8 RTD (PT100) Input Schematic 2

[Morcego]

Hi,

I need advice in designing a board for temperature measurement using PT100.

I know that I need a stable current generator (of about 1ma) to the PT100 and then read the voltage drop.

My questiom is: Do I need a current generator for each PT100, or can I use one of all the 8 ?

The current generator is generally feed by a stable voltage reference. This APP note

ww1.microchip.com/downloads/en/AppNotes/00687b.pdf

from microchip shows that kind of circuit. Can I use Vref (2,5V) to drive 8 current generators like the one show on figure 4 ?

Thanks in advance for any help

Joaquim

 

[IRstuff]

Depends on how you're going to interrogate the PRTs

TTFN

 

[itsmoked]

Well you can use a mux to drive the current thru each RTD in turn. Since it is a current source you can pretty much ignore any inherent resistances in the mux.

But for sure, you should just use one.

Alternatively you can just use the RTDs as the R in an RC circuit and hence use any old voltage reference for the charger.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[danw2]

Did you plan on measuring the 3rd or 4th wire resistance(s), to compensate for lead length resistance?

Is this only for Pt 100 ohm 385 IEC or for other alphas, like the 100/120 ohm nickel or 10 ohm copper bulbs?

 

[Morcego]

Hi,
Thanks for all the awnsers. My plan is to use a 3 wire layout. Standard PT100, the voltage drop is then measured with an 8 input, 24 bit, serial ADC (LTC2418) from linear.

I already have this working with one channel. But when I add more channels using the same current source I get some erratic readings. On the bench everything works ok, but in the field, with remote PT100's the erratic readings start.
And the problem is not on the location, because other comercial readers (with only one channel) work ok.

Joaquim

 

[itsmoked]

You can't run the one current source thru them all at once! This would make the RTDs all "interact" with each other.

Look at:

http://www.ee.washington.edu/stores/DataSheets/cd4000/cd4051.pdf#search="4051"

for switching the ref between all the different RTDs.

Also watch the 1mA as that may cause heating of the RTD and strange drifts.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[cbarn24050]

Hi, It looks to me like a poor concept. I can see you having problems getting stable accurate results. A much better option would be traditional bridge arrangement.

 

[VEBill]

"...the 1mA as that may cause heating of the RTD..."

Is it common practice to cycle the current so that it is only ON when actually required (resulting in a very low duty cycle, for the heating, based on the measurement time compared to the desired update cycle)?

 

[LionelHutz]

Last time I checked a 3-wire compensating RTD circuit required 2 matched current sources. The current really should be switched to avoid self heating.

Why not use something like an Analog Devices AD7711 for an all-in-one solution?

Analog Devices also have some IC's built to use AC excitation which helps eliminate measuring parasitic thermocouples. An example is the AD7730.

 

[logbook]

I don't know what sort of error magnitude is acceptable to you, but 1mA continuously seems fine to me. The self heating is then less than 1mW which isn't much.

It smoked is correct with the comment that you need one current source per PRT.

The current source has a huge number of error sources. Remember that a 100R PRT has 0.38 ohms per degreeC on a 100R basic. That is 0.4% error gives 1 degree error! You need 0.1% resistors to define the currents. You should then use one measurement channel to measure the 2.5V reference to give a ratio measurement and therefore a more stable measurement.

 

[itsmoked]

Lets see where to start..

Three-wire is pretty much the de facto standard. You can compensate for the lead resistance, close to perfectly, with three leads. Four leads are often used when your circuitry needs to be rudimentary, and you don't mind bigger sensor heads and extra cost for the RTDs.

You do not need two current sources for the three wire solution.

1mA continuously is no problem IF you can calibrate your system to offset the self heating.

1mA or even a few mA switched briefly in, during measurement, works very well and is quite common as long as the "switched on" time is rigidly enforced otherwise, you will get horrendous reading variations that are not subject to calibration.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[nbucska]

Before we start designing, we need SPECS!!!!

Plesae read FAQ240-1032
My WEB: <

http://geocities.com/nbucska/>

 

[LionelHutz]

Keith, do you have an example of a single current source 3-wire RTD circuit that compensates for all lead resistance? Just curious because I've never seen such a circuit. 4-wire yes but not 3.

 

[itsmoked]

Yo LionelHutz!

I have used this method to great effect many times. Until you work thru the description and math you won't "get it".

Here's the description(at bottom):

http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1155,C1001,C1152,P1860,D4458

Here's similar:

http://ww1.microchip.com/downloads/en/AppNotes/00687b.pdf

And here's what I'm telling the OP:

Text:

http://www.elecdesign.com/Articles/Index.cfm?AD=1&ArticleID=1724

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[LionelHutz]

I see what you mean. Pretty simple - no real math because it just measures the voltage drop in the wire and compensates for it. I was thinking of two current sources where the current flows will cancel - actually not really cancelling but they create the same polarity voltage drop in both measurement wires.

To answer one origional question - I don't see any reason why you couldn't use the same Vref to feed 8 current sources. You could use a multi-channel input A/D and by just adding more of the op-amp circuits build the 8-input circuit.

 

[itsmoked]

Um.. the problem is, every current source will be different. That can cause a headache in calibration. Often it is better to have one good one and just monitor it.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[Skogsgurra]

AD7783 in the diagram.

Electronic Design has often a problem with prefixes. Now it happened again. The 200 milliamps shall read 200 microamps. Just to avoid any confusion.

Gunnar Englund

http://www.gke.org

 

[itsmoked]

Yeah! Talk about self heating!! 20W - RTD light bulbs.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[LionelHutz]

Yes, monitoring one current source certainly is the easiest way to know what you have Keith but if the OP is using a circuit like fig 5 of the Microchip app note link then the current likely isn't being monitored. In that case, I don't see why adding more current sources would really make a difference. That ADG711 is a neat IC and seems perfect for this application but doesn't the CD4051 have an awfully high on resistance for this application? Not knowing the circuit design as it stands doesn't help.


Morcego - I don't fully understand but in your other post did you mean that the single channel version of the circuit is not working in the field? Does this mean there a problem with the basic one channel circuit that needs addressing first? Is this monitor for a motor?