Maker Pro
Maker Pro

Position measurement.

  • Thread starter Dirk Bruere at Neopax
  • Start date
D

Dirk Bruere at Neopax

Jan 1, 1970
0
I'm looking for a small(ish) non contact solution to measure an oscillating
surface to within a few tens of microns over a stroke of some 50mm at a
frequency of around 100Hz. I need to take measurements at 100uS intervals. Any
ideas? Cost as always is a factor so anything under (say) $300 in mass production...

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org
 
I'm looking for a small(ish) non contact solution to measure an oscillating
surface to within a few tens of microns over a stroke of some 50mm at a

Angled laser, optics and a CMOS image sensor?
 
T

Tim Wescott

Jan 1, 1970
0
Dirk said:
I'm looking for a small(ish) non contact solution to measure an
oscillating surface to within a few tens of microns over a stroke of
some 50mm at a frequency of around 100Hz. I need to take measurements at
100uS intervals. Any ideas? Cost as always is a factor so anything under
(say) $300 in mass production...
- Better than 1ppm accuracy
- Better than 100Hz bandwidth
- Measurements at 10kHz

Sure!

- price < $300

Oh dear.

Keyance makes a non-contact laser distance measuring instrument that
gets your 10um accuracy over less than 1mm stroke with about 1/10th your
rep rate and your full bandwidth -- they charge about $10k.

Mostly I mention the Keyance's product because you may be able to copy
some of it's salient features for your application.

While you're asking your question, you may be able to significantly
lower the price of your solution if

- You have control over the surface, and can either prepare it in a
certain way or (better) embed something into it to make the job easier.
- The oscillating part is significantly less than 50mm stroke (I'm
thinking a 2-part sensor like a capacitive or inductive proximity sensor
on the end of a movable probe).
 
D

Dirk Bruere at Neopax

Jan 1, 1970
0
Angled laser, optics and a CMOS image sensor?

I'm thinking of that.
Maybe some kind of interferometer setup with either fringe counting or convert
the fringe rate to a velocity measurement.

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org
 
D

Dirk Bruere at Neopax

Jan 1, 1970
0
Tim said:
- Better than 1ppm accuracy
- Better than 100Hz bandwidth
- Measurements at 10kHz

Sure!

- price < $300

Oh dear.

Keyance makes a non-contact laser distance measuring instrument that
gets your 10um accuracy over less than 1mm stroke with about 1/10th your
rep rate and your full bandwidth -- they charge about $10k.

Mostly I mention the Keyance's product because you may be able to copy
some of it's salient features for your application.

While you're asking your question, you may be able to significantly
lower the price of your solution if

- You have control over the surface, and can either prepare it in a
certain way or (better) embed something into it to make the job easier.
- The oscillating part is significantly less than 50mm stroke (I'm
thinking a 2-part sensor like a capacitive or inductive proximity sensor
on the end of a movable probe).

I'm wondering whether I can get away with measuring velocity.
Still, this is some time away as a project so I'm just scouting round for
options. Maybe someone will come up with a really clever idea I have not even
thought of.

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org
 
I

Ian Stirling

Jan 1, 1970
0
Dirk Bruere at Neopax said:
I'm looking for a small(ish) non contact solution to measure an oscillating
surface to within a few tens of microns over a stroke of some 50mm at a
frequency of around 100Hz. I need to take measurements at 100uS intervals. Any
ideas? Cost as always is a factor so anything under (say) $300 in mass production...

Is the stroke strictly one-dimensional?
 
I can get you the resolution and about half the stroke, using a
capacitive measurement.
Our prototype here at the University is a two opamp triangle generator,
a 25 mm diameter pair of disks to make the capacitor , a one opamp
preamp/clipper, a two opamp precision rectifier and a passive low pass
filter. Its crude but we can see distance changes on the order of 5 nM
with it. A properly shaped linear capacitor could do it with our
circuit, which is by no means a origional concept. Adding a
synchronous rectifer as a sort of rude lock-in would really improve
things but we havent got around to that yet. Would you like a scanned
diagram emailed to you?

Steve Roberts
Research Associate,
Polymer Science, Akron U.
 
