Maker Pro
Maker Pro

Linear position transducer?

M

mikem

Jan 1, 1970
0
I need to measure the position of an object which moves slowly (< 1Hz)
along one axis. Max travel is about 6 inches. I want to read its
position to a resolution about 1 part out of 1024 (10 bits), but the
absolute accuracy only needs to be 1%. I can mechanically attach
something like a pushrod to the object.

I'm familiar with LVDTs and the old NE5521 chip, but dont want to
spend a $1000 for a 6" LVDT.

Anybody got any ideas?

MikeM
 
F

Frank Bemelman

Jan 1, 1970
0
mikem said:
I need to measure the position of an object which moves slowly (< 1Hz)
along one axis. Max travel is about 6 inches. I want to read its
position to a resolution about 1 part out of 1024 (10 bits), but the
absolute accuracy only needs to be 1%. I can mechanically attach
something like a pushrod to the object.

I'm familiar with LVDTs and the old NE5521 chip, but dont want to
spend a $1000 for a 6" LVDT.

Anybody got any ideas?

Vishay (and no doubt others) has linear displacement sensors, basically
just a potentiometer. $100 perhaps.

I wonder what laser displacement sensors cost these days (like the ones
from Keyence).
 
S

Spehro Pefhany

Jan 1, 1970
0
I need to measure the position of an object which moves slowly (< 1Hz)
along one axis. Max travel is about 6 inches. I want to read its
position to a resolution about 1 part out of 1024 (10 bits), but the
absolute accuracy only needs to be 1%. I can mechanically attach
something like a pushrod to the object.

I'm familiar with LVDTs and the old NE5521 chip, but dont want to
spend a $1000 for a 6" LVDT.

How much life do you need? A linear conductive plastic pot might do
the trick.

Best regards,
Spehro Pefhany
 
M

Michael Gray

Jan 1, 1970
0
I need to measure the position of an object which moves slowly (< 1Hz)
along one axis. Max travel is about 6 inches. I want to read its
position to a resolution about 1 part out of 1024 (10 bits), but the
absolute accuracy only needs to be 1%. I can mechanically attach
something like a pushrod to the object.

I'm familiar with LVDTs and the old NE5521 chip, but dont want to
spend a $1000 for a 6" LVDT.

Anybody got any ideas?

MikeM

Just a thought:

How about using a scrapped HP inkjet printer?
They have an optical position sensor on the print carriage, that reads
a plastic tape with tiny binary markings photographically etched on
it.
They are very accurate, more than long enough, and don't wear out.
I'm not sure what code the sensor outputs the position in, but a
little experimentation would tell you that.
 
R

Rob

Jan 1, 1970
0
mikem said:
I need to measure the position of an object which moves slowly (< 1Hz)
along one axis. Max travel is about 6 inches. I want to read its
position to a resolution about 1 part out of 1024 (10 bits), but the
absolute accuracy only needs to be 1%. I can mechanically attach
something like a pushrod to the object.

I'm familiar with LVDTs and the old NE5521 chip, but dont want to
spend a $1000 for a 6" LVDT.

Anybody got any ideas?

MikeM

Servo pot with a machined pulley & a return "clock" sping.
Use light gauge nylon coated fine stranded stainless steel fishing trace
wire to go from the pulley to the moving structure.

Have fun
rob
 
F

Frank Bemelman

Jan 1, 1970
0
Michael Gray said:
Just a thought:

How about using a scrapped HP inkjet printer?
They have an optical position sensor on the print carriage, that reads
a plastic tape with tiny binary markings photographically etched on
it.
They are very accurate, more than long enough, and don't wear out.
I'm not sure what code the sensor outputs the position in, but a
little experimentation would tell you that.

It's Gray code! ;-)
 
R

Roger Hamlett

Jan 1, 1970
0
Frank Bemelman said:
Vishay (and no doubt others) has linear displacement sensors, basically
just a potentiometer. $100 perhaps.

I wonder what laser displacement sensors cost these days (like the ones
from Keyence).
A couple of other ideas.
If the motion is very slow/infrequent, consider a digital caliper. The
plastic versions of these are available in the UK, for under £20, and give
resolutions in the order of 1thou. Most designs don't draw power from (or
need) their internal batteries if connected to an external data cable. These
have four connections for power, and serial data. The sensor on these is a
strip of PCB, cut to form a 'tooth' pattern, which couples to a second
similar strip inside the sensor head, with the tooth pattern at a slightly
different spacing. The strips don't have to touch, so in fact the head can
be positioned over a different sensor strip, and arranged so there is no
contact at all. For the cost, they are unbeatable.
Another alternative, is to use a mouse. Provided you have an ability to
'resync' at intervals (a limit switch perhaps), you can either use a normal
mouse ball, or connect a drive directly to the optical sensor spindle. Again
the price (even with some added hardware), is the real 'feature'.
Third alternative, is to use a optical encoder strip. Honeywell make the
quadrature encoders used inside some mice, as a complete assembly. This has
the pair of decoders, and the quadrature decoder itself, to give a pair of
'step/direction' signals, from a slotted strip at the right interval passing
through the decoder. Look at 'hlc2705'. These are under £2, in 'one off'
quantities. The 'strip' to operate these sensors, can be made on plastic
film designed for printing slides for overhead projectors on laser printers.
I have printed such strips in the past, using a basic CAD package to make a
sheet of the lines, and printing an entire A4 sheet, then cutting strips
about 1/4" wide off this. I usually end up throwing away a couple of strips
with visible 'faults' (dust etc.). The entire system, then becomes the
optical detector, a suitable IR LED, with a gap between, and the strip
mounted onto the moving components (I usually 'hard mount' the strip at one
end, and pull the other with a rubber band or spring). Total cost is well
under £10. Resolution is in the order of 0.5mm with the system as described
(less than you need). However you can 'cheat', and improve the resolution on
this (provided the system that prints the sensor strip is accurate enough).
What you do, is instead of printing a 'linear' strip, you print the lines
angled. So, instead of a strip like:
||||||||||

you generate a strip with the lines like:
//////////

Then if the sensor sits parallel to the lines, you get an increased
positional signal change as the strip moves. At 30 degrees, you can get
resolutions of about 0.25mm. This approaches the sort of accuracy you need,
and for the price, might well be worth considering. :)

Best Wishes
 
F

Fred Abse

Jan 1, 1970
0
It's Gray code! ;-)

I've just looked at one on my HP printer, and it isn't Gray code, just
full width stripes. I assume there must be two photosomethings displaced
by 1/4 line pitch to read it, giving sin and cos, hence direction.

Plain ol' dumb incremental.
 
Top