Unstable LM92 question

J

Jurgen Priem

Guest
Hi all,

I've been experimenting a while with an LM92 temperature sensor:
http://cache.national.com/ds/LM/LM92.pdf
And I'm a bit dissapointed, is it normal that a sensor with 0.33°c
accuracy and a resolution of 0.0625 °C has its lower 2 to 3 bits jumping
around? Resulting in a 0.2°C jumpy display? I would have thought it
would be more stable than that? Or am I missing something?
I have the lm92 connected as followed :

SDA & SCL with a 4k7 pull up to a PIC
T_CRIT and INT connected to a 10K resistor to V+
A0 and A1 to V+
V+ = 5v
22nF over V+ and GND


I'm trying to build a reasonably accurate ambient air sensor, i've tried
using an LM35 but gave this up because of the difficulty to get a good
linear amplification to boost the signal to a usable level for A/D and
thought a digital sensor would be the best option...
Anybody got a better idea for a sensor that has got a good accuracy, not
to pricy and has good availability or can fix my LM92 problem?

Thanks!
 
Mika Lindblad wrote:
On 2008-05-27, Jurgen Priem <jurgen.//NO//SPAM//priem@pandora.be> wrote:
I've been experimenting a while with an LM92 temperature sensor:
http://cache.national.com/ds/LM/LM92.pdf
And I'm a bit dissapointed, is it normal that a sensor with 0.33°c
accuracy and a resolution of 0.0625 °C has its lower 2 to 3 bits jumping
around? Resulting in a 0.2°C jumpy display? I would have thought it
would be more stable than that? Or am I missing something?
I have the lm92 connected as followed :

SDA & SCL with a 4k7 pull up to a PIC
T_CRIT and INT connected to a 10K resistor to V+
A0 and A1 to V+
V+ = 5v
22nF over V+ and GND

22nF capacitor for power supply decoupling might me insufficient, try 100nF
or 220nF instead.

Is +5V supply voltage stable and clean? Is the supply of microcontroller properly
decoupled? Poorly decoupled microcontroller could introduce disturbance to
supply voltage. You might try to add LC-filter (ferrite bead + small
capacitor) in the power supply of microcontroller and/or temperature sensor.


Decoupling the sensor with a series resistor from V+ to a big capacitor
Click to expand...
may also help, if the noise is high frequency enough. With lower
frequency noise a separate regulator to the temperature sensor V+ would
be a help.

I'd have a scope on that V+ line right at the sensor chip, to see how
noisy it is.

If worse comes to worst you could always oversample and average. It's a
band-aid, but if it works...

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" gives you just what it says.
See details at http://www.wescottdesign.com/actfes/actfes.html
 
Jurgen Priem wrote:
Hi all,

I've been experimenting a while with an LM92 temperature sensor:
http://cache.national.com/ds/LM/LM92.pdf
And I'm a bit dissapointed, is it normal that a sensor with 0.33°c
accuracy and a resolution of 0.0625 °C has its lower 2 to 3 bits jumping
around? Resulting in a 0.2°C jumpy display? I would have thought it
would be more stable than that? Or am I missing something?
I have the lm92 connected as followed :

SDA & SCL with a 4k7 pull up to a PIC
T_CRIT and INT connected to a 10K resistor to V+
A0 and A1 to V+
V+ = 5v
22nF over V+ and GND


I'm trying to build a reasonably accurate ambient air sensor, i've tried
using an LM35 but gave this up because of the difficulty to get a good
linear amplification to boost the signal to a usable level for A/D and
thought a digital sensor would be the best option...
Anybody got a better idea for a sensor that has got a good accuracy, not
to pricy and has good availability or can fix my LM92 problem?

Thanks!
why don't you simply average the readings. that will smooth it out.
Click to expand...


--
"I'd rather have a bottle in front of me than a frontal lobotomy"

"Daily Thought:

SOME PEOPLE ARE LIKE SLINKIES. NOT REALLY GOOD FOR ANYTHING BUT
THEY BRING A SMILE TO YOUR FACE WHEN PUSHED DOWN THE STAIRS.
http://webpages.charter.net/jamie_5"
 
Jurgen Priem wrote:
Hi all,

I've been experimenting a while with an LM92 temperature sensor:
http://cache.national.com/ds/LM/LM92.pdf
And I'm a bit dissapointed, is it normal that a sensor with 0.33°c
accuracy and a resolution of 0.0625 °C has its lower 2 to 3 bits jumping
around? Resulting in a 0.2°C jumpy display? I would have thought it
would be more stable than that? Or am I missing something?
I have the lm92 connected as followed :

SDA & SCL with a 4k7 pull up to a PIC
T_CRIT and INT connected to a 10K resistor to V+
A0 and A1 to V+
V+ = 5v
22nF over V+ and GND


I'm trying to build a reasonably accurate ambient air sensor, i've tried
using an LM35 but gave this up because of the difficulty to get a good
linear amplification to boost the signal to a usable level for A/D and
thought a digital sensor would be the best option...
Anybody got a better idea for a sensor that has got a good accuracy, not
to pricy and has good availability or can fix my LM92 problem?
Click to expand...
Maybe your domain is more in the digital area. If an amp was difficult
and you can't get it all quiet with the bypassing that the others have
suggested I'd advise to seek help from a consultant. Preferably local so
he/she could come in. Must be experienced in low-noise analog/mixed stuff.

Or just post schematic and layout.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
 
On 2008-05-27, Jurgen Priem <jurgen.//NO//SPAM//priem@pandora.be> wrote:
I've been experimenting a while with an LM92 temperature sensor:
http://cache.national.com/ds/LM/LM92.pdf
And I'm a bit dissapointed, is it normal that a sensor with 0.33°c
accuracy and a resolution of 0.0625 °C has its lower 2 to 3 bits jumping
around? Resulting in a 0.2°C jumpy display? I would have thought it
would be more stable than that? Or am I missing something?
I have the lm92 connected as followed :

SDA & SCL with a 4k7 pull up to a PIC
T_CRIT and INT connected to a 10K resistor to V+
A0 and A1 to V+
V+ = 5v
22nF over V+ and GND
Click to expand...
22nF capacitor for power supply decoupling might me insufficient, try 100nF
or 220nF instead.

Is +5V supply voltage stable and clean? Is the supply of microcontroller properly
decoupled? Poorly decoupled microcontroller could introduce disturbance to
supply voltage. You might try to add LC-filter (ferrite bead + small
capacitor) in the power supply of microcontroller and/or temperature sensor.
 
Jurgen Priem schreef:
Hi all,

I've been experimenting a while with an LM92 temperature sensor:
http://cache.national.com/ds/LM/LM92.pdf
And I'm a bit dissapointed, is it normal that a sensor with 0.33°c
accuracy and a resolution of 0.0625 °C has its lower 2 to 3 bits jumping
around? Resulting in a 0.2°C jumpy display? I would have thought it
would be more stable than that? Or am I missing something?
I have the lm92 connected as followed :

SDA & SCL with a 4k7 pull up to a PIC
T_CRIT and INT connected to a 10K resistor to V+
A0 and A1 to V+
V+ = 5v
22nF over V+ and GND


I'm trying to build a reasonably accurate ambient air sensor, i've tried
using an LM35 but gave this up because of the difficulty to get a good
linear amplification to boost the signal to a usable level for A/D and
thought a digital sensor would be the best option...
Anybody got a better idea for a sensor that has got a good accuracy, not
to pricy and has good availability or can fix my LM92 problem?

Thanks!
Click to expand...

Thanks all for the tips,
turned out that the bypass on the pic had a bad solder joint, also
changed the bypass over the LM92 to 100nF
Now I have a much more stable temp reading!
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top