J

John Perry

Jan 1, 1970
0
Tim said:
- Better than 1ppm accuracy

50mm / 50um = 1000

He needs around 0.1% accuracy. Does that make it any easier?

jp
 
T

Tim Wescott

Jan 1, 1970
0
John said:
50mm / 50um = 1000

He needs around 0.1% accuracy. Does that make it any easier?

jp

Dangit.

Why yes, it does.
 
D

Dirk Bruere at Neopax

Jan 1, 1970
0
John said:
50mm / 50um = 1000

He needs around 0.1% accuracy. Does that make it any easier?

Not really since I'd already done the calcs for a 12 bit converter.

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org
 
J

Jonathan Kirwan

Jan 1, 1970
0
50mm / 50um = 1000

He needs around 0.1% accuracy. Does that make it any easier?

I don't even see accuracy requested. It might just be precision and
repeatability, which could be calibrated for accuracy if needed (but
not necessarily.)

Jon
 
D

Dirk Bruere at Neopax

Jan 1, 1970
0
I can get you the resolution and about half the stroke, using a
capacitive measurement.
Our prototype here at the University is a two opamp triangle generator,
a 25 mm diameter pair of disks to make the capacitor , a one opamp
preamp/clipper, a two opamp precision rectifier and a passive low pass
filter. Its crude but we can see distance changes on the order of 5 nM
with it. A properly shaped linear capacitor could do it with our
circuit, which is by no means a origional concept. Adding a
synchronous rectifer as a sort of rude lock-in would really improve
things but we havent got around to that yet. Would you like a scanned
diagram emailed to you?

Steve Roberts
Research Associate,
Polymer Science, Akron U.

That's very impressive.
I had thought about capacitative measurement but the distance seemed too great.
How would increasing the dia of the disk affect the spacing limits?

Also, this has given me a different idea. Would it be possible to put a thin
film electret on the device and sense its proximity with the plate?

Anyway, I would definately like to see the circuit as I have other position
measurement stuff to do that may be related. Email as above.

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org
 
F

Frank Bemelman

Jan 1, 1970
0
Dirk Bruere at Neopax said:
I'm looking for a small(ish) non contact solution to measure an oscillating
surface to within a few tens of microns over a stroke of some 50mm at a
frequency of around 100Hz. I need to take measurements at 100uS intervals. Any
ideas? Cost as always is a factor so anything under (say) $300 in mass
production...

Total travel per second = 100 * 100 => 10.000 mm. Travel between
100uS samples => 1mm. Accuracy 0.02 mm while object is moving...
You say 'tens' not 'tenths'. I assume 'tens' not 'tenths'.

Doesn't sound as a real time control of something, but just a
recording/measurement apparatus. If it were a real time control,
the slightest delay in processing could cause problems.

I'd try a linescan camera with ((1/0.02) * 50) => 2500 pixels,
perhaps 2048 or 4192 pixels. Don't know if you can get those
for <$300. Optics could be a simple pinhole, no lens needed.
 
a bigger set of disks increases sensitivity.
could also be used in bridge mode with a little thinking to see your
electret.
parts count = LM412 X 1 Lm318 X 1
2 diodes 7 caps 12 Rs 1 pot runs off +-12 V

we use a 40 Khz 5V p/p triangle as excitation

NASA used to hand out data in NASA TECH BRIEFS on something they
called capaciflector which was a scaled up multiple electrode sensor
that could have one side "blind" as you need,

Steve
 
D

Dirk Bruere at Neopax

Jan 1, 1970
0
Frank said:
production...

Total travel per second = 100 * 100 => 10.000 mm. Travel between
100uS samples => 1mm. Accuracy 0.02 mm while object is moving...
You say 'tens' not 'tenths'. I assume 'tens' not 'tenths'.

Doesn't sound as a real time control of something, but just a
recording/measurement apparatus. If it were a real time control,
the slightest delay in processing could cause problems.

I'd try a linescan camera with ((1/0.02) * 50) => 2500 pixels,
perhaps 2048 or 4192 pixels. Don't know if you can get those
for <$300. Optics could be a simple pinhole, no lens needed.

Thanks - that is definately something I have not considered.

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org
 
Top