EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: Electro Fan on August 29, 2020, 06:43:57 pm
-
Some questions....
Any reason that this:
https://www.keysight.com/en/pd-2309587-pn-34152A/pt100-rtd-4-wire-class-a-sensor-kit?cc=US&lc=eng (https://www.keysight.com/en/pd-2309587-pn-34152A/pt100-rtd-4-wire-class-a-sensor-kit?cc=US&lc=eng)
would perform significantly better than this:
https://www.ebay.com/itm/303658131860 (https://www.ebay.com/itm/303658131860)
plus a couple of these:
https://www.pomonaelectronics.com/products/hardware/double-banana-plug-solder-turrets (https://www.pomonaelectronics.com/products/hardware/double-banana-plug-solder-turrets)
Seems like for accuracy that PT100 Class A might be sufficient with 4 wires but any reason to consider PT1000?
Class A specifies more accuracy (tighter tolerances) than Class B (but there are classes with better accuracy than A)?
FWIW, these are pretty informative:
https://reotemp.com/wp-content/uploads/2015/11/TB3-4WIRE-06143-Wirevs4-WireRTD.pdf (https://reotemp.com/wp-content/uploads/2015/11/TB3-4WIRE-06143-Wirevs4-WireRTD.pdf)
https://reotemp.com/wp-content/uploads/2015/11/TBRTDTOL-0614RTDToleranceClasses.pdf (https://reotemp.com/wp-content/uploads/2015/11/TBRTDTOL-0614RTDToleranceClasses.pdf)
https://www.thermometricscorp.com/acstan.html (https://www.thermometricscorp.com/acstan.html)
https://blog.wika.us/products/temperature-products/pt100-and-pt1000-sensors-important-facts-and-differences/#:~:text=Pt100%20sensors%20have%20a%20nominal,%C2%B0C%20is%201%2C000%CE%A9.&text=The%20temperature%20coefficient%20of%20resistance,10%20compared%20to%20Pt100%20sensors. (https://blog.wika.us/products/temperature-products/pt100-and-pt1000-sensors-important-facts-and-differences/#:~:text=Pt100%20sensors%20have%20a%20nominal,%C2%B0C%20is%201%2C000%CE%A9.&text=The%20temperature%20coefficient%20of%20resistance,10%20compared%20to%20Pt100%20sensors.)
Net, net: Anyone have any recommendations on a PT100 Class A 4 wire RTD that is cost-effective (closer to the eBay 4 wire RTD price than the Keysight price)? Or anything else you like better with other specs (PT1000, etc.)?
Edit: Looks like still more configuration choices:
https://www.ebay.com/itm/Keysight-PT100-RTD-4-Wire-NIST-Class-A-Temperature-Sensor-Kit-34152A-34153A-Temp/203083931638?hash=item2f48bedff6:g:1nEAAOSwTbFfQ9UO (https://www.ebay.com/itm/Keysight-PT100-RTD-4-Wire-NIST-Class-A-Temperature-Sensor-Kit-34152A-34153A-Temp/203083931638?hash=item2f48bedff6:g:1nEAAOSwTbFfQ9UO)
1. exposed sensor - fastest response?
2. surface patch
3. stainless steel probe (vs. platinum?)
For simply measuring ambient temperature would one of the 3 above be preferred? For some DUT situations, others might be preferred?
Not sure if these questions lead to a sub-branch of TEA/Volt Nut/Ohm Nut/Temp Nut .... :)
-
If you are trying to choose between different class A RTDs then the accuracy of your temperature measurement is obviously important to you. I would suggest getting a system calibration of your meter / data logger with the sensors you want to use is as important as the choice of which sensor you buy. By system calibration I mean the meter/logger and sensor are calibrated together at the same time.
Probably best to get a cheap type A PT100 sensor and use the money saved to get the calibration done?
As all decent sensors will be 4 wire dont worry too much about the connectors / changing them as it will not effect the accuracy.
-
From my (long) experience:
The sensor is even more important than the readout part.
The sensor is mechanically fragile, can suffer from temperature shocks etc.
I had problems with custom probes from Southern Temperature Sensors: they didn't remove humidity before potting the sensor.
