DC - DC LVDT voltage comparison

N

news.cfl.rr.com

Guest
I need to deliver a logic signal based upon the comparison of two dc
voltages coming from a DC - DC LVDT (sheet
http://www.sensotec.com/pdf/ms3s3c.pdf ) . These inputs will be in the range
of +- 5V and are temperature stabilized and filtered. Since the discrete
movements being measured are ~ .00015", the device will have to be very
precise. Any suggestions would be appreciated.
 
"news.cfl.rr.com" wrote:
I need to deliver a logic signal based upon the comparison of two dc
voltages coming from a DC - DC LVDT (sheet
http://www.sensotec.com/pdf/ms3s3c.pdf ) . These inputs will be in the range
of +- 5V and are temperature stabilized and filtered. Since the discrete
movements being measured are ~ .00015", the device will have to be very
precise. Any suggestions would be appreciated.
Click to expand...
Keep in mind that the filtering used to reduce the noise and ripple
voltage exiting these units will slow their response to movement,
also. That filtering appears to be specified as high as 10 mv with a
10 volt output range. So even though the average output resolution is
infinite, there will be short term errors equivalent to .01/10=0.1% of
full scale. If your unit has a full scale stroke of 0.4 inch, this
noise represents a position error of about 0.0004 inch. There are
also steady state errors caused by temperature change and nonlinearity
that no amount of filtering can cure.
--
John Popelish
 
Thanks John. This system response time can be seconds and I can smooth any
output with a capacitor on the output. I am using this to check for backlash
in a ball screw. So I can command a movement of .00015 then wait for a
noiseless comparison then continue. Wouldn't this eliminate all problems
except the .25% non linearity within the LVDT specs?

"John Popelish" <jpopelish@rica.net> wrote in message
news:403E1873.6AF37BE4@rica.net...
"news.cfl.rr.com" wrote:

I need to deliver a logic signal based upon the comparison of two dc
voltages coming from a DC - DC LVDT (sheet
http://www.sensotec.com/pdf/ms3s3c.pdf ) . These inputs will be in the
range
of +- 5V and are temperature stabilized and filtered. Since the discrete
movements being measured are ~ .00015", the device will have to be very
precise. Any suggestions would be appreciated.

Keep in mind that the filtering used to reduce the noise and ripple
voltage exiting these units will slow their response to movement,
also. That filtering appears to be specified as high as 10 mv with a
10 volt output range. So even though the average output resolution is
infinite, there will be short term errors equivalent to .01/10=0.1% of
full scale. If your unit has a full scale stroke of 0.4 inch, this
noise represents a position error of about 0.0004 inch. There are
also steady state errors caused by temperature change and nonlinearity
that no amount of filtering can cure.
--
John Popelish
Click to expand...
 
I meant a switched capacitor on the output.

"kdoney" <kdoney@*earthlink.com> wrote in message
news:m5r%b.140226$jH.981930@twister.tampabay.rr.com...
Thanks John. This system response time can be seconds and I can smooth any
output with a capacitor on the output. I am using this to check for
backlash
in a ball screw. So I can command a movement of .00015 then wait for a
noiseless comparison then continue. Wouldn't this eliminate all problems
except the .25% non linearity within the LVDT specs?

"John Popelish" <jpopelish@rica.net> wrote in message
news:403E1873.6AF37BE4@rica.net...
"news.cfl.rr.com" wrote:

I need to deliver a logic signal based upon the comparison of two dc
voltages coming from a DC - DC LVDT (sheet
http://www.sensotec.com/pdf/ms3s3c.pdf ) . These inputs will be in the
range
of +- 5V and are temperature stabilized and filtered. Since the
discrete
movements being measured are ~ .00015", the device will have to be
very
precise. Any suggestions would be appreciated.

Keep in mind that the filtering used to reduce the noise and ripple
voltage exiting these units will slow their response to movement,
also. That filtering appears to be specified as high as 10 mv with a
10 volt output range. So even though the average output resolution is
infinite, there will be short term errors equivalent to .01/10=0.1% of
full scale. If your unit has a full scale stroke of 0.4 inch, this
noise represents a position error of about 0.0004 inch. There are
also steady state errors caused by temperature change and nonlinearity
that no amount of filtering can cure.
--
John Popelish
Click to expand...
 
