Motorcontrol - speed regulation

You are building a voltage-controller oscillator - the more volts you put on your motor, the faster it spins and the higher the frequency coming out of your Hall element.

Get a data sheet for a 4046 phase-locked loop chip and read the theory from there. There are better texts, but they are more expensive.

You have a super lowcost motorcontroller, do you know how to program it ??

What can be lower cost then software ??

I think you should start learning how to program your super low cost motorcontroller first, then learn how to control its internal timer, than learn how to make a PWM.

good luck

donald

Ohh no, please let me clarify. I do not have anything yet, I have made the plans to BUILD it. I plan to build the motorcontroller entirely from simple components such as opamps. I already have figured out how to make the PWM signal from a sawtooth generator and a comparator. This i've got running already.

My problem is to adjust the speed (i.e the pulsewith to the MOSFET) when the speed drops. My PWM level is determined via a trimpotentiometer, so I need to feed the feedback from my hall-sensor to this circuit somehow.

It would be easy for me to do this if only I could use a uC, but unfortunately I cannot. The reason is the added cost from writing and (possibly) maintaining firmware and the production process where firmware has to be downloaded to the device. We wish to make this design using only "standard" components, if such a term exists....

Hope this clarifies my problem.

Thanking you all in advance.

Best regards Henrik

I would guess that not using a 50 cent micro will cost more in parts, design costs, and support.

Another option is a 555 timer driving a mosfet, but this does not incorporate feedback from your sensor for speed control.

Luhan

On a sunny day (4 Dec 2006 11:38:12 -0800) it happened "Luhan" wrote in :

If you start a one shot from the hall sensor, and lowpass that, then you have a frequency discriminator, feed that into an opamp and on the other input of the opamp a reference voltage. The output from the opamp then changes the voltage your trimpot sets.

lowpass R var DC hall sensor -- oneshot -----===----------------------------|\\ _ _ | + | \\________ A ___| |___| |_ === C | | / | [ ]------------|\\ |__| |____| | \\__________ PWM /\\/\\ ------| / triangle |/

You have to dimension things so the things stay in range, maybe add some PID to get better lopp stabilisation.

Got it; your current need is to make a voltage proportional to the motor speed for comparison to the potentiometer position.

The easy way is to use the Hall sensor pulse to (1) sample-and-hold a capacitor's voltage, (2) discharge the capacitor (use a one-shot triggered by the sensor to do the discharge after you're sure the HOLD function is complete). Then simply connect a current source to the capacitor (or use an op-amp integrator). When the motor is slow, the HOLD voltage output is high, when the motor is fast, it gets lower. And, the HOLD output voltage will not change except once per cycle, so should be a suitable signal to feed your PWM stage. Some linear combination of the setpoint potentiometer and the HOLD voltage will be a suitable speed-error indicator.

For best accuracy, using the speed-error with an added integral term (PID control, but differential = 0.00, remembering the hold-cycle glitches) may be useful.

This won't run as well when it is sitting next to a DC motor with 7A running through the commutator. Fully digital control scheme look a lot nicer in noisy environments.

If you were expert with analog and good at electromagetic compatibility, you might do better - I'm both, and didn't do too well the last time I tried to get a saw-tooth generator and comparator to PWM a couple of amps running through a Peltier junction (which doesn't even have a commutator). It was just for prototyping, so I couldn't apply all the tricks in my armoury, but it was still a bit chastening.

Microcontrollers are about as standard as components get these days. Look at the range of Arizona Microsystems PIC processors that Farnell stocks, and compare it with their range of analog comparators.

The added cost of writing and maintaining firmware is peanuts compared with the costs of debugging a cranky and sensitive analog circuit - good analog engineers can get comparators to work, but they tend to need four layer boards with buried ground and power planes and a couple of reoganisations of the sensitive bits of the layout. They will generally try and design out as many of the comparators as possible in the process.

Less experienced analog engineers are braver, and much more expensive.

If you're planning on going into full-on production, then it's insane not to spring a hundred bucks on a development system, and a few weeks writing a PWM.

Then, if the product takes off, you just order them mask-programmed.

Just for perspective, I've worked at places where each item needed programmed ROMs, and we just had a "PROM programming stateion", where we'd pre-program them, and the final assemblers would just stuff them.

Good Luck! Rich

On a sunny day (Wed, 06 Dec 2006 22:25:50 GMT) it happened Rich Grise wrote in :

The only issue _was_ in the past that the flash could go duff in a few decennia, and the hardware based solution would still work after 40 years. Thse days I think flash wil lalso last 100 years? But think of those poor archeologist who dig up a cellphone 10000 years from now (they could be apes) and then find it does not work ;-) OK , forget it .... LOL