The effect was several megs of leakage resistance to the sensor's inox shell and related very unstable readings.
There are dozens of companies which specialize in manufacturing sensors but only few of them reached stability and reliability level which I would call 'entry level metrology'.
I'm not talking here about SPRT but high class industrial PRT.
I can recommend one of best products regarding value for the money:
https://www.labfacility.com/pt100-precision-probe-with-2-metre-lead-type-prt.html (https://www.labfacility.com/pt100-precision-probe-with-2-metre-lead-type-prt.html)
P.S. Soon I'm planning to release my new product: affordable 10 channel RTD readout with 1mK resolution and very good long term stability.
It is the successor of the RFS2804A. I got regular info from metrology lab who uses newest version of RFS2804A that they observe 1ppm per year resistance measurement drift.
-
Thanks for the replies.
I'm looking forward to trying a 4 wire sensor.
Any thoughts on using a patch vs an exposed sensor vs a metal probe?
Seems like the exposed sensor might be the most flexible for measuring ambient temp plus more specific/small DUTs.
-
I can recommend one of best products regarding value for the money:
https://www.labfacility.com/pt100-precision-probe-with-2-metre-lead-type-prt.html (https://www.labfacility.com/pt100-precision-probe-with-2-metre-lead-type-prt.html)
I can also recommend Labfacilty, I have bought from them (in various roles) for many years now. If things do go wrong, they are very helpful and keen to sort them out. You can also phone them up and talk to a real human being who knows their stuff, about your application.
-
Thanks for the recommendations on Labfacility.
As you can see, I'm just getting to metrology 101, or maybe just metrology 100.
Nothing very scientific (in the form of repeatable precise measurements) here but given that I have some confidence in the Keysight and that the Fluke has shown reasonably similar measurements when both used theromocouples....
It looks like with long lead relatively thin gauge test leads the Keysight reads within about 1 degree F of the Fluke + thermocouple. (Moral of this story? Maybe if you have a meter that measures temperature but you don't have a thermocouple, you might be able to get by with test leads.... and maybe even do better if you have the right leads...., ie, low resistance test leads.)
When using shorter thin gauge test leads the readings come within roughly 0.7 degree F.
When using short larger diameter leads terminated in banana plugs the readings look to be within about 0.1 degree F.
Again, nothing very scientific here but these tests do indicate resistance impacts temperature readings - and it seems that to get it all precisely dialed in looks like it might take some extra budget, time, and patience. Just seems to depend on on accurate is accurate enough.
Other moral of the story - getting close to time to turn on the air conditioning :)
-
There are more accurate sensors than class A. There is class AA and DIN/3.
PT1000 is less influenced by cable length and self heating. If you can choose, choose 1000.
Omega is another brand that sells RTDs. They also have a lot of info on their website. I wouldnt bother with Fluke and Keysight ones, they just re-brand other RTDs.
-
There are more accurate sensors than class A. There is class AA and DIN/3.
PT1000 is less influenced by cable length and self heating. If you can choose, choose 1000.
Omega is another brand that sells RTDs. They also have a lot of info on their website. I wouldnt bother with Fluke and Keysight ones, they just re-brand other RTDs.
Thanks NANDBlog,
I'm not sure what the tradeoffs are yet between 4 wire PT100 and 2 wire PT1000 (or maybe there is such a thing as 4 wire PT1000?).
PT1000 - only 2 wires? Looks like these go for $3.55 each. Just need the wires....
https://www.ist-ag.com/en-us/products/pt1000-iec-60751-f03-3750-ppmk (https://www.ist-ag.com/en-us/products/pt1000-iec-60751-f03-3750-ppmk)
-200 °C to +600 °C
< 0.04 %
PT100 - these look to include the sensor with 4 wires $26.35 each.
