Page 1 of 1

Motor Speed Control Issue

PostPosted: Sat Apr 27, 2013 2:10 am
by Ralph
Rather then jump directly to a sync-driven loop to control the motor speed of my new Nipkow televisor, I wanted to play with simulating the use of a synchronous motor. As shown below, I opted to modify the Peter Smith/Club motor control circuit. The picture that is displayed shows the schematic at low resolution, but you can download the full-resolution version if needed.

The strategy seemed simple - the 4046 PLL would be driven from the optical detector on pin 3 and, instead of the standard detected sync pulses, a crustal-controlled 400 Hz square wave on pin 14. The optical sensor is reflective, and looks at an evenly spaced set of black and white pairs - 32 pairs in all. It works fine. The 400 Hz tone source is a 4060 with a 3.2768 MHz crystal-controlled internal oscillator and I am taking the signal at pin two of the divider network (/8192) and the result is a 400 Hz square wave.

The resistors on either side of the 10K speed control are actually 10K pc trimmers so that I can play with the bias network as opposed to a pair of 4700-ohm resistors. I am using a 12V motor rated at 1800+ RPM at 2A under full load - the running load is notably less.

With the 400 Hz signal at pin 14 of the PLL and the optical chopper input at pin 3, I expected to be able to lock the system at 750 RPM with proper adjustment of the 10K speed pot and the associated bias resistors. That is not what I get!

With the 4046 in the circuit, I can get the motor up to slightly less than 750 RPM if I set the high bias resistor to 0 ohms and the low one to 10K. If the 10K ipeed pot is set to maximum, the motor speed settles into something a bit below 750 rpm.

If the optical sensor input to pin 3 of the PLL is disconnected, the motor speeds up and will peak out at its normal maximum speed at 12V - a bit over 1800 RPM. When I reconnect the sensor to pin 3 of the PLL, the motor immediately starts to slow down. As it passes through 750 rpm (determined by watching the image display), the picture jerks few times, as if it were trying to lock, but then it slows to something just under 750 RPM and stays there. A few times, by letting the disc speed up by breaking the connection at pin three and then restoring the connection, the system WILL lock at 750 RPM (confirmed by the image display and the strobe), but that happens rarely and eventually it snaps of of lock and settles to a constant speed just a bit slower.

With pin 3 disconnected, I can juggle the speed pot and bias resistors to get it to free-run at just about 750 RPM, but when the optical line is reconnected to pin 3, the disc immediately slows way down to something around 375 RPM.

If someone can point me to what I have over-looked, it would certainly be appreciated....


PostPosted: Sat Apr 27, 2013 4:28 pm
by gary
Ralph, This circuit is notoriously difficult to debug over the forum and so I make not pretence that I know what your problem is but I have a couple of questions and suggestions that *may* lead to something.

Firstly your circuit looks fine other than to say there are some issues involved with changing the values of the components so marked * in the club circuit. If you have the latest news letter Karen explains these beautifully.

Is your encoder with or without the missing pulse? In other words do you have 32 pulses coming in from the xtal osc. and only 31 from the encoder?

When this circuit is locked the output of pin 13 is at high impedance, therefore the circuit behaves as if the PLL isn't there at all. So to start with, it is always a good idea to either remove the pll temporarily or disconnect pin 13 whichever is easiest. Having done this then all components and voltages should be so modified that when VR1 is at the midway position the disk is turning at or close to 750 RPM. The PLL is then reconnected and lock should be obtained by adjusting VR1.

If hunting occurs then C4 should be adjusted accordingly.

I must say I have never been able to get this and similar circuits to work reliably without further pulse shaping of the encoder output - however others routinely do it seems, so I do not proffer that as a solution merely as something to look at *perhaps*.

It is possible, I think, for this circuit to lock at 375 RPM as it is an even harmonic - it should result in a steady and periodic stream of error pulses coming out of the PLL which, when filtered would result in a constant voltage. I have noted this happening with similar circuits. That is the only explanation I can see for this phenomenon unless your xtal osc was 200 hz, and I am damn sure you have checked for that!

That's all for now, maybe others have further ideas. Good luck!

PostPosted: Sat Apr 27, 2013 5:23 pm
by Steve Anderson
I agree fully on the points mentioned by Gary above. As you've mentioned 375RPM (half-speed) there may be something in this.

Check the encoder output (pin 3 of the 4046), it should toggle between 0V and 12V cleanly and rapidly, I emphasize that as these devices are not rapid when configured such. The datasheet for the QRD1113 mentions a 10us rise-time and a 50us fall-time...and that's into a 100R load, it probably will be a lot slower into 10k.

This could cause the input stage of the 4046 to internally bounce as the encoder output slowly transitions the 4046 isn't a Schmitt-trigger input. As Gary mentions I would apply some processing on the output of the detector as a routine.

I will also mention my old favorites of decoupling/bypassing and good grounds/earths. A small (100n) disc-ceramic capacitor across the motor terminals may help as the motor is sharing the same supply as the logic. You could try separating these supplies as an experiment. Better still, use a 12V Lead-Acid battery for the motor temporarily.

Steve A.

PostPosted: Mon Apr 29, 2013 5:58 am
by Ralph
Gary and Steve,

Thanks for tge useful input. I am already and work on the Mk. 2 version of the circuit with more silicon on the board. Both pulse inputs to the 4046 will have Schmidt-trigger buffers. My plans include a lot of odd experiments and that will save the trouble of worrying about pulse rise and fall times (within limits), no matter what I hook in. And yes Gary, the circuit is designed for the equivalent of 32 pulses at both inputs.

Let you know what happens in a week or so......


PostPosted: Sat May 04, 2013 3:48 pm
by Ralph
The Schmitt trigger was the answer! The optical sensor was entirely up to the job (reflective mode) once the rise and fall times were sharpened up. From a cold start the televisor will lock up in under 5-seconds. There was a very slow drift with respect to framing, but that was tuned out with the trimmer capacitor in the 3.2768 MHz oscillator circuit. The televisor has been running for over an hour now and it is evident that it will work very well for talks and demonstrations.

Thanks for the advice guys....


PostPosted: Sat May 04, 2013 6:17 pm
by Steve Anderson
Good news indeed! I must admit I have been suspicious of the detectors in the past, but having not built such a circuit wasn't able to confirm or otherwise my thoughts.

There are examples (admittedly few) of these detectors with the Schmitt-trigger built-in and a standard logic output. However all examples I can find are either 5V or 3.3V, not 12. Though that's a simple transistor buffer to get around that problem.

Steve A.