Accelerometers for accurately indicating position?

[Forty_TuT]

maybe accelerometers aren't what i'm looking for but I'm hoping someone much more knowledgable than me can tell me if my idea is bogus or not.

I recently purchased a milling machine and need to set up Digital Readouts for the x and y. I was reading through a forum about installation and someone brought up a product called touchDRO which is readout display software that can apparently be downloaded onto tablet devices. This made me wonder if I could get around installing these large Linear Scales on my machine and instead maybe a single module on the table that can give me both at the same time. This brings me to this topic.

- Can accelerometers be used to measure distance traveled?
- If so, can a device using an accelerometer like this be repeatable, zero-able, and accurate to thousandths of inches (or hundredths of millimeters i think)?
- Is an accelerometer not what I'm looking for?
- Is there a product that can already do this?
- IS there a way to do this?

my understanding is a smart phone like iPhone can determine its position in 3D space?

hope that all made sense...
thanks!

 

[Bluejets]

Touch DRO uses encoder scales such as glass scales and others.
I think your understanding of how it works might be about 10 miles left of centre.

 

[OBW0549]

- Can accelerometers be used to measure distance traveled?

In principal, yes; it's done by processing the accelerometer's output through an integrator to obtain a signal proportional to velocity, and then passing that signal into another integrator to obtain a signal proportional to displacement.

In practice, accuracy and drift with time are major limitations.

- If so, can a device using an accelerometer like this be repeatable, zero-able, and accurate to thousandths of inches (or hundredths of millimeters i think)?

No.

 

[Nanren888]

In principle yes. In practice: no.
.
Acceleration integrated/accumulated is velocity.
Velocity intergrated/accumulated is distance.
Any DC offset at either stage integrates/accumulates to either a increasing velocity error, or an increasing distance error.
.
So zero-ing not really possible over any significant time.
The error relates to the offset, the time and the values.
Absolute positioners, difficult to avoid.
.
You can mostly get round the problem with something like a smart phone, by using the camera and preferable refenece points it can watch go by. More difficult than including a method of reading the absolute position.

 

[hevans1944]

For a milling machine, you need something that is accurate and repeatable. Linear (optical or magnetic) tape "scales" interfaced with electronic up-down counters come in many varieties and are recommended for the precision you need. Ask the vendor of your mill if they offer a retrofit readout

If you need better, or if a retrofit is not available or practical, there is the plane-mirror interferometer that Hewlett-Packard (and perhaps others) made. These were quite pricey but only plane mirrors attached to the moving table, along with some H-P beam-splitting cubes and H-P photo-detectors and a HeNe laser (all mounted external to the moving table) are needed. The output of the interferometer is a series of up or down pulses representing incremental displacements, at 1/8 wavelength intervals, IIRC. You need FAST counters to keep up and not lose any pulses if the mirrors are attached to the moving bed of a milling machine. Some used equipment is available on eBay, but I have no idea how much effort would be required to adapt it to your mill.

I used one of these H-P plane-mirror interferometer systems on a Mann-Data microdensitometer "upgrade" in the 1980s, interfacing the interferometer up-down pulse outputs to TTL counters installed on two complex, wire-wrapped, full-length prototyping boards inside a PC-AT personal computer. You probably don't need, nor can afford, this level of complexity but it worked like gang-busters. The PC loaded and read registers as 32-bit "words" on the interface and received high-priority interrupts when the stage had moved the desired distance. Meanwhile, smaller displacements were measured and used to generate A/D converter sampling commands for a photo-multiplier tube (PMT) "looking" at a small spot of light projected on a reconnaissance film transparency. At the end of each scan line, the buffered-in-RAM PMT data was read, formatted, and written out to a 9-track magnetic tape drive. Meanwhile, the interface was loaded with a step-over displacement value and when the stage reached that position it automatically resumed the scan and the data collection in the opposite direction, thus implementing a serpentine raster scan with minimal supervision by the PC.

H-P had a digital read-out module available of course, using the HPIB protocol to interface to a PC, but I determined that it would be waaay too slow for this application. The idea was to manually locate areas of interest on the film and then slew over to those areas and scan them. With our position read-outs displayed on the PC, this could have been a simple matter to program. Perhaps a joy-stick to move the optical stage to an area of interest, press a button and let the PC calculate the parameters defining the area of interest. Move on to the next area of interest, wash, rinse, repeat until all had been identified, then let 'er rip!

All this has now been replaced by digital imaging technology, although some universities still hang on to their microdensitometers.