https://www.ist-ag.com/en-us/products/realprobetemp (https://www.ist-ag.com/en-us/products/realprobetemp)
-50 °C to +200 °C
< 0.04 %
Looks like three types of 4 wire configs:
https://www.ist-ag.com/en-us/news-events/news/new-platinum-rtd-4-wires-high-precision-temperature-measurements... (https://www.ist-ag.com/en-us/news-events/news/new-platinum-rtd-4-wires-high-precision-temperature-measurements...) always another layer to decode :)
Looks like they are looking for a Measuring Technician :)
https://www.ist-ag.com/en-us/about-us/career/jobs/messtechniker-mwd (https://www.ist-ag.com/en-us/about-us/career/jobs/messtechniker-mwd)
Main Next Question: Back to basics on RTDs, any reason to go with a metal (platinum?) probe vs a bare sensor (just the wire leads)? Maybe it's ok to go with bare wire sensor leads if the insulation can handle the temp, otherwise the metal probe would be better?
-
IME it's not that important what sensor is used internally but rather how it is attached to wires. I've seen plenty of very expensive RTD which were not stable long term because of crappy crimping to wires. 0.1 ohm here or there makes a hell of a lot of difference.
-
IME it's not that important what sensor is used internally but rather how it is attached to wires. I've seen plenty of very expensive RTD which were not stable long term because of crappy crimping to wires. 0.1 ohm here or there makes a hell of a lot of difference.
So does that imply that maybe a bare wire sensor might actually be advantageous (both for good measurement results and also because it could be more easily placed either in the open (ambient) air or touching a relatively small DUT?
I realize the image below is just a K thermocouple - but it seems to just be an exposed bare wire? yes/no?
Same/similar with the image of the PT100 RTD - just an exposed (bare wire?) sensor? yes/no?
https://www.ebay.com/itm/Keysight-PT100-RTD-4-Wire-NIST-Class-A-Temperature-Sensor-Kit-34152A-34153A-Temp/203083931638?hash=item2f48bedff6:g:1nEAAOSwTbFfQ9UO (https://www.ebay.com/itm/Keysight-PT100-RTD-4-Wire-NIST-Class-A-Temperature-Sensor-Kit-34152A-34153A-Temp/203083931638?hash=item2f48bedff6:g:1nEAAOSwTbFfQ9UO)
- under the Probe Type drop down Keysight lists it as an Explosed Sensor (Fastest Respon...)
I don't need any particular brand, just need to figure out the preferred configuration of the sensor end.
Seems like to manage to the tenth of an ohm it would be good to go with 4 wire, and a bare sensor - but I'm just starting near zero trying to figure it out.
Thx
-
@Electro Fan - have a read of this article as it gives the differences between thermocouples, RTDs and 2 & 4 wire sensors.
https://www.picotech.com/library/application-note/improving-the-accuracy-of-temperature-measurements (https://www.picotech.com/library/application-note/improving-the-accuracy-of-temperature-measurements)
It was a bit of a shock when I realized I wrote that article over 20 years ago - I think its all still relevant though.
-
There are more accurate sensors than class A. There is class AA and DIN/3.
PT1000 is less influenced by cable length and self heating. If you can choose, choose 1000.
Omega is another brand that sells RTDs. They also have a lot of info on their website. I wouldnt bother with Fluke and Keysight ones, they just re-brand other RTDs.
Thanks NANDBlog,
I'm not sure what the tradeoffs are yet between 4 wire PT100 and 2 wire PT1000 (or maybe there is such a thing as 4 wire PT1000?).
PT1000 - only 2 wires? Looks like these go for $3.55 each. Just need the wires....
https://www.ist-ag.com/en-us/products/pt1000-iec-60751-f03-3750-ppmk (https://www.ist-ag.com/en-us/products/pt1000-iec-60751-f03-3750-ppmk)
-200 °C to +600 °C
< 0.04 %
PT100 - these look to include the sensor with 4 wires $26.35 each.
https://www.ist-ag.com/en-us/products/realprobetemp (https://www.ist-ag.com/en-us/products/realprobetemp)
-50 °C to +200 °C
< 0.04 %
Looks like three types of 4 wire configs:
https://www.ist-ag.com/en-us/news-events/news/new-platinum-rtd-4-wires-high-precision-temperature-measurements... (https://www.ist-ag.com/en-us/news-events/news/new-platinum-rtd-4-wires-high-precision-temperature-measurements...) always another layer to decode :)
Looks like they are looking for a Measuring Technician :)
https://www.ist-ag.com/en-us/about-us/career/jobs/messtechniker-mwd (https://www.ist-ag.com/en-us/about-us/career/jobs/messtechniker-mwd)
Main Next Question: Back to basics on RTDs, any reason to go with a metal (platinum?) probe vs a bare sensor (just the wire leads)? Maybe it's ok to go with bare wire sensor leads if the insulation can handle the temp, otherwise the metal probe would be better?
