Optical Sensor Speed

[Moses]

From my reading here I thought that an optical sensor would basically not have any speed issues. But now that I have finally fully assembled my first winder, I can only get to speeds around 1k RPM. This is an estimate based on treating revolutions like BPM - I watched the counter and tapped my foot every time the tens changed. My interuptor disc is a CD painted black with maybe an inch cut out the circumference. It is not a perfect plane, so it's position inside the sensor's arms changes, but it never comes into contact. Now, that speed is really probably fast enough, but my winder is a vacuum with a light dimmer control, and my control over speed is not exactly amazing. I can get by, but I was wondering if anyone had any suggestions on improving performance. I can get (bad) pics if needed.

[Joe Gwinn]

Search the forum on "duty cycle". The cutout should be about 50% of a turn.

[Moses]

00

I was under the nonsensical impression that optical sensors didn't suffer from the duty cycle problem, or at least that they were just fast enough so that it was negligible. Now that I think about it, though, is it really a problem with the sensor or the counter? Thought that's not important. That's an easy fix. I figured my cutout would be enough after looking at pictures of Red House's winder here.

[Berman]

This is not a sensor problem, Its a problem of the counter (nearly each counter)
If you modify the disk so it has a larger opening it will probably the solution of your problem.
But think: there are many counters in the market, that can only count at quite low speeds. in the datasheet of your counter you can find the value of the maximun counting speed (often in Hz).
Example:
If it is 30 Hz, the counter can count 30 x 60 = 1800 turns per minute at a duty cycle of 50% (half dark, half light).

And one thing, many plastics are transparent for infrared light, but not for visual light. A good material for index disks is good old aluminium with black color.

I hope it helps ab bit.

Hermann

[Moses]

Well, like I said, I've painted the CD with black latex paint, I'm hoping that that is good enough. I've also cut away about half of the CD, if I accidentally cut off a little too much, resulting in more than 50%, it won't be a problem, will it? I really wasn't too careful, but it looks pretty good. Now that I've done that, it goes up to "Way too fast, rattling the entire workbench it's clamped to" speed, and as far as I can tell isn't missing any counts.

[David King]

Start with a fresh CD disc and just paint 1/2 of it, as long as the light can get through the clear part without reflecting off you'll be all set and have a balanced shaft that won't jump around.

[Magnut]

They always throw an "uncoated" cd on the top of the stack if you buy a container with 50 or 100. That would be a good one to use. You would not have to remove any of it and that would help with the rotational balance.

[RedHouse]

I was under the nonsensical impression that optical sensors didn't suffer from the duty cycle problem, or at least that they were just fast enough so that it was negligible. Now that I think about it, though, is it really a problem with the sensor or the counter? Thought that's not important. That's an easy fix. I figured my cutout would be enough after looking at pictures of Red House's winder here.

What counter are you using?

[Moses]

What counter are you using?

Cub 3.

[Joe Gwinn]

Well, like I said, I've painted the CD with black latex paint, I'm hoping that that is good enough. I've also cut away about half of the CD, if I accidentally cut off a little too much, resulting in more than 50%, it won't be a problem, will it? I really wasn't too careful, but it looks pretty good. Now that I've done that, it goes up to "Way too fast, rattling the entire workbench it's clamped to" speed, and as far as I can tell isn't missing any counts.

It's not a "problem" - counters like the CUB3 are designed that way so they will ignore contact bounce when driven by mechanical switches.

The 50% need not be exact.

As others have suggested, using an intact disk with black paint on half will work, and will remain dynamically balanced.

[RedHouse]

Cub 3.

That's the problem, pulse width is a big factor in the CUB3.
(see data sheet)

Next time get a CUB4 or CUB5, they trigger on the rising (or falling) edge of the input signal and have a alternate high speed input that can count up to 5kHz which is a theoretical limit of 300,000 RPM. Duty-cycle isn't a problem/issue for these counters, and you can get 'em fairly cheap off the bay.

(Note: not the CUB4L8 which has different input circuitry)