Fluke 54 II B Digital Thermometer reads very high from actual temp [SOLVED]

[enut11]

I have acquired a Fluke 54 II digital thermometer in good physical condition but with an unusual fault - it reads about 40degC high on both inputs. Tried 2 different thermocouples with same result. I have posted it here reading room temperature ahead of a full search on the possible fault.

TiN has done some work on Fluke thermometers and I will be reading this thread shortly:
https://www.eevblog.com/forum/testgear/fluke-5x-ii-series-thermometer-tear-down-and-hacks/

Meanwhile, any ideas on possible causes?
enut11

[mendip_discovery]

Random thought, could the CJ sensor be faulty. But that would normally cover for 20°C ish.

Is it 40°C error linear ? It maybe an offset or two.

Have you got access to a mV source? Or a temp calibrator?

[enut11]

Yes, I have an accurate millivolt source and a HP34401A 6.5D DMM. I guess this has something to do with calibration? I will try and chase up a service manual.
The meter accommodates a maximum offset of +/-5 deg C.

[mendip_discovery]

Using this, https://www.pyromation.com/Downloads/Data/emfk_c.pdf

You should be able to generate a few temps and then use that to work out if the error is 40C through the range, or 40 at 20, 50 at 30 etc. If you use a good thermometer you can use that to guess the Cold Junction Temp. Use some low emf copper connectors, you can buy them in k-type (often white in colour).

I know you can use Ice and Hot water to do the checks as well but doing with mV rules out the probe.

If you put a good copper wire short across the terminals you should see what the units 0C with a CJ of 0C is, which usually is what the meter things the CJ is.

[JimLev]

I’ve got the Fluke 52 II, looks almost the same as your 54 II B except yours has a recall and logging button.
Have you gone into the setup? There is an offset setting. Do you have the manual?
 

[floobydust]

Fluke 51,52,52,54 Series II Thermometer Service Manual does not have schematics.  I'd uploaded it to BAMA long ago but most of the batch did not get posted.
You'd have to take it apart and post pics to see what the CJ sensor is, assuming it is broken.

You could check the calibration, for both the cold-junction and A/D.
Reference Junction CAL:
Type K thermocouple in ice bath.
Enter CAL mode (poke the tiny CAL button hole on the backside), LCD shows one of 4 menus: CAL A-d, CAL rEF, CAL donE, CAL Abrt.

Select CAL rEF, press enter, arrow keys up/down until temperature matches known temperature, press enter, shows CAL donE, press enter to store.
Or up/down arrows to CAL Abrt, press enter, to abort cal changes.
Calibrating the A to D needs a precision voltage source where you can dial in +80.000mV and -8.0000mV

There's an offset feature too in SETUP you could see if it's zero.

[enut11]

I’ve got the Fluke 52 II, looks almost the same as your 54 II B except yours has a recall and logging button.
Have you gone into the setup? There is an offset setting. Do you have the manual?

Maximum offset capability is +/- 5C

[enut11]

Fluke 51,52,52,54 Series II Thermometer Service Manual does not have schematics.  I'd uploaded it to BAMA long ago but most of the batch did not get posted.
You'd have to take it apart and post pics to see what the CJ sensor is, assuming it is broken.

You could check the calibration, for both the cold-junction and A/D.
Reference Junction CAL:
Type K thermocouple in ice bath.
Enter CAL mode (poke the tiny CAL button hole on the backside), LCD shows one of 4 menus: CAL A-d, CAL rEF, CAL donE, CAL Abrt.

Select CAL rEF, press enter, arrow keys up/down until temperature matches known temperature, press enter, shows CAL donE, press enter to store.
Or up/down arrows to CAL Abrt, press enter, to abort cal changes.
Calibrating the A to D needs a precision voltage source where you can dial in +80.000mV and -8.0000mV

There's an offset feature too in SETUP you could see if it's zero.

Internal pic attached. Which is the CJ (Cold Junction?) Edit: found it.

