way!
With a single pulse of say 1ms (1 thousand of a second), you would get much less and you can brake the motor when it's not pulsed, so it will stop much faster.

[definitionofis]

I like the braking idea during no pulse. Thank-you.
I forgot about that technique, and yes, the load will slow it.

So I might give this a go with one small 12 pitch gear or no gears.

I'll report back what percent of 100 rpm I got it down to, after I finish.

[definitionofis]

Here is my first data point regarding a practical lower limit, using PWM on my 100 RPM motor.

This likely will be useful to somebody as a future reference.

My phidgets board can do 125 pulses per second as a maximum PWM frequency.
It is not a real PWM controller. It is just a input/output board (Phidget model 0/16/16).

So I decided to send an arbitrary pulse of 1/125 of a second on-time and then a long pause.

The rotation of my motor with no load was 1/72 of one revolution, or
in other words, five degrees.  This was much smaller than I expected since
my manual slash-a-wire-across-the-terminals test was no shorter than 1/4 turn, ie. 90 degrees.
Note that this 1/125s pulse result, is very much the same as 100rpm/60seconds/125 of a rotation.
So the gear box and motor are responding very precisely to a slice of on time.

Before doing this, I thought the small on-time might produce no torque.
I could not physically hold the 3/4 inch shaft still with my fingers.
So although I expect a smaller number of degrees after I attach the chain and sprockets,
and 22 lb. mass, it might not be much smaller than 5 degrees; ie. the torque is tremendous.

The internals of my motor's gear box is a worm gear driving a large gear.
So the hold positioning is strong (no roll back) and the result is very repeatable.
This might not be so with a different kind of gear system and motor,
even if it was also 100 rpm..

I did not try less than a 1/125 second on-pulse as yet.
I think that will work too.

I am very pleased with this result because it means that after I
attach my 2:1 chain reduction, and given my 25 inch diameter wheels,
I might be able to move in 0.17 inch increments or less.
(25/2/72 = 0.17)

I was fearing 25/4 = 6.25 inch increments with no chain/sprocket reduction.
I might have been able to attach these motors directly to a wheel shaft without
chain and sprocket and still achieved small incremental motion.

I'll report load conditions and other shorter pulses another day.
I did not yet try braking during the off-cycle.