Voltage Measurement Across Externally Excited Thermistor

[leo.su]

Hello,

I am attempting to collect temperature data by measuring the voltage across a temperature input terminal of a piece of industrial equipment. A thermistor is currently wired to the terminal, and my USB-6212 analog input terminal is connected in parallel. The equipment needs to remain functional, so the thermistor cannot be removed.

The problem is, the USB-6212 significantly disturbs the measurement - for example, the equipment's temperature reading spikes from 45degF to 60degF whenever the USB-6212 is connected. Using a Fluke DMM does not cause the same problem. I tried measuring the analog input impedance for the USB-6212 with a Fluke 87-III. I get out of range (>50M Ohm) with the 6212 turned on and ~12M Ohm when off. This seems to meet specifications and should be high enough to not disturb the measurement. I don't know the rating of the thermistor, but estimate it is around 20K Ohm based on DMM voltage measurements with and without parallel resistors.

Can anyone suggest a way to fix this problem, or an alternative way of making the measurement? The primary constraint is that the thermistor must remain in the temperature terminal.

Thanks,

Leo

[johnsold · Accepted solution]

Leo,

This is only  a guess, although a somewhat educated guess.

Two clues: Measurement with a DMM does not distuirb the result. Measurement with the USB-6212 disturbs the result.

The DMM is most likely battery powered and not connected to any ground reference. The USB device is definitely connected to the USB ground which in turn is connected to the computer ground.

My guess: A ground loop between the DAQ/USB/computer and the industrial equipment. Unless the industrial equipment specifies that  one end of the thermistor is grounded and the equipment and computer grounds are connected to the same ground point, it will be difficult to make a measurement as you have described.

Things to try:

1. Using the DMM measure the voltage (both AC and DC) between each end of the thermistor and the equipment ground.

2. With the computer/DAQ disconnected from the equipment measure (DMM) the voltages between the computer or DAQ ground and the equipment ground.

3. If all the voltages measured in 1 and 2 are within the common mode specification for input voltages to the DAQ device, then try a differential connection with suitable bias resistors to ground/

Do you have a technical manual or schematics for the industrail equipment?

Lynn

[leo.su]

Thank you Lynn. I had not considered grounding. Unfortunately I do not have wiring information for the equipment control board (which contains the temperature input terminals).

However, I will make your suggested measurements on Monday and post my results. I'm not constrained on the number of USB-6212 ports, so I can use either differential or single-ended as necessary.

Leo

[Henrik_Volkers]

If you measure multiple channels with the 6212 I highly recommend a seperate differential amplifier as a buffer.

Multiplexed inputs (like the 6212 have) are prone to output switching noise. For each sample the (small) sample&hold capacitor is switched in parallel to the thermistor, with the result of small current spikes that can make trouble!

But, as Lynn recommended, also check the voltages from the thermistor pins to AGND of the 6262!  Do this with your handheld DMM and measure in DC and AC mode!

[leo.su]

Here are some interesting results...

It seems like the thermistor is grounded to the equipment ground.

There was no voltage between the thermistor (-) terminal the the equipment ground, either AC nor DC.

There was a voltage (~100mVDC, 10VAC) between the equipment ground and 6212 "ground", and the same voltage between the equipment ground and all 6212 terminals.

I say "ground" in quotations because the computer attached to the 6212 is a laptop. I checked the power plug and it didn't have a 3rd prong, perhaps that means the laptop is "grounded" to the neutral.

I hooked up the thermistors to a differential terminal on the 6212, and, as before, the measurements were disturbed. However, I then unplugged the laptop (running on battery), and the measurements were no longer disturbed. Also, when the laptop was plugged in, I measured ~1M Ohm across the differential terminals. After the laptop was unplugged, the resistance was >50M Ohm.

I reasoned that the laptop power connection must be introducing noise into not just the 6212 ground, but all 6212 terminals. Of course, I can't run on battery permanently, so my solution was to cut open the USB cable, extract the 5VDC pair, and connect those to a 5VDC source directly on the equipment control panel (grounded to the equipment ground). I left the data wires in the USB cable so it can still communicate with the computer. The measurement is no longer disturbed, regardless of whether the laptop is plugged in.

[leo.su]

While I am able to make measurements now, there is an additional issue.

If the USB (now data only, no power) is disconnected, the 6212 turns "off", and its input resistance (and maybe other characteristics) changes. When this happens, the measurement (seen by the equipment) is again disturbed.

In my particular setup this is a problem, because if the computer were to suddenly crash or restart, the equipment's reading of the thermistors will become incorrect. In this case, it becomes incorrect in such a way that damage will occur to the equipment.

I had hoped that connecting the 5VDC pairs of the USB to the equipment controller, the 6212 would stay "on" even if the computer crashes. Turns out not to be the case.

I still have to do more searching, but does anyone know how to keep the 6212 "on" with power but no data?

[johnsold]

Now, that is scary!

---

...my solution was to cut open the USB cable, extract the 5VDC pair, and connect those to a 5VDC source directly...

---

1. The USB port expects to provide some power to the external device. When the device is first connected (or the host powers up), the host and external device negotiate how much power will be provided to the external device. I have no idea how the host will behave when the power consumed by the device does not match the negotiated value.

2. You now have the 10 VAC superimposed on the USB data signals! It did not go away, it is just hidden where it is not so easily measured. But it is also now in a position where it may be more likely to corrupt data or damage equipment.

3. It is unlikely that the laptop ground is connected to the neutral. WIth the widespread use of universal power supplies the supplies must work on power sources which do not have a grounded neutral. I suspect that what you have is mostly inductive or capacitive coupