This was for the app note with a bunch of opamps and resistors.
If there is a small resistor, the sensor element starts to burn power and it creates errors due to self-heating.
-- Tauno Voipio
It "burns power" at any resistance. At 200°C it will dissipate about
160uW in the above circuit. The 10K is what controls the self-heating.Self-heating in air would add less than +0.1°C to the measurement for a typical sensor. Presumably this is not in air, so the contribution to the error budget may not be great.
The 4K99 simply meets the requirements of the LTC2433, just as Bob said.
It should be > 1/2 the max sensor resistance or the ADC will saturate.
It should not be too large or you will lose ADC resolution (otherwise, you could use just one resistor).
Normally you'd want a bit of overrange so that the measurement doesn't die right at the highest temperature setting, so some standard E96 value around 4.02K (15% over) to 4.99K (40% over) could be used. 4.22 (20% over) might be a good choice too.
Best regards, Spehro Pefhany
-- "it's the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
So that's what a 3-wire RTD is for! Thanks!
Bob
he
n tble
sing
ate
Completely wrong, spend most money get worst possible outcome.
You'll lose most of the resolution, but there
y
Ignore him he knows nothing!
The word crappy is in the wrong place
y at
More drivel
=A0I'm
He's right there, you do need compensation leads.
a=A0
NTC, yuk
What you need is a bridge circuit, your ptc100 plus 3 precision temp comp resistors. Believe it or not but they actually make these for this very job. you can find the circuit on the web, it's only been around a few hundred years. You drive the bridge with a sqare wave, use a spare port from the pic. Then you need a 741 (or similar) AC coupled to amplify the signal to something usefull. Put it into the pics adc, sample in the middle of pos and neg periods, do some simple math, 0.5 deg without breaking sweat. they were getting 0.1 deg out of this circuit when a op amp was made up of 3 germanium transistors.
[%X]
[%X]
As one who suggested the OP uses Wheatstone bridge arrangements for his sensor earlier in this thread I can concur that the accuracy is much better. I drive resistive sensors (Carbon Ceramic) in such bridge formations and can get 0.05 degree Kelvin resolution without breaking much of a sweat. Yes we use square wave excitation and are fairly careful with our signal cabling runs of about 150 metres. Then we have a few Tesla of magnetic flux to deal with in the proximity of the sensor housings.
I hope my search suggestions in the previous posting were helpful.
-- ******************************************************************** Paul E. Bennett............... Forth based HIDECS Consultancy Mob: +44 (0)7811-639972 Tel: +44 (0)1235-510979 Going Forth Safely ..... EBA. www.electric-boat-association.org.uk.. ********************************************************************
Paul, I think you and Chuck have forgotten the OP: "I'm struggling to implement an RTD in my design as the interfacing solutions I have seen require a lot of opamps and a precise reference voltage." That's why I suggested a simple R chain and a ratiometric 16 bit delta-sigma. The ADC I suggested will return 0.02C resolution and notches out 60Hz (a likely problem with the OP's transformerless PS). You and Chuck are correct that AC drive and a bridge can be cheaper and more accurate, but given the lack of analog skills in the OP, do you really think a precision bridge and AC drive and conversion is going to lead to success for him? He specifically asked if there was an alternative to all that.
Bob
tter.
d can
we
ng
deal
Right, so loads of precision engineering right there.
The ADC I suggested will return 0.02C resolution
Not without a massive amount of design effort it wont
The AC bridge does that to.
No real analoge skill required, cheap common op amp with two resitors for gain and a blocking capacitor, I think thats alot easier than some tempramental 16bit adc.
do you
It should do, parts are commonplace, AC drive is easy from his micro, adc is olso onboard his micro, I dont see how it gets any easier.
He specifically asked if there was an
It's hard to know why he's even bothering, digital temp controllers can be had for $50 or so.
Thanks to everybody you've been a lot of help thus far.
Not that this is some device that must be mass produced or anything but I would still like to put some sound engineering into it and design the whole thing thus no $40 temperature controller.
Anyways, when initially googling of course I ran into all kinds of circuits using a wheatstone bridge configurations and most of them seemed as simple as described. Its not that I want to avoid analog electronics so much as after reading the microchip application note that I linked to before:
I really second guessed what it would take to accomplish my goal here. A few resistors and a differential amp is a long shot from the circuit they lay out there with a constant current source, signal conditioning, separate ADC, all of which requires a constant reference voltage source--implementing all of this seems like more trouble than its worth.
Why would microchip do all of that if there is a much easier way?
--------------------------------------- Posted through
If you use the right ADC, PT100 (4-wire) is a doddle. All you need is an ADC with differential main input, and a precision resistor.
You power the PT100 via a high stability precision resistor, with a stable low-noise supply ( supply does not need to be accurate or long-term stable) You connect the ADC reference across the precision resistor ( you may or may not need an ADC with differential ref inputs, depending how you arrange things). You connect the ADC inputs to the PT100 sense terminals
This measures the ratio of the PT100 output to the precision resistor, cancelling out effects of the PT100 current. better than 0.01 deg.C accuracy is easily achievable.