There are 4 wire PT1000s sensors. What you posted, is only a sensing element. So they typically solder/crimp cables to these sensors. It depends on the accuracy, how many meters of cable your measurement can tolerate. With the same cable, a PT1000 could tolerate 10x the cable length with the same accuracy. If you just want to measure something 1m away, you dont need to worry about it. Same if you just want 3 degrees accuracy. Also worth noting, that cables can be a mayor PITA when it comes to measuring temperature. Normal silicone cables cannot handle more than 150 degrees C. Teflon coated cables go to 260C, then they become poisonous. Even your sensing element goes to 600C, your cable might not. Then you end up with enameled wires with steel braid, thermowells and other weird equipment, which costs $$$.
What you posted there is not an RTD, that is a thermocouple. They have their drawbacks and benefits, but they are not as accurate as an RTD (broadly speaking). They typically use Nickel and Platinum RTDs. The P in PT100 means Platinum.
You should decide first, what you measure. Liquid or not. Surface temperature. Is the cable waterproof. How close is the cable. What temperature range. How long is the cable. What accuracy. What is the measuring equipment, what sort of connector does that have.
If by DUT, you mean measuring an IC temperature, then a PT1000 is not the right part for that. for that, you should buy miniature thermocouples. T type for example is solderable. You will loose some accuracy, but usually that doesnt matter.
-
Wow, NANDBlog and ADT123, THANKS FOR THE VERY HELPFUL REPLIES!! (ADT, congrats on writing an article that has stood for 20 years - definitely shows you were on it and ahead of it.)
I'm going to re-read the posts and will be back with some more questions. Thanks again very much.
- just one quick combination standards big picture plus practical equipment question in the meantime:
ADT123's article says that the DIN 43760 standard, also called the ‘European Curve’, is 0.385 Ω/°C for a Pt100 sensor.
Keysight says that their 34465A DMM uses the measurement for the IEC751 standard RTD, which has a sensitivity of 0.385%/°C. So, perhaps DIN 43760 and IEC751 are both based on the same (or similar?) specs.
No doubt standards evolve and I'm still coming up to speed on the standards but from a practical standpoint unless the 34465A can support PT1000 (which I haven't figured out yet), I think I might have to focus on PT100. I'm going to do some more reading now..... Thx!!
-
Still not sure if a 34465A can support Pt1000 (although it appears that Pt1000 meets the IEC751 standard, so maybe) .....
but for anyone following this RTD thread the following articles address various questions about Pt100 vs Pt1000.
https://blog.wika.us/products/temperature-products/rtd-sensors-pt100-faqs/
https://blog.wika.us/products/temperature-products/pt100-and-pt1000-sensors-important-facts-and-differences/
-
Still not sure if a 34465A can support Pt1000 (although it appears that Pt1000 meets the IEC751 standard, so maybe) .....
but for anyone following this RTD thread the following articles address various questions about Pt100 vs Pt1000.
https://blog.wika.us/products/temperature-products/rtd-sensors-pt100-faqs/
https://blog.wika.us/products/temperature-products/pt100-and-pt1000-sensors-important-facts-and-differences/
It does. You can set the 0 degrees resistance to 1000 Ohm.
-
Just reporting back on my first RTD. From eBay, Hong Kong, arrived in 12 days.
