DL06 Analog Sensor Design -> Accelerometer -> Sensitivity - OT?

Hi everyone.

I've got a DL06 with the 08ADH-2 8-channel analog input. I have that option card because I have to determine the angle of a carriage. To do that, I've built an acceleromter-based sensor that has been configured as a tilt sensor.

The code works well, and I get pretty good results. The voltage out is nice and smooth, and the results are within the right range. The oscilloscope shows a nice flat output voltage with a 0-5V range. This is paired with the 0-5V jumper on the ADH card.The PLC code takes the value, converts it into a +/- 45 degree value and displays it on the LCD panel. It works great. (TIP: When building a sensor like this, use a lot of EM shielding and nice filter caps. It makes an enourmous difference.)

Then I wake up and take the sensor out of ideal conditions. (banging the table, slamming desk drawers, running machines next to it, etc.) The sensor is sensitive enough to pick up the vibrations and give erroneous values.

The PLC converts the values every 200ms. I've tried the analog filter Iboxes from the DS5 package and/or the filtering functions listed in other threads, but they seem to make the problem worse, not better. (And yes, I've played with the numbers.) Changing the sensor design has allowed me to make it twice as sensitive to vibration, but that's exactly the opposite of what I was looking for.

I built a little anti-vibration cage to mitigate the problem, but that's only taking the edge off.

What can I do to make the sensor less sensitive to vibration? Is an accelerometer even the right tool for the job? I only require 1 degree accuracy on the angle measurement.

Thank you.

How about using a pot, or maybe an encoder instead...if you hung an excentric weight on the encoder shaft gravity would help you out...and if it is too sensitive,you could add a "dampener" like a felt or foam washer between the weight and the encoder case...


Happy New Year!!!!

David

Further to what I posted yesterday, I found a few things while going thru some reading materials today...

Heidenhain has a cool unit: www.heidenhain.com , unit # ERP 4080.

Admotec also has the KXL and KXS units : www.admotec.com

David

Well, that's a pretty good suggestion.

The problem is that the structure is a carriage on a gangway on a pier, so the tides will change the angle when the motor is off - that means the angle is going to change when I'm not changing it. From what I understand, that means a recalibration every time the tide changes.

That's why I went with an accelerometer. The angle has to be known absolutely, without a relative home position - since any given "home" position will change throughout the day.

Sorry, I didn't mean to pose a trick question.

MASEngr said:

That's why I went with an accelerometer. The angle has to be known absolutely, without a relative home position - since any given "home" position will change throughout the day.

Sorry, I didn't mean to pose a trick question.

Click to expand...

MASEngr,

How about a absolute or graycode encoder with a weight on a lever attached to the shaft? No home possition needed.

milldrone said:

MASEngr,

How about a absolute or graycode encoder with a weight on a lever attached to the shaft? No home possition needed.

Click to expand...

Accelerometers are MEMS (micro-electrical mechanical system) devices, so they're basically a weight on a lever attached to an encoder. It's just in a little SOIC-16 package.

For those who may care (probably just me ) I found a solution.

Short version: Adding a threshold into the programming fixed the jitter.


Long version: The application notes for the accelerometer had an application for a "tap" sensor for a pager. Basically, if you want it to stop buzzing, you can just smack it. Now, to get around the vibration setting off the tap sensor, it has to have a threshold.

So what I did was sketch out the schematic, feeding the output back to the comparator to get a changing threshold. The plan was to wire it backwards and signal to the PLC that the angle was valid in the absence of a shock.

Then I thought, "hey, I can write this in software instead of pulling out the station". So I just added a few lines of > and < to the PLC code to fix the problem. (Fixed like a cat!)

Now I can shake it to my heart's content and still read the angles. It's still quite buggy at this point, but that's the way I'm solving the problem...

...unless you think that'll make the papers.

Glad to hear that you were able to work it out...keep us informed as you go along ...

David