Now that temperature sensors are discussed, does anybody have
experience with remote temperature sensing. With remote I mean
a temperature sensor with about 30m (100 feet) of wire attached
to it. On the other side of this wire there is an opamp with
some gain followed by a PIC A/D.
Temperature range I need is 0..50 degrees Celsius, +/- 1 degree
accuracy. I'm currently using a National LM35 Centigrade temp. sensor
with a three wire connection (+5V, GND and Vtemp). But I'm
experiencing
quite a linearity error.
well, you're trying to send a mV signal over a long wire. There are
two main problems you will face. First, the resistance of the 30m wire will
introduce an error, due to the fact that its resistance is not zero. Have
you tried measuring the output of the sensor and the input at the receiving
end? Second, and probably more serious, is that you are bound to get EMI
over your 30m wire.
You will need to condition the signal on the LM35 side, before it is
sent over the wire. Turning the mV signal into a mA signal would make your
sensor much more robust and less sensitive to EMI. You might get away with
sticking your existing op-amp on the sensor side, so that you at least have
a larger signal going over the wire. But IMO, I don't think you will get
very far with that.
At 07:10 PM 12/18/00 +0100, you wrote:
>Hi all,
>
>Now that temperature sensors are discussed, does anybody have
>experience with remote temperature sensing. With remote I mean
>a temperature sensor with about 30m (100 feet) of wire attached
>to it. On the other side of this wire there is an opamp with
>some gain followed by a PIC A/D.
>
>Temperature range I need is 0..50 degrees Celsius, +/- 1 degree
>accuracy. I'm currently using a National LM35 Centigrade temp. sensor
>with a three wire connection (+5V, GND and Vtemp). But I'm
>experiencing
>quite a linearity error.
Unh, the LM35 output is directly from an emitter follower.... Are you
using the recommended 2K series resistor into the cable (sensor end)?
You should also bypass the chip at the sensor end with something
like 0.1uF. I'd use an X7R ceramic capacitor.
Then you may need a buffer amplifier (with an input LPF) at the
receiving end.
Best regards,
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Spehro Pefhany --"it's the network..." "The Journey is the reward" speffKILLspaminterlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
Contributions invited->The AVR-gcc FAQ is at: http://www.bluecollarlinux.com
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
-- http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads
> At 07:10 PM 12/18/00 +0100, you wrote:
> >Hi all,
> >
> >Now that temperature sensors are discussed, does anybody have
> >experience with remote temperature sensing. With remote I mean
> >a temperature sensor with about 30m (100 feet) of wire attached
> >to it. On the other side of this wire there is an opamp with
> >some gain followed by a PIC A/D.
> >
> >Temperature range I need is 0..50 degrees Celsius, +/- 1 degree
> >accuracy. I'm currently using a National LM35 Centigrade temp.
sensor
> >with a three wire connection (+5V, GND and Vtemp). But I'm
> >experiencing
> >quite a linearity error.
>
> Unh, the LM35 output is directly from an emitter follower.... Are
you
> using the recommended 2K series resistor into the cable (sensor
end)?
> You should also bypass the chip at the sensor end with something
> like 0.1uF. I'd use an X7R ceramic capacitor.
>
> Then you may need a buffer amplifier (with an input LPF) at the
> receiving end.
Yep, did all that. Just like the datasheet said. However, I have to
correct my earlier question a bit. I'm not really experiencing a
linearity error, but a slope error. So I need to do some slope
compensation in sw.
The only thing that bugs me, is that National guarantees a linear
10mV/degr.Celsius output, and I'm not getting it from the device even
on the small temperature range I require.
Did you have better results with the LM35 Spehro?
Daniel...
-- http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads
>well, you're trying to send a mV signal over a long wire. There are
>two main problems you will face. First, the resistance of the 30m
wire
>will introduce an error, due to the fact that its resistance is not
zero.
That's true. But to increase the LM35's capability to drive capacitive
loads (=cable), the datasheet advises to add a 2k resistor in series
with the LM35 output at the sensor side. So I think the cable
resistance (2 or 3 Ohms to be exact) can be neglected. Or am I wrong
here?
>Have you tried measuring the output of the sensor and the input at
the
>receiving end?
Not yet. But the output impedance of the sensor shall be around 2600
Ohms. On the input side of the measuring unit, there's a 1k series
resistor followed by a 5V1 zener to gnd and a 100nF capacitor to gnd.
After that the 'cleaned' signal is fed to the non inverting input of
an LMV321 opamp with a gain of 8. The amplified temperature signal is
then routed via a 100 Ohm resistor to the 8 bit PIC A/D.
>Second, and probably more serious, is that you are bound
>to get EMI over your 30m wire.
