PT100 from China - is it a good idea?

[ckocagil]

I want to DIY a temperature reference to measure/calibrate thermocouples. I'd like to have 0.15 C accuracy if possible, but up to 0.3 C is good enough. As far as I understand class A PT100 RTDs will provide these features with no calibration required.

Here are some I found from China. The sellers seem genuine: they sell a lot of temperature instruments, they include details, prices don't look weird...

https://www.aliexpress.com/item/32967406914.html
https://detail.tmall.com/item.htm?id=585260179478

1) Will the 3 wires on these units provide me good enough accuracy? The descriptions says the wires have the same length. If the resistance difference between them is a few percent that should only cause a tiny amount error, according to my quick and dirty calculations.

2) Is there a cheap commercial unit that can satisfy these requirements instead?

3) Is it a better idea to buy a thin film PT100 (like these) and attaching more wires to it to get a 3 or 4 wire unit?

[Kleinstein]

Cheap sensors from China may work, but it is pretty hard to trust the quoted accuracy. 0.15 C accuracy is already a tight speced PT100.

There is a chance one can get away with a cheap sensor, if one does a calibration relative to a bath of ice water. It is not trivial, but it can be done to the 0.1 to 0.2 K level. The ultimate solution would be a DIY tripple point cell  . This is not absolutely impossible.

With some care a 3 wire interface can be sufficient. If using cheap thin film sensors, one could use PT1000 instead and get away with just 2 wire interface. Usually the thin film ones are sufficient at low temperature, but they are often no very accurate and would thus need calibration.

One can attache double wires to get a 4 wire connection to a thin film PT100 sensor.

[edpalmer42]

I'd be very cautious about buying a PT100 sensor that didn't quote the alpha value.  A standard PT100 sensor should have an alpha value of 0.00385.  This value, when inserted into the appropriate equation gives you the resistance at any particular temperature.  I don't know if "Class A" is an equivalent method of stating that.  I'm far from an expert in PT100 sensors.

A few years ago, I picked up a bunch of cheap PT100 sensors from ebay.  They were actually PT100 (100 ohms @ 0C), but the alpha value was way off so the only way to calibrate them would be to determine the actual alpha value and do a custom calibration.

The three-wire interface tells me that it's intended to be used in a bridge circuit because that cancels out the lead resistance.  For best accuracy, you'll need to use that type of circuit topology.

[bgugi]

a well prepared ice point isn't out of the question... nist claims .01 K uncertainty for a simple preparation https://www.nist.gov/system/files/documents/2017/04/28/Ice-Melting-Point-Validation-Method-for-Data-Loggers.pdf ISO E563 and some other sources claim that an ice point prepared and maintained very carefully can achieve uncertainties as low as 2-3 mK.

performing an ice point check can improve your confidence in the quality of the RTD, it does not guarantee matching to the alpha, beta, or delta specified for a "standard" rtd curve. You'd need additional checks outside the range of use to verify these.

To use a 3 wire rtd, you'll need a readout that actually performs 3 wire measurements, or the ability to switch between two-wire measurements of the rtd and the connected leads and do manual subtraction, don't try to cheat by splitting one of the wires at the end to do 4 wire measurements.

[maginnovision]

https://www.thorlabs.com/thorproduct.cfm?partnumber=TH100PT&gclid=Cj0KCQiArdLvBRCrARIsAGhB_syOiNQ-qpV5wlV-mLS_rI07n491QXRE88BCgE_7v5HJwZIeSiKscDIaArLWEALw_wcB

Something like this might work ok as well. Pretty cheap and a simpler 2 wire setup.

[ak634]

What is the temperature range?
Some of the cheap RTD on ebay seem to be marked "WZP". Assuming this means anything, there is a datasheet here: http://www.cheemi-tech.com/pdf/Cheemi%20PT100%20series%20RTD%20Temperature%20Sensor.pdf. They do not specify alpha, but provide a t to R table, which well matches alpha = 0.00385 between 0 and 100 deg. C, but the difference quickly adds up outside this range. E.g. it is 3 deg. C at 200 deg and ~55 deg C at 650 deg C. I remember some ebay sellers used to attach the curve to the listings, but now I can't see any.
I have a couple of them.

EDIT. Actually, there is a "TRC" in the table, which is for some reason, 0.003581. Using this as alpha brings the table data within +- 4 deg C, but this is still far outside of their tolerance range.

[bgugi]

ak634, pt100's aren't linear, and typically shouldn't be used as such. the sensor you linked matches the standard industrial curve.

[GeorgeOfTheJungle]

I want to DIY a temperature reference to measure/calibrate thermocouples.

Keep in mind that thermocouples have a much wider temp range, but the upper limit of platinum detectors is ~ 500°C, and many PT100 sensors are not even designed to sense temps that high. If the thermocouples you want to test are going to be used outside of the temp range of the PT100 you use to calibrate, I'd say don't do that.

There's an awesome breakout board on ebay/amazon/aliexpress with a MAX31856 for a couple $ that works with 2, 3 or 4 wire PT100 RTDs. That and an arduino sketch is all you'd need: search for "max31865 pt100".

Edit: There's plenty of good info @ the wikipedia: https://en.wikipedia.org/wiki/Resistance_thermometer

[mzzj]

I'd be very cautious about buying a PT100 sensor that didn't quote the alpha value.  A standard PT100 sensor should have an alpha value of 0.00385.  This value, when inserted into the appropriate equation gives you the resistance at any particular temperature.  I don't know if "Class A" is an equivalent method of stating that.  I'm far from an expert in PT100 sensors.

Class A would or should refer to IEC 751 which is  alpha = 0.00385

1/10 class B would be common solution. I'd look for something like these:
https://www.tcdirect.co.uk/Default.aspx?level=2&department_id=230/21

[mzzj]

https://blog.beamex.com/pt100-temperature-sensor

[MegaVolt]

Someone on the forum praised this sensor: https://www.aliexpress.com/item/-/32917771457.html