Digital thermometer?

[bdunham7]

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).

How are you getting that value on each of them? 

[Fungus]

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.

For playing around you can probably just wind the wire around the connector a bit to make it shorter.

[Kleinstein]

One gets the cold junction temperaure with a copper wire as a short. A bit higher than room temperature can be correct for a meter that runs a bit warm and this way heats the terminals / connectors.

The extra heat from the bench DMM can effect the cold junction compensation. So bench DMMs can be a bit problematic with thermcouples.
A TC with 4 mm plugs is not a great solution to start with as the cold junction temperature is measured at the wrong place.

For some uses one can also use NTC thermistors - some of them are reasonable accurate / selected.

[Fungus]

Yep, as noted in Dave's video; the cold junction temperature is usually measured quite a long way from the actual cold junction.

I'd expect bench meters to be more affected by this than handhelds simply because they're warmer inside.

[BillyO]

Here's the Brymen with 4 of the better cheapies.  It agrees fairly well with these ones.  Is this evidence that I need to work with adjusting the Siglent?

Also shown is the shorting block I use for cold junction testing.

[BillyO]

A TC with 4 mm plugs is not a great solution to start with as the cold junction temperature is measured at the wrong place.

I can't see it making that much difference.  Unless the meter has a "K" type connector socket you are going to have to go through some sort of adapter with 4mm plugs.

BTW, after several hours of being on, the Siglent is now actually measuring the cold junction higher than the Brymen.

[bdunham7]

One gets the cold junction temperaure with a copper wire as a short.

For some uses one can also use NTC thermistors - some of them are reasonable accurate / selected.

That seems obvious in retrospect, especially since I already know that it works on various Fluke meters as I'm doing with the 189 in the previous photos.  It never occurred to me til just now that it should always work that way, partly because I've seen examples where it doesn't seem to.  But the only example I have here now is a 34401A and I know why it doesn't work on those.  The reason I asked how he had gotten the value was to explore how and where those errors were introduced.  I'm thinking the Siglent bench meter has a calibration adjustment that may change that displayed cold junction value.  It has a user adjustment as well, but that's separate AFAIK.

Every meter using TC actually depends on a calibrated NTC thermistor or similar device anyway, so using one directly seems to make perfect sense.  It's just that getting a direct readout in degrees C or F takes bit more work.

[bdunham7]

Unless the meter has a "K" type connector socket you are going to have to go through some sort of adapter with 4mm plugs.

BTW, after several hours of being on, the Siglent is now actually measuring the cold junction higher than the Brymen.

A miniature thermocouple connector right on the meter would be ideal, but that apparently runs afoul of safety ratings.  A standalone TC thermometer might be the best solution for TCs.  I believe the better adapters are constructed to minimize junction errors, but I've no idea of the details. 

A bench meter should have a higher-than-ambient cold junction.  I don't see how it could be 6C lower than ambient unless it had been in a colder environment immediately before testing.  Are you measuring this just by shorting the inputs?  Can you try what I suggested earlier?  Also, you might try putting a probe in a glass of water to guarantee a stable temperature.  Then take a reading with the Brymen and your best "cheap meter" and record both the temperature displayed and the voltage in mVDC.  Then do the same with the Siglent, using the "ALL" display in the temp mode to display the mVDC and then using the mVDC range to read voltage directly.  Then install a short and again read temperature and mVDC on the same range as used with the probe.

[BillyO]

Are you measuring this just by shorting the inputs?

Yes.

Can you try what I suggested earlier?  Also, you might try putting a probe in a glass of water to guarantee a stable temperature.  Then take a reading with the Brymen and your best "cheap meter" and record both the temperature displayed and the voltage in mVDC.  Then do the same with the Siglent, using the "ALL" display in the temp mode to display the mVDC and then using the mVDC range to read voltage directly.  Then install a short and again read temperature and mVDC on the same range as used with the probe.

Yes, but it will take a couple of days.  Band practice tonight and out of town tomorrow.

[bdunham7]

Here's the Brymen with 4 of the better cheapies.  It agrees fairly well with these ones.  Is this evidence that I need to work with adjusting the Siglent?

I missed this post, it might have saved me a bit of typing!