Yes, that is really something I worry about. In any case I would like
to use twisted pair cable or shielded cable. And besides that I hope
the input filtering will do hf cancelling. Anymore ideas?
>You will need to condition the signal on the LM35 side, before it is
>sent over the wire. Turning the mV signal into a mA signal would make
>your sensor much more robust and less sensitive to EMI. You might get
>away with sticking your existing op-amp on the sensor side, so that
you
>at least have a larger signal going over the wire. But IMO, I don't
think
>you will get very far with that.
The LM35 datasheet has a remote temperature sensor example using a two
wire approach. I built this application, but the results were much
worser
than using the standard 3 wire interface. Pretty strange.
How cost sensitive is the project ? As another suggestion, what about using
a PIC12CXX with analog input to measure temperature at source and then send
digital value back to main unit. Depending on the space requirements, you
could then have the sensor very close to the interface and do away with some
of your signal conditioning and use RS485 to place your sensors up to
hundreds of meters away if desired.
Regards
David Huisman
-- http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads
> Now that temperature sensors are discussed, does anybody have
> experience with remote temperature sensing. With remote I mean
> a temperature sensor with about 30m (100 feet) of wire attached
> to it. On the other side of this wire there is an opamp with
> some gain followed by a PIC A/D.
My first reaction is to put the sensor, amp, and PIC at the end of the wire.
You feed it ground and +5, it sends back a serial bit stream right off the
PIC UART.
At 11:02 PM 12/18/00 +0100, D. Schouten wrote: {Quote hidden}
>Hi Phil,
>
>Thanks for your input!
>
> >well, you're trying to send a mV signal over a long wire. There are
> >two main problems you will face. First, the resistance of the 30m
>wire
> >will introduce an error, due to the fact that its resistance is not
>zero.
>
>That's true. But to increase the LM35's capability to drive capacitive
>loads (=cable), the datasheet advises to add a 2k resistor in series
>with the LM35 output at the sensor side. So I think the cable
>resistance (2 or 3 Ohms to be exact) can be neglected. Or am I wrong
>here?
>
> >Have you tried measuring the output of the sensor and the input at
>the
> >receiving end?
>
>Not yet. But the output impedance of the sensor shall be around 2600
>Ohms. On the input side of the measuring unit, there's a 1k series
>resistor followed by a 5V1 zener to gnd and a 100nF capacitor to gnd.
>After that the 'cleaned' signal is fed to the non inverting input of
>an LMV321 opamp with a gain of 8. The amplified temperature signal is
>then routed via a 100 Ohm resistor to the 8 bit PIC A/D.
If you are having slope errors with the LM35 *and* you have a 5V1 zener as
a clamp, try removing the zener and doing your tests again. You may be
surprised! For the record, I *never* use zener clamps on a/d inputs. If
adequate series resistance can't protect against over-voltage, I use
steering diodes that go to zener clamps. I have found that zener diodes,
especially low voltage zener diodes, simply have too much leakage
current. Worse, that leakage varies with both applied voltage as well as
temperature.
The setup I would use is as follows: At the LM35 end, I'd have a 100n mono
cap across V+ and GND and a 2k2 resistor in series with the output (before
it hits the cable). At the PIC end, I'd have a RC filter (10k input
resistor, 1 uF 35V tantalum cap) which feeds the PIC a/d input. Only 2
components at each end of the link cable. Give it a try!
> Now that temperature sensors are discussed, does anybody have
> experience with remote temperature sensing. With remote I mean
> a temperature sensor with about 30m (100 feet) of wire attached
> to it. On the other side of this wire there is an opamp with
> some gain followed by a PIC A/D.
>
> Temperature range I need is 0..50 degrees Celsius, +/- 1 degree
> accuracy. I'm currently using a National LM35 Centigrade temp. sensor
> with a three wire connection (+5V, GND and Vtemp). But I'm
> experiencing
> quite a linearity error.
>
> Anyone using the LM35 in remote applications too?
I have used the LM35 and have noticed the same linearity errors,
annoying. My solution? The LM75. It is an I2C chip, the temperature is read
and converted on site, a great way to solve the problem. 30m might be a
little longer than the I2C spec specifies, but it does work, just run it
slower and take multiple readings to verify results. TTYL
-- http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads
We use the LM34 together with a 4-20mA current loop driver over a
hundred feet with no problem. Particularly in an industrial environment
with a lot of interference a low impedance current loop will have an
advantage except for the additional cost on the current loop driver. We
use Burr brown XTR110 in this application. Again, don't forget the 75
ohms and the 1uF at the output, without it will cause several degrees
off.
If your problem persist, you can call National Semiconductor direct, I
heard that they are very helpful.
