Digital thermometer?

[BillyO]

Does anyone know of a decent digital thermometer that is decently accurate and linear without costing and arm and a leg?

I have dozens of DMMs here, some costing many hundreds of dollars and if I lay out the 10 best of them with the thermocouples that they came with I'll get 10 different temperature readings.  Some of the most expensive meters being at the opposite ends of a +/- 15% range!  Also, the linearity sucks!  They all seem to get closer to each other at 0 and 100, but in between it's mayhem.  I have no idea which is the one to trust.

[KungFuJosh]

Have you tried a thermocouple wire that didn't come with any of the meters?

[IanB]

Does anyone know of a decent digital thermometer that is decently accurate and linear without costing and arm and a leg?

https://www.thermoworks.com/

You can get accuracy within 0.5°F with a NIST traceable calibration certificate. There are all sorts of options for monitoring and connectivity, depending on how much you want to pay. The Thermapen ONE gives good basic accuracy for a hand-held thermometer. I use it as a cross check for thermocouple readings.

[bdunham7]

Does anyone know of a decent digital thermometer that is decently accurate and linear without costing and arm and a leg?

Some of the most expensive meters being at the opposite ends of a +/- 15% range!  Also, the linearity sucks! 

Name the brands and models here.  It could be that some of the meters lack cold junction compensation measurement or some are just crap.  For example, the HP 34401A can be modded to read thermocouples, but it does a crap job because of the lack of cold junction measurement.  You can enter the value manually, but that's pretty useless.  Also, your meters have to be in a stable temperature for a while for the compensation settle. 

Here's 4 random meters, 5 TCs and a short.  I'm missing an adapter, so one meter is using the internal measurement.  I can swap the TCs around and all of my errors will be less than 0.5C after 5 minutes.  This picture was taken using the TCs that came with the meters except the 189 which was retrieved from a junk bin.

[tautech]

Does anyone know of a decent digital thermometer that is decently accurate and linear without costing and arm and a leg?

Do you have one of the SDM3000 benchmeters Bill ?

See p47 and later for tweaking thermocouples for better accuracy:
https://int.siglent.com/u_file/download/23_10_10/SDM3055_UserManual_EN03B.pdf

[IanB]

Here's 4 random meters, 5 TCs and a short.  I'm missing an adapter, so one meter is using the internal measurement.  I can swap the TCs around and all of my errors will be less than 0.5C after 5 minutes.  This picture was taken using the TCs that came with the meters except the 189 which was retrieved from a junk bin.

I think reading room temperature is not a good test for a thermocouple. For a better test, you need to read a temperature far from room temperature, such as boiling water. Then you may see variations.

[Mahagam]

Do you have one of the SDM3000 benchmeters Bill ?

SDM3055 SW 1.02.01.28 Relative mode doesn`t work for temperature measurement. Why?

[tautech]

Do you have one of the SDM3000 benchmeters Bill ?

SDM3055 SW 1.02.01.28 Relative mode doesn`t work for temperature measurement. Why?

Please report in detail here for Siglent to investigate:
https://www.eevblog.com/forum/testgear/siglent-new-bench-dmm-sdm3055/

[bdunham7]

I think reading room temperature is not a good test for a thermocouple. For a better test, you need to read a temperature far from room temperature, such as boiling water. Then you may see variations.

Why, because the TC compensation would make it nearly a null reading?  AFAIK the OP was complaining specifically about room temperature readings.  In any case, here's 3 TCs and a Class A 4W RTD in a kettle.  The two Fluke TCs tracked the RTD perfectly all the way up even though they aren't in the exact same spot in the kettle.  The Cheap-Tech meter actually was ahead of them all by a few degrees at one point.  The heat from the kettle was probably affecting the cold junction compensation of the smaller Fluke but I didn't think of that until afterwards.  Still, they're at +/- 0.4C from the RTD in a randomly moving pot of water.

As the kettle is cooling down, the temperatures diverge a bit, but at 70-80C both Flukes and the RTD converge to +/- 0.2C if I stir the pot.  The Cheap-Tech is off by a degree or so at 75C, so it might be a bit non-linear.  But the decent products all agree to less than the apparent actual variation in the pot of water.  IOW your typical errors in temperature measurement are not going to be dominated by TC error here.

[joeqsmith]

I use these platinum thermistors.

[CatalinaWOW]

Depending on the temperature change you want and how portable you want the measurement the temp/humidity measurement ICs are surprisingly good and affordable.  I have used the TI HDC1010EVM, an evaluation board that you just plug into USB and use and been happy with the results.  There are app notes indicating the chip itself has accuracy in the 0.1 degree range.  The evaluation board has the chip on a break off section allowing you to put it remote from the power and USB interface circuits which cause a very significant amount of self heating.  With the sensor broken off the results are very good.