By all means do the additional tests so we can figure out how it all adds up.  However, I'll take a preliminary guess that the discrepancy is due to the difficulty of doing a proper CJ comp in a bench meter.  I'm thinking that they made some tweaks in software regarding the CJ comp and those were sorted out for normal room temperatures--say 18-28C--but don't work so well in your environment. 

[G0HZU]

I've always compared my two temperature meters by putting both probe tips in the same cup of water. Maybe there are reasons this method is not ideal but it's a lot better than doing it with the probes left in free air. If one probe has been touched and has moisture on it, the temperature will be lower for quite a while due to evaporation etc. So it's easy to get confusing results in free air.

I've used the ice fragments in water method to get close to 0 degC and used boiling water to get close to 100degC.

The water can also be any temperature inbetween when comparing. I generally get very good agreement with these methods.
One meter is a Digitron 2751K and the other is a DMM that came with its own thermocouple.

[bdunham7]

I've always compared my two temperature meters by putting both probe tips in the same cup of water.

Close to room temperature that works well, but colder or hotter you'll see significant differences unless you stir.  In the previous post, as the kettle was cooling I'd see >5C differences that went away when I gave it a good stir.

I've used the ice fragments in water method to get close to 0 degC and used boiling water to get close to 100degC.

Distilled, degassed (boil before freezing) water and ice improve on that.  As for boiling, you should be able to guess my altitude within 50M.

[G0HZU]

I usually hold the two tiny thermocouple probe tips so they are face to face in the cup of water and are virtually kissing each other when using the (cold or ambient or warm) cup of water. This gives repeatable results as they are occupying the same droplet sized area of water in the cup.

[BillyO]

Can you try what I suggested earlier?  Also, you might try putting a probe in a glass of water to guarantee a stable temperature.  Then take a reading with the Brymen and your best "cheap meter" and record both the temperature displayed and the voltage in mVDC.  Then do the same with the Siglent, using the "ALL" display in the temp mode to display the mVDC and then using the mVDC range to read voltage directly.  Then install a short and again read temperature and mVDC on the same range as used with the probe.

The mug of water was left out over night.  Same probe was used with all and was taped in place.  I have not turned the Siglent off since I took the first picture.



For some reason the Kaiweets is not in as close agreement with the Brymen as before. 

[bdunham7]

For some reason the Kaiweets is not in as close agreement with the Brymen as before. 

You'll see variations like that which is why even though you might often see closer agreement, you have to figure +/-1C as reasonable tolerance.  Things just move around a bit.

I assume from what I'm seeing that your room air temperature is about 13.8C, give or take a bit?  So what we're seeing is that the Siglent agrees with the Brymen as to the measured voltage and the short is effective enough, but there's a bit discrepancy between the internally measured cold junction temperature and the actual effective cold junction temp--the latter being a bit of a complex topic.  In your cold room, the internal measurement is probably correct for where it is but then there is likely a large temperature gradient between that point and the actual terminals.  I think this will be more pronounced in a cold environment and your results seem to indicate that the actual effective cold junction temp is about 14C.  Try putting some insulating materials--cloth, bubble wrap, some masking tape--to cover and insulate the entire right front quarter of the meter including top and bottom.  Seal it up as best you can and wait a few hours.  I'll bet that if you put two probes in your cup tied together near the end, one to the Siglent and one to the Brymen, that after a few hours your readings will be much closer. 

[Poroit]

I assume you are aware of the Brymen's & Siglent's accuracy specs.

Temperature (Models 789, 786 only) RANGE -200.0C to 1090C Accuracy 1) 2) 1.0%+1.0C

Type K Siglent  -100 °C ~ 1372 °C 0.5 °C .(Page 7 of the attached).

It looks like your Siglent needs a tweek. It looks a reasonably good unit so worth doing it you want it as your Reference Meter.

Your Siglent also has a 4 Wire RTD Input. That is much more accurate for low temperature measurements.

[BillyO]

I assume from what I'm seeing that your room air temperature is about 13.8C, give or take a bit?

Yes, around about that this time of year.  I'll try what you suggest sometime, bt for now I'll just reply on the Brymen and/or tweak the Siglent.

[bdunham7]

Type K Siglent  -100 °C ~ 1372 °C 0.5 °C .(Page 7 of the attached).

It looks like your Siglent needs a tweek. It looks a reasonably good unit so worth doing it you want it as your Reference Meter.