Slightly limited temp range (top end is 200C) but supposedly PT100. Hard to compare it to much (just have a Fluke and a thermocouple), but so far my impression is good. Only slight issue is that the leads are pretty fine where they connect to banana terminals but performance-wise so far so good, I think.
https://www.ebay.com/itm/4-Wires-Class-A-RTD-PT100-Temperature-Sensor-Probe-50-200-C-58-392-F-2-M/303658131860?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2060353.m2749.l2649 (https://www.ebay.com/itm/4-Wires-Class-A-RTD-PT100-Temperature-Sensor-Probe-50-200-C-58-392-F-2-M/303658131860?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2060353.m2749.l2649)
4 Wires Class A RTD PT100 Temperature Sensor/Probe, -50~200 °C / -58~392 °F, 2 M
Item Location: HK
Postage: Air Mail & Ship Fast with Tracking no.
Packet: Simple environmental friendly pack, no box, no batteries, well protected
Product Details
Spec1 Temperature Range -50 ~ 200°C (-58 ~ 392°F)
Spec2 Platinum resistance thermometers (PRTs) PT100, 4 Wires connection
Spec3 Tolerance of PT100 Ω (Alpha = 0.003850 @ 0°C) , Class A : ± 0.06 Ω OR ± 0.15°C
Spec4 Length : 2000 mm, 2 Meters (long)
Spec5 Probe head: 4 mm x 30 mm
-
Here is another. This one instead of a phone camera photo was a screen capture using the Keysight Web Control interface - kinda nice.
What I think I noticed is that when the temperature measurement just ran overnight it was a pretty smooth curve; when I arrived at the bench and spent some time within arms reach of the 34465A and the RTD, or just a foot or so beyond, the curve seemed to reflect my presence - so I think it's pretty sensitive.
-
Anders Celsius is disappointed.
(https://upload.wikimedia.org/wikipedia/commons/b/b1/Anders-Celsius-Head.jpg)
mod: Is there any way to make him fit on the screen?
-
Nice photo of a person who figured out a lot of valuable info - see below in case anyone isn’t aware of his contributions beyond temperature.
.... so I’ll bite, why is he disappointed?
In 1730, Celsius published the Nova Methodus distantiam solis a terra determinandi (New Method for Determining the Distance from the Earth to the Sun). His research also involved the study of auroral phenomena, which he conducted with his assistant Olof Hiorter, and he was the first to suggest a connection between the aurora borealis and changes in the magnetic field of the Earth. He observed the variations of a compass needle and found that larger deflections correlated with stronger auroral activity. At Nuremberg in 1733, he published a collection of 316 observations of the aurora borealis made by himself and others over the period 1716–1732.[1]
Celsius traveled frequently in the early 1730s, including to Germany, Italy and France, when he visited most of the major European observatories. In Paris he advocated the measurement of an arc of the meridian in Lapland. In 1736, he participated in the expedition organized for that purpose by the French Academy of Sciences, led by the French mathematician Pierre Louis Maupertuis (1698–1759) to measure a degree of latitude.[1] The aim of the expedition was to measure the length of a degree along a meridian, close to the pole, and compare the result with a similar expedition to Peru, today in Ecuador, near the equator. The expeditions confirmed Isaac Newton's belief that the shape of the earth is an ellipsoid flattened at the poles.[2]
In 1738, he published the De observationibus pro figura telluris determinanda (Observations on Determining the Shape of the Earth). Celsius's participation in the Lapland expedition won him much respect in Sweden with the government and his peers, and played a key role in generating interest from the Swedish authorities in donating the resources required to construct a new modern observatory in Uppsala. He was successful in the request, and Celsius founded the Uppsala Astronomical Observatory in 1741. The observatory was equipped with instruments* purchased during his long voyage abroad, comprising the most modern instrumental technology of the period.
He made observations of eclipses and various astronomical objects and published catalogues of carefully determined magnitudes for some 300 stars using his own photometric system (mean error=0.4 mag).[2][3][4] He proposed the Celsius temperature scale in a paper to the Royal Society of Sciences in Uppsala, the oldest Swedish scientific society, founded in 1710. His thermometer was calibrated with a value of 0 for the boiling point of water and 100 for the freezing point. In 1745, a year after Celsius's death, the scale was reversed by Carl Linnaeus to facilitate more practical measurement.[5]
*some TEA?
https://en.m.wikipedia.org/wiki/Anders_Celsius
-
The best temperature sensor really depends on what you want to know the temperature of which if you explained clearly.