>
> Hi all,
>
> Now that temperature sensors are discussed, does anybody have
> experience with remote temperature sensing. With remote I mean
> a temperature sensor with about 30m (100 feet) of wire attached
> to it. On the other side of this wire there is an opamp with
> some gain followed by a PIC A/D.
>
> Temperature range I need is 0..50 degrees Celsius, +/- 1 degree
> accuracy. I'm currently using a National LM35 Centigrade temp. sensor
> with a three wire connection (+5V, GND and Vtemp). But I'm
> experiencing
> quite a linearity error.
>
> Anyone using the LM35 in remote applications too?
>
> Regards,
>
> Daniel..
>
> ----- Original Message -----
> From: Gary Tompkins <garyspam_OUTATRBIOTECH.COM>
> To: <@spam@PICLISTKILLspamMITVMA.MIT.EDU>
> Sent: Monday, December 18, 2000 15:52
> Subject: Re: [PIC]:Simple temperature sensors
>
The Maxim MAX6576 is a temperature sensor with frequency output proportional
to the temperature, 10us/K; if you connect a resistor 1.5K to VCC, the
power supply current will be so a square wave with a bias of 250uA and a
maximum value of 250uA+(Vcc-0.4V)/R. You can connect the sensor with only 2
wires and with the MAX4173F at the receiving side you will have TTL signal
or other levels dependig to a resistor.
__________________ _____________ 27
_____+5V
--!---
| -|------|--
| |
1.5K | |
|
|_____|
| |_____OUT TTL
> Hi, Daniel,
>
> We use the LM34 together with a 4-20mA current loop driver over a
> hundred feet with no problem. Particularly in an industrial environment
> with a lot of interference a low impedance current loop will have an
> advantage except for the additional cost on the current loop driver. We
> use Burr brown XTR110 in this application. Again, don't forget the 75
> ohms and the 1uF at the output, without it will cause several degrees
> off.
>
> If your problem persist, you can call National Semiconductor direct, I
> heard that they are very helpful.
>
> Eric
>
> "D. Schouten" wrote:
> >
> > Hi all,
> >
> > Now that temperature sensors are discussed, does anybody have
> > experience with remote temperature sensing. With remote I mean
> > a temperature sensor with about 30m (100 feet) of wire attached
> > to it. On the other side of this wire there is an opamp with
> > some gain followed by a PIC A/D.
> >
> > Temperature range I need is 0..50 degrees Celsius, +/- 1 degree
> > accuracy. I'm currently using a National LM35 Centigrade temp. sensor
> > with a three wire connection (+5V, GND and Vtemp). But I'm
> > experiencing
> > quite a linearity error.
> >
> > Anyone using the LM35 in remote applications too?
> >
> > Regards,
> >
> > Daniel..
> >
> > ----- Original Message -----
> > From: Gary Tompkins <KILLspamgaryKILLspamATRBIOTECH.COM>
> > To: <RemoveMEPICLISTTakeThisOuTMITVMA.MIT.EDU>
> > Sent: Monday, December 18, 2000 15:52
> > Subject: Re: [PIC]:Simple temperature sensors
> >
>Temperature range I need is 0..50 degrees Celsius, +/- 1 degree
>accuracy. I'm currently using a National LM35 Centigrade temp. sensor
>with a three wire connection (+5V, GND and Vtemp). But I'm
>experiencing
>quite a linearity error.
You may be better off using a device like the Analog Devices AD590 which is a 2
terminal current device. Its only requirement is to have a minimum of about 5
volts across it, and it will then have a current flow of 1uA/K, so 0C is 273uA.
This converts quite nicely into a usable voltage for A/D conversion with a 10k
resistor. The AD590 is available in various temperature tolerance selections,
the best being +/-0.5C. A 1nF capacitor across the terminals seems to keep most
RF induced problems away.
This device has the advantage that being a current operated device there is no
problem with "ground bounce" or "earth loop" type problems that 3 terminal
devices can have.
We also use the AD590 in a ground water monitoring application where the
sensor is on the end of a 100 metre cable. It works well.
Peter
>>Temperature range I need is 0..50 degrees Celsius, +/- 1 degree
>>accuracy. I'm currently using a National LM35 Centigrade temp. sensor
>>with a three wire connection (+5V, GND and Vtemp). But I'm
>>experiencing
>>quite a linearity error.
>
>You may be better off using a device like the Analog Devices AD590 which
is a 2
>terminal current device. Its only requirement is to have a minimum of about 5
>volts across it, and it will then have a current flow of 1uA/K, so 0C is
273uA.
>This converts quite nicely into a usable voltage for A/D conversion with a
10k
>resistor. The AD590 is available in various temperature tolerance selections,
>the best being +/-0.5C. A 1nF capacitor across the terminals seems to keep
most
>RF induced problems away.
>
>This device has the advantage that being a current operated device there
is no
>problem with "ground bounce" or "earth loop" type problems that 3 terminal
>devices can have.
>
>--
>http://www.piclist.com hint: The list server can filter out subtopics
>(like ads or off topics) for you. See http://www.piclist.com/#topics
>
I too used a 4-to-20 mA loop converter with a voltage-ouput temperature
sensor. For this project, I used a Nat Semi LM135H temp sensor and an
Analog Devices AD593AQ. The AD593 is pricey, about $22.00 in single
quantities, but it works really well and is perfect for the application
(electic utility substation SCADA I/O). If 4-to-20 mA loop power is not
required, I have used an Analog Devices AD590LH. It is a linear current
limiting device that allows 1 microamp per degree Kelvin to flow over the
range of -55 to 150 degrees Celcius, is accurate to +/- 1 degree Celcius,
and costs about $16. Between the sensor lead pair and between each lead
and a grounded shield connection, I connect a 0.1 microfarad capacitor in
parallel with a 30 volt bidirectional TVS for surge suppression. I have
run this across a substation yard for hundreds of feet with no trouble.
Surge suppression is my main concern in this case since there are large EMF
fields and ground potential rises whenever switching or faults occur. It
truly is a nasty environment. The sensors used to get popped before I
added the caps and TVS's.
In reply to some of the posters I would like to respond as follows :
David / Olin / Herbert :
Unfortunately the project is quite cost sensitive, however I don't
need to save -every- penny. The device to which the remote temp.
sensor needs to be connected, is a small PIC powered dashboard
measurement instrument to read out several parameters, including
remote ambient temperature from 0..50 degr. Celsius. The temperature
sensor will be sold as an option, so I don't want to spend too much
money on interface hardware in the readout device itself.
The option of implementing a 12Cxx with an RS485 transmitter is
probably the most ideal solution, but unfortunately too expensive.
Also I can't handle the hardware overhead needed in the readout device
(which is, unlike the temp. sensor, very cost sensitive) and am
already using the PIC UART (for RS232), an internal I2C bus and an
internal SPI bus.
Using an LM75 with I2C output might be an option, but then I will need
a second I2C bus (I don't want any interference on my already existing
I2C bus) and I'm quite short on I/O pins. :(
Dwayne :
Thanks for the zener tip. I will remove it right away next week and
see how the reading reacts to it. The way you would use the LM35 is,
apart from the heavier low pass filter on the readout device side,
exactly the same as I'm using it right now.
Eric / Maurizio :
Using the LM35 in combination with a 4-20mA current loop device might
be an option, but I will need to investigate how these chips are
priced. I don't have any experience with 4-20mA current loops. I
assume it can be used for analog signals too?
The use of the Maxim temperature-to-frequency chipset will totally
depend on pricing. But I will definately look at it.
Alan / Mark / Peter :
The AD950 seems the best solution for me since it already works as a
current loop device. I only hope it's pricing is acceptable for my
application. I will investigate it tomorrow.
I seem to have missed this discussion, have you looked at the DS1820
temperatures sensor which uses a 1 wire bus. Here is a link to a site where
the device has been used. Big advantage is there is no calibration and
because it is digital there should be no loss...
Why not use a LM75, the I2C temp sensor.
Size the pull-up resistors so that your driving close to the max 3mA drive
current.
Then you don't need a PIC with the A/D, you don't use any extra pins if you
already
are using the I2C for an EEPROM.
Run your data rate slow and 100 feet will be no problem with cheap phone
cable and
even RJ11 connectors if you want.
Why tread in the analog world if you don't have to? :)
Works for me.
Jason Wolfson
___________________________________________
Lipidex Corp
50 Franklin Terrace
Duxbury MA 02332
781-834-1600
781-834-1601 FAX spamBeGonejasonspamBeGonelipidex.com
____________________________________________
At 02:20 PM 12/20/00 -0500, you wrote:
>Why not use a LM75, the I2C temp sensor.
>Size the pull-up resistors so that your driving close to the max 3mA drive
>current.
>Then you don't need a PIC with the A/D, you don't use any extra pins if you
>already
>are using the I2C for an EEPROM.
>Run your data rate slow and 100 feet will be no problem with cheap phone
>cable and
>even RJ11 connectors if you want.
>Why tread in the analog world if you don't have to? :)
>Works for me.
You want to connect a 100 foot antenna to your EEPROM pins?
The digital world may turn out to have been a mirage. ;-)
Best regards,
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Spehro Pefhany --"it's the network..." "The Journey is the reward" TakeThisOuTspeffEraseMEspam_OUTinterlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
Contributions invited->The AVR-gcc FAQ is at: http://www.bluecollarlinux.com
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
ATRBIOTECH.COM>