On Thu, 26 Feb 2004 15:13:50 GMT, "news.cfl.rr.com" <kdoney*@cfl.rr.com>
wrote:

I need to deliver a logic signal based upon the comparison of two dc
voltages coming from a DC - DC LVDT (sheet
http://www.sensotec.com/pdf/ms3s3c.pdf ) . These inputs will be in the range
of +- 5V and are temperature stabilized and filtered. Since the discrete
movements being measured are ~ .00015", the device will have to be very
precise. Any suggestions would be appreciated.
Click to expand...
---
With a full-scale deflection of 0.4" and an uncertainty in the output of
+/- 0.017% of reading/°F, it seems to me that your error budget's going
to be shot in just over 2°F. Add to that the zero error of 0.006% of
full scale/°F and it'll be shot with less than a 1°F change in temp
unless you've got the thing in an oven or something, so it may be that
the accuracy/stability of the comparison circuitry is the least of your
problems at the moment.

Assuming that I'm wrong and/or that you've got that problem under
control, when you say "two dc voltages" I'm assuming that you mean you
want to know the location of the plunger at two points along its travel
and want to determine that by measuring the voltages which correspond to
those points. Right?

--
John Fields
 
kdoney wrote:
Thanks John. This system response time can be seconds and I can smooth any
output with a capacitor on the output. I am using this to check for backlash
in a ball screw. So I can command a movement of .00015 then wait for a
noiseless comparison then continue. Wouldn't this eliminate all problems
except the .25% non linearity within the LVDT specs?
Click to expand...
Right. It will help reduce the random noise and the ripple voltage on
the outputs but will not make the absolute measurement any more
accurate. Good for small changes, though.
--
John Popelish
 
When I said 2 DC voltages I was referring to the reference voltage (my first
reading from the LVDT) and the present voltage coming from the LVDT. I will
move the LVDT with a servo motor controlled ball screw until I get a
voltage. Then I will reverse the servo, one step at a time (.00015") until I
get a different voltage. This will give me the backlash. I really only use
the LVDT to detect movement. I don't mind if the voltages have errors since
I will be subtracting the reference voltage. The servo movement actually
gives me the measurement. If the temperature is going to change the reading
during testing beyond the difference that .00015" movement would make then
it won't work.

"John Fields" <jfields@austininstruments.com> wrote in message
news:79gs30t1fqel3bp18045jrgqoga65qe6ct@4ax.com...
On Thu, 26 Feb 2004 15:13:50 GMT, "news.cfl.rr.com" <kdoney*@cfl.rr.com
wrote:

I need to deliver a logic signal based upon the comparison of two dc
voltages coming from a DC - DC LVDT (sheet
http://www.sensotec.com/pdf/ms3s3c.pdf ) . These inputs will be in the
range
of +- 5V and are temperature stabilized and filtered. Since the discrete
movements being measured are ~ .00015", the device will have to be very
precise. Any suggestions would be appreciated.

---
With a full-scale deflection of 0.4" and an uncertainty in the output of
+/- 0.017% of reading/°F, it seems to me that your error budget's going
to be shot in just over 2°F. Add to that the zero error of 0.006% of
full scale/°F and it'll be shot with less than a 1°F change in temp
unless you've got the thing in an oven or something, so it may be that
the accuracy/stability of the comparison circuitry is the least of your
problems at the moment.

Assuming that I'm wrong and/or that you've got that problem under
control, when you say "two dc voltages" I'm assuming that you mean you
want to know the location of the plunger at two points along its travel
and want to determine that by measuring the voltages which correspond to
those points. Right?

--
John Fields
Click to expand...
 
kdoney wrote:
I meant a switched capacitor on the output.
Click to expand...
Not sure where you are going to switch the capacitor in. What total
travel LVDT are you using?

--
John Popelish
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top