There are some examples in this datasheet (fig.50,51)
From memory I think TI's delta-sigma ADCs are cheaper than LT so worth a look there as well
It's just an 'example' circuit that happens to illustrate the use of several uChip parts. You really cannot expect a complete optimized commercial-quality design for free in an application note. It might happen, but not likely. The simple bits are made overly complex and much of the real-world stuff (protection, defined response to breaks, etc.) is just not there.
Best regards, Spehro Pefhany
-- "it's the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
We know its a doddle, I guess your now going to explain the hard way.
Here we go, extra parts, expensive hard to get parts no less.
and a precision resistor.
e low-noise supply
Now we need a low noise supply as well
( supply
may not need an ADC with
Now your not sure quite what you need or how to go about it.
cancelling out effects of
Not on this planet it isn't.
look there as well
And/or because the person writing the application note did not know any better...
-- Roberto Waltman
[ Please reply to the group. Return address is invalid ]
Call it a half-bridge if it will make you feel better.
Well, I slipped a bit (the LTC2433 has a bipolar differential range so I lose a bit for the negative half that I never use) so the resolution is actually 0.04C - still plenty good for a control loop. If you really need more, then step up to a 20 or 24 bit ADC.
Yes, if you sync to the line F. But the delta-sigma also gets you a ton of averaging - averaging that you would have to do yourself if you try to use an on-chip uP ADC.
Chuck, delta-sigmas are not temperamental. 16 to 24 bits is easy at low conversion rates. I know: I've used this RTD circuit almost exactly as shown in thousands of units that are currently in the field.
BTW: I didn't mention it before, but the real reason for the 10k resistor in the chain is to make it easy to detect an open RTD (the ratiometric conversion goes undefined if you put in a zero reference when the chain is broken so you need to detect the case of zero current).
It's all about options, isn't it? He's probably going to spend $30 or more on the oil bath RTD probe. It sounds like a one-off or a small run, so a COTS controller *would* make far more sense from a money+time perspective.
Bob
Can I ask what is the reason for driving the bridge with a square wave?
--------------------------------------- Posted through
Simple single op-amp at the read-back and digitally driven excitation driver with four switching transistors. Nothing so difficult about that.
I don't recall if the OP was USA or elsewhere on the planet. Some of us deal with 50Hz.
There are good circuits in "The Operational Amplifier Cookbook" (Walter G Jung) for suitable differential input op-amp circuits. These come with formulae so that you can make adjustments to your own design.
-- ******************************************************************** Paul E. Bennett............... Forth based HIDECS Consultancy Mob: +44 (0)7811-639972 Tel: +44 (0)1235-510979 Going Forth Safely ..... EBA. www.electric-boat-association.org.uk.. ********************************************************************
Err. extr compared to what? Previous messages have been talking about voiltage refs, opamps etc. You can do it with an ADC, one precision reference and a few cheap passives.
In what way is a diff input precision ADC hard to get? Have you not heard of Digikey? Precision is always going to be expensive - smart design is about minimising the number of expensive parts. You can always scale the components you use according to your requirements.
You always need a stable reference. A high precision resistor is cheapest reference you can use to measure a resistive sensor.
low-noise supply
Resistor, capacitor, or a reference IC if you're really fussy. Any precision solution will need clean power.
not need an ADC with
WTF are you talking about? I'm just suggesting some options - apparently you don't understand that every application is different. e.g. you can put the Rref in the ground leg and use a single-ended reference from the junction of teh pt100 and the resistor, but you might prefer to have the bottom of the PT100 grounded, so a differential ref across a high-side resistor may be a better choice.
cancelling out effects of
I don't know what planet you're on but I have a customer on my planet who has been shipping precision thermometers calibrated to better than .005 deg.C for several years with exactly this method.
The notch in the LT parts defaults to covering 50 *and* 60Hz. Minimum
147 db common mode rejection from 49 to 61.2 Hz. You can also supply your own clock if your interference is somewhere else.The OpAmp Cookbook is a good suggestion.
Bob
Resolution is no good without accuracy, 0.04c requires something out of the ordinary to achieve. How is the op going to calibrate it?
still plenty good for a control loop. If you really
No need to line sync at all.
But the delta-sigma also gets you a ton
Unlikely to need any averaging, but if you do its' kinda like childs play.
dAt 16 bits they are, it's not just the adc, everything else has to be well engineered as well.
16 to 24 bits is easy at lowI strongly suspect you have fooled yourself into this belief.
If you think a bridge is undefined when the rtd is open then you need to go back to school.
o
=A0
It gives you an ac output from your bridge. This allows you to ac couple the amplifier removing the amplifiers drift problem. You can add band pass filtering if needed. By samling the singnal in sync with the bridge excitation to can make a "lock in amplifier" which enables you to resolve the signal even if it is well below the noise floor.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.