Maker Pro
Maker Pro

Temperature compensated current measurement

C

Christian Walter

Jan 1, 1970
0
Hello,

I am currently looking for a way to measure and control the current
through a resistive load. I found same standard circuits often used for
voltage/current conversion but all of them are temperature sensitive
because they simply measure the current through an additional sense
resistor. This does not provide the desired accuracy over the entire
temperature range. For example if I use a 50ppm resistor and a resistive
value of 100R I have about 0.25R over a temperature range of 100K. This
is 0,25% which allows maybe 8-9Bits of resolution.
Does anybody know a good solution for this problem or can provide some
hints on that?

Kind regards,
Christian Walter
 
P

Phil Allison

Jan 1, 1970
0
"Christian Walter"
I am currently looking for a way to measure and control the current
through a resistive load. I found same standard circuits often used for
voltage/current conversion but all of them are temperature sensitive
because they simply measure the current through an additional sense
resistor. This does not provide the desired accuracy over the entire
temperature range. For example if I use a 50ppm resistor and a resistive
value of 100R I have about 0.25R over a temperature range of 100K. This is
0,25% which allows maybe 8-9Bits of resolution.
Does anybody know a good solution for this problem or can provide some
hints on that?


** Easy - don't let the resistor get hot.

Use a wire wound type ( typical tempco +/- 30ppm), mount it on a heatsink.

If its temp don't change, neither will its resistance.

Surely you are not subjecting the whole circuit to temps of 120 C ???



......... Phil
 
L

linnix

Jan 1, 1970
0
Hello,

I am currently looking for a way to measure and control the current
through a resistive load. I found same standard circuits often used for
voltage/current conversion but all of them are temperature sensitive
because they simply measure the current through an additional sense
resistor. This does not provide the desired accuracy over the entire
temperature range. For example if I use a 50ppm resistor and a resistive
value of 100R I have about 0.25R over a temperature range of 100K. This
is 0,25% which allows maybe 8-9Bits of resolution.
Does anybody know a good solution for this problem or can provide some
hints on that?

Kind regards,
Christian Walter

What's wrong with temperature compensating the data? Anything is
possible with digital data.
 
Hello,

I am currently looking for a way to measure and control the current
through a resistive load. I found same standard circuits often used for
voltage/current conversion but all of them are temperature sensitive
because they simply measure the current through an additional sense
resistor. This does not provide the desired accuracy over the entire
temperature range. For example if I use a 50ppm resistor and a resistive
value of 100R I have about 0.25R over a temperature range of 100K. This
is 0,25% which allows maybe 8-9Bits of resolution.
Does anybody know a good solution for this problem or can provide some
hints on that?

The answer is fairly simple - spend more money on your resistor.

Farnell stocks a number of Vishay bulk foil resistors offering
temperature coefficients of +/-0.2ppm per degree Celcius. Look at the
100R 0805 part (order code 110-924) the 1206 part (110-9037) or the
through hole part 120-3343 - the last only offers a temperature
coefficient that is less than 0.6/ppm per degree Celcius form 0 to 60
Celcius. They cost around 15 euro (say $20) each.

Farnell also stocks several 15ppm metal film resistor - the MMA0204
range has a 100R part (order code 308-5703) which is very cheap, while
their E96 ranges of axial lead resistors include a couple of 100R
parts for about one euro each.

If you are doing seriously precise work you could go directly to
Vishay or other specialist suppliers to get four-leaded (Kelvin
connection) resistors.
 
D

D from BC

Jan 1, 1970
0
Hello,

I am currently looking for a way to measure and control the current
through a resistive load. I found same standard circuits often used for
voltage/current conversion but all of them are temperature sensitive
because they simply measure the current through an additional sense
resistor. This does not provide the desired accuracy over the entire
temperature range. For example if I use a 50ppm resistor and a resistive
value of 100R I have about 0.25R over a temperature range of 100K. This
is 0,25% which allows maybe 8-9Bits of resolution.
Does anybody know a good solution for this problem or can provide some
hints on that?

Kind regards,
Christian Walter

One time I made a current sensing resistor so small*, I got worried
about noise from amplification.

*To reduce self-heating, to reduce R drift.


D from BC
British Columbia
Canada.
 
C

Christian Walter

Jan 1, 1970
0
Phil said:
"Christian Walter"


** Easy - don't let the resistor get hot.

Use a wire wound type ( typical tempco +/- 30ppm), mount it on a heatsink.

If its temp don't change, neither will its resistance.

Surely you are not subjecting the whole circuit to temps of 120 C ???

I care more about the environment. The power dissipated by the resistor
is very small. Certainly less than 125mW. Of course I don't want to
subject the circuit to 120C. And you are right. If I calibrate the
device at 20C the whole spawn is only 0.125R if I go up to 70 and down
to -30 using R = R0*(1+alpha*delta_T).

Regards,
Christian
 
P

Phil Allison

Jan 1, 1970
0
"Christian Walter"
I care more about the environment. The power dissipated by the resistor is
very small. Certainly less than 125mW. Of course I don't want to subject
the circuit to 120C. And you are right. If I calibrate the device at 20C
the whole spawn is only 0.125R if I go up to 70 and down to -30 using R =
R0*(1+alpha*delta_T).


** I think you have misread the spec for resistor tempco.

The figures are all " +/- xx ppm " = not just plus or minus.

You would do well to get some sample components and see just how the value
varies over your required temp range.

The actual variation may well be no where near as bad as you assume.




........ Phil
 

neon

Oct 21, 2006
1,325
Joined
Oct 21, 2006
Messages
1,325
An lm117 regulator can do that connect it as a current source. must remenber that temperatures do not rise or fall quickly.
 
J

Joerg

Jan 1, 1970
0
Christian said:
I care more about the environment. The power dissipated by the resistor
is very small. Certainly less than 125mW. Of course I don't want to
subject the circuit to 120C. And you are right. If I calibrate the
device at 20C the whole spawn is only 0.125R if I go up to 70 and down
to -30 using R = R0*(1+alpha*delta_T).

If it is really, really critical you could pick the lowest tempco
resistor money can buy and then peltier-cool that. But be prepared to
shell out close to $100 if you want a TE controller module that is ready
to use (meaning no design work there) out of the box.
 
Top