I would choose a very different sensor if I wanted to measure things fast than slow and I would use a different one if I wanted to work very cold (like 4 K) than if I wanted to go hot (1000 C). I've done measurements across this entire range.
The temperature sensor tells you the temperature of the sensor at best. Using an off the shelf PRT without calibration of the PRT should give you the class rating you mention but beware stress in the metal film can change the alpha of the metal for instance. Also be aware there are effects like hysteresis in these sensors. If you work reasonably hard and get a sensor that costs about 500 dollars you should be able to get long term accuracy of better than 100 mK.
Lakeshore would be a good place to look at all the different technologies for temperature. One thing many people don't realize is that you can use a humble diode as a very accurate temperature sensor if you have a good stable current source and an excellent volt meter. If you have another reference sensor and are willing to calibrate them you should be able to use almost any regular pn junction diode.
For casual temperature measurement not requiring scientific traceability the normal K-type thermocouple is normally enough. Meanwhile thermistors tend to have better sensitivity than anything else but require more care to calibrate.
Naked sensors not in nice stainless cans of course respond faster because there is less specific heat to the package to have to supply joules to heat or cool.
Another way to measure temperature is of course pyrometry which is what thermal cameras are doing. You need to know the emissivity of the surface to get a real number, but sometimes it is the best way when you have a very hot chamber and cannot put something metal in it because for instance that metal would contaminate your process.
You could also measure temperature by looking at noise or dissipation in something. Fluctuation-dissipation theorem connects noise and dissipation so you can measure either of any system (resistor, oscillator) and measure the temperature.
Another way that is quite accurate to measure a temperature is to look at wavelength shifts of something like a Brag grating. If you know the materials and their thermal properties than a wavelength shift with the correct lasers can get you a very good temperature measurement. One of my old colleagues at NIST I think is working on an optical temperature measurement with integrated optics. If you have a good spectrometer or wavemeter around you can look at the wavelength shift of a VCSEL or other diode laser. They tend to move sub-nm per degree. LEDs will do the same stuff depending on the structure but the emission is broader so knowing the center is less defined.
Anyway, you can turn any process that depends on temperature into a thermometer. If your leads are short a 2 or 3 wire PRT is fine for any slow measure to better than 0.2-0.5 degree K probably. If you want better accuracy (not precision which will be a lot better than 0.2 degree) then you will want to not just buy any old thing.
-
.... so I’ll bite, why is he disappointed?
You measure in Fahrenheit. I know, you probably was brought up that way, people probably forced you to use it. But it is not too late. There is light at the end of the tunnel.
-
.... so I’ll bite, why is he disappointed?
You measure in Fahrenheit. I know, you probably was brought up that way, people probably forced you to use it. But it is not too late. There is light at the end of the tunnel.
Roger that. I get it and agree. Thanks
I had a hunch maybe that was what you were saying.
To be square, the reason I used F this time is that it shows me slightly more granularity (or at least in my mind that's the way it works sometimes). But I know that lots of (most) useful measurements (and overall measuring systems) are referenced in C (and metric in general vs. imperial). It's a habit and an environment - if there is one thing that could probably propel people stuck in Imperial farther forward faster it would be to just get with and stick with with metric. In the meantime it helps keeps the U.S., Myanmar, and Liberia synchronized when it comes to units of measurement. (Although rumor has it Myanmar and Liberia might be substantially on Metric.)
I appreciate your optimism (light at the end of the tunnel) and encouragement (it's not too late). :-+
-
The choice of PT100 vs PT1000 also depends on the temperature to measure. The PT1000 sensors are usually thin film type. With PT100 one can get both thin film type and more expensive but also more stable wire type sensors.
Thin film sensors age faster when used at high temperatures.
Another point at higher temperatures like > 400 C is that isolation can become tricky. Leakage currents are more of a problem with PT1000.
In some cases one even has PT25 or PT10.
Pt100 sensors are usually made to be used with 1 mA. Pt1000 may be calibrated for some 300 µA - the usually DMMs use either 1 mA or 100 µA.