Please refer to note 3 below that spec:

[3] Relative to cold junction temperature, accuracy is based on ITS-90. Built-in cold junction temperature refers to the
temperature inside the banana jack and its accuracy is ± 3.5 ℃

The Siglent is only going to be the reference meter here if he gets a 4W RTD.  The TC accuracy won't ever be consistent over any range of ambient temperatures.

[bdunham7]

I'll try what you suggest sometime, bt for now I'll just reply on the Brymen and/or tweak the Siglent.

You can do it if you want to satisfy your curiosity and verify that it will do what I said, but I don't think there's much you can actually do to fix the issue.  I would not "tweak" it because that can't fix the issue of varying temp gradients.  I thought that perhaps Siglent already had a "tweak", but it looks like they just calibrate it using the Ext Ref setting and avoid the problem entirely.  That seems pointless for regular bench use.  Either just forget about using the bench meter with a TC or get the 4W platinum RTD I mentioned earlier.

[BillyO]

Either just forget about using the bench meter with a TC or get the 4W platinum RTD I mentioned earlier.

For now I'll forget it, but will eventually look into the 4W RTD.

[hpw]

I'll try what you suggest sometime, bt for now I'll just reply on the Brymen and/or tweak the Siglent.

You can do it if you want to satisfy your curiosity and verify that it will do what I said, but I don't think there's much you can actually do to fix the issue.  I would not "tweak" it because that can't fix the issue of varying temp gradients.  I thought that perhaps Siglent already had a "tweak", but it looks like they just calibrate it using the Ext Ref setting and avoid the problem entirely.  That seems pointless for regular bench use.  Either just forget about using the bench meter with a TC or get the 4W platinum RTD I mentioned earlier.

I use with my SDM3065 & plot over time using a chip PT1000 resistor at up to 100°c and glue them on the surface...

Note on PT100 V PT1000:

The main difference between Pt100s and Pt1000s in general is the electrical resistance at 0⁰C, which is the number in the name: a Pt100 is 100Ω at 0⁰C and a Pt1000 is 1000Ω at ⁰C. This makes Pt1000s more accurate for small temperature changes as they would result in larger changes in resistance when compared to Pt100s

[David Hess]

My experience is that uncalibrated thermocouple measurements have horrible accuracy.  The specification given for the meter represents the best case with a perfect thermocouple.

I have been calibrating my own home-built thermocouple probes and thermocouple converters in pairs, but this requires a thermocouple converter which allows user calibration, like an old Fluke 80TK.  B&K used to have a multimeter with thermocouple socket and offset adjustment built in.

I measure absolute atmospheric pressure with my sensor GPS unit so that I can use a boiling water bath as a 100C reference.

[tautech]

My experience is that uncalibrated thermocouple measurements have horrible accuracy.  The specification given for the meter represents the best case with a perfect thermocouple.

I have been calibrating my own home-built thermocouple probes and thermocouple converters in pairs, but this requires a thermocouple converter which allows user calibration, like an old Fluke 80TK.  B&K used to have a multimeter with thermocouple socket and offset adjustment built in.

I measure absolute atmospheric pressure with my sensor GPS unit so that I can use a boiling water bath as a 100C reference.

 

Who'd have guessed Temp measurements are another rabbit hole ?

[David Hess]

My experience is that uncalibrated thermocouple measurements have horrible accuracy.  The specification given for the meter represents the best case with a perfect thermocouple.

I have been calibrating my own home-built thermocouple probes and thermocouple converters in pairs, but this requires a thermocouple converter which allows user calibration, like an old Fluke 80TK.  B&K used to have a multimeter with thermocouple socket and offset adjustment built in.

I measure absolute atmospheric pressure with my sensor GPS unit so that I can use a boiling water bath as a 100C reference.

Who'd have guessed Temp measurements are another rabbit hole ?

I know not what you speak of.

I estimate that my calibration was good to within 0.25 degrees Fahrenheit, where I doubt that I was previously within even 5 degrees.

I needed the baffle because with the thermocouple tip exposed, the response was so fast that it was measuring the temperature of the steam bubbles in the water.

[IanB]

I needed the baffle because with the thermocouple tip exposed, the response was so fast that it was measuring the temperature of the steam bubbles in the water.

Where did you get those red plastic probe supports? Are they a standard kitchen item for sugar thermometers, possibly?