[Fungus]

Bottom line: K-type thermocouples are cheap and cheerful but they're not very accurate.

To make them work well you need to calibrate each individual probe to their meter.

Some meters (eg. The Fluke 189 in that photo) allow you to manually add an offset to the reading but you'll need a temperature reference to be able to figure out what it should be.

[BillyO]

Sorry for the slow reply, been busy with another snow storm.

So, the two meters I'm most concerned with are my Brymen BM786 and my Siglent SDM3055.  They are supposed to be my "best" two meters but they are the furthest apart in their readings.  All my cheaper meters fall in between.  So, I have no idea which is correct or if they are both wrong and nothing to calibrate them against.

In these pictures, the two meters have been on for 30 minutes with two supposedly identical probes just dangling beside each other in the air.  If I exchange the probes (or use entirely different probes) the results are the same.  Maybe a .3 degree difference from what you see here.

[IanB]

Something's wrong. Thermocouple probes are usually accurate to ±1°C or so. Your picture shows a much bigger discrepancy than that. I would say one or other of your meters is malfunctioning.

[KungFuJosh]

Is the correct load set in the Siglent? Choosing the wrong probe type will give all kinds of wonky results.

[Fungus]

So, the two meters I'm most concerned with are my Brymen BM786 and my Siglent SDM3055.  They are supposed to be my "best" two meters but they are the furthest apart in their readings.

It's nothing to do with the meters. It's 100% to do with the probes.

(try swapping the probes around...)

[BillyO]

Is the correct load set in the Siglent? Choosing the wrong probe type will give all kinds of wonky results.

They are type "K" probes and the meter is set to KITS90.

[BillyO]

(try swapping the probes around...)

I mentioned that.  There is about a 0.3 degree difference.

[Fungus]

(try swapping the probes around...)

I mentioned that.  There is about a 0.3 degree difference.

Do you have one with the yellow plug that comes apart. Try shortening the wire, see what happens.

(it might go the wrong way...  )

You'll still need a temperature reference though.

[KungFuJosh]

Is the correct load set in the Siglent? Choosing the wrong probe type will give all kinds of wonky results.

They are type "K" probes and the meter is set to KITS90.

Did you give it time to settle after switching probes? I put a type K Fluke in my SDM3065X also set to KITS90. It started off 4 degrees F too high, then settled within about .5 degrees to the digital thermometer next to it within a few minutes, with no warmup time on the meter. I think my touching the probe is maybe  why it needed time to settle. 🤷

[BillyO]

Did you give it time to settle after switching probes?

Yes, about 10 minutes.  There is a little settling but it slows down pretty quickly.

[KungFuJosh]

Did you give it time to settle after switching probes?

Yes, about 10 minutes.  There is a little settling but it slows down pretty quickly.

How is the meter's accuracy otherwise?

[BillyO]

How is the meter's accuracy otherwise?

Great.  Both of those meters agree with each other well withing their specifications.  The both also agree likewise with my Fluke 8840AF.

However, they don't even agree on the cold junction temperature.  In that case the Siglent is much (6 degrees) lower than the Brymen.  The Brymen, at least, is fairly close to the temperature measured by the probe (room temp) (yes, it a cold room).

[BillyO]

Do you have one with the yellow plug that comes apart. Try shortening the wire, see what happens.

Several, but they are crimped connections.  I don't have any spare terminals.

[bdunham7]

Show us the "cheap" meters as well.  Low power meters probably do the best with TCs and bench meters tend to be the worst.  The issue is thermal gradients and cold junction compensation accuracy are much harder to control in devices that use more power.  As I was playing with TCs and the RTD yesterday to post those photos, I did notice that the Fluke 289 was noticeably worse than the 117 and 189 in random deviations from the RTD, but still well within spec.  IDK what to expect from the Siglent bench meters with TCs, my bench meters either don't read them at all or do a terrible job.  I've no idea why the 786 would be off, how far is it from the other units?

IDK how you can sort this out without getting some very good reference device of some kind, as you mentioned in your OP.  I got a Class A 4W RTD385 Pt100 probe from Evolution Sensors for $60USD and with the bench meter I should be accurate to less than 0.2C without any need to characterize it.  I would assume your Siglent would be able to do something similar with an RTD, perhaps not 0.2C but certainly better than 0.5C, without too much doubt or variance. 

https://evosensors.com/collections/rtd-sensors?srsltid=AfmBOopP2mRcsTX_v6JRe1AQXu7ew_gawkoMEcF5cn8Jm1VI8G9Yn9dY

Can you use the "ALL" readings mode for the Siglent?  This will give you the actual voltage measured and then using the ITS-90 table you can deduce the cold junction compensation value that it is using.  Also, make sure you are actally using the internal measurement.  See pp 48-49 of your manual.

https://www.thermocoupleinfo.com/type-k-thermocouple.htm