[enut11]

More pix

[floobydust]

I see thermistor RT1 near R21-R24 and pin 90 of MCU. It doesn't make sense to use a possible MCU single-slope A/D for that, they must be switching-in different resistors for best resolution within a range. It's odd, with the zillion mux's on the board just use the 24-bit A/D to read the thermistor. Try warm RT1 up, see what the readings/voltages TP14/TP15 do.

I would make sure if it's just an offset in readings or more - such as the inability to read thermocouple negative voltages i.e. below freezing temperatures.
I made a thermocouple simulator with Wheatstone bridge to generate say +/-100mV for testing these things. The power supply voltages I would check, ensure TP11 has the -ve rail; not sure if the MCU runs off 3.3V or 5V here.

[TiN]

CJ sensor is on copper-clad interface PCB (with connector pads for thermocouples).
Probably calibration is borked. Or check for broken connections between copper interface PCB and main PCB, I've seen few times those pads on pin header get open. Don't just visually look at it, broken connection can have hair-line crack and that's enough. Instead beep beep connections with DMM.

My old hacking post:

https://xdevs.com/article/fluke_51ii/

Calibration process:

https://xdevs.com/guide/cal_fluke52_2/

[enut11]

Thanks @TiN. Checked connections as per your suggestion and all is ok in this area.

@floobydust, heating RT1 made no diff to display reading. I injected 0-1mV into ports and the response was linear but or course too high by about 40C.
Power supply for the meter is 4.5v. All measurements were below this.
I have yet to check the AD response
enut11

[enut11]

This should better define the problem with the Fluke 54 II B thermometer.
A direct comparison with 2 other thermometers set to read an oven cooling from about 43C to 25C.
There is also a reading taken in a glass of melting ice water.

As you can see from the data, the F54 is consistently higher by ~39C.

[enut11]

The AD test resulted in a low output: 7965 vs 8000 expected. EDIT: THE VOLTAGE SOURCE WAS NOT STABLE ENOUGH
Faulty AD chip??

The meter also failed all of the mV vs Temp compliance checks by 40-60C.

[floobydust]

I always start a H/W repair measuring the power supply voltages, and here the LTC2400 A/D voltage reference seen at pin 2 would be something additional to check. How's that TP11?
Although it looks like the CJC is off, there must be an A/D offset seen. That could be the op-amp, voltage ref, CMOS switches etc. and a lot of hardware this beast has inside.

[enut11]

Before poking around measuring voltages that I had no idea what to expect, I decided to do a proper calibration from p9 of the Fluke service manual.

One has to be careful with the signal, especially when the red wire in both Tye K and Type T thermocouples is negative polarity.

Calibration calls for a signal of +80mV and -8mV be applied to the thermometer AD converter. This time, using one of my LM399 Precision Voltage References and a Kelvin Varley divider, I was able to generate accurate levels and ended up with a result of '7997' against an aim of ~8000.

The next step was to calibrate the Reference Junction with a known temperature between 0-50C. Melting crushed ice is the preferred way for an accurate 0C.

Well, I am happy to report that the +40C offset error has gone and am now in the process of checking accuracy over a range of temperatures. It is still a mystery how the offset occurred in the first place.

A lesson for me here was that these instruments need microvolt accuracy test signals.

Thanks to Forum members who contributed to the solution. I for one have learned a lot more about thermometers and thermocouples in the process of this 'repair'.
enut11

[floobydust]

Glad you have it working, it could be a flaky EEPROM with corrupted CAL values for CJC, or a cal lab/user that didn't know what they were doing and bungled it.
I am always suspicious of hardware problems that can sometimes get "calibrated out", because such a large shift is unusual, and for both channels. I would keep an eye on it. I didn't see pics of the CJC sensor(s).

[enut11]

Thanks @floobydust. Maybe we will never know. Now that it appears to be working I am reluctant to pull it apart again for more pix.

Follow-up test after re-calibrating the F54 shows the 2 Fluke thermometers tracking within 0.1C which is nice. Horizontal scale is minutes.

BTW, if you ever come across service info that is missing from the official Fluke 51-54 thermometers manual, I would appreciate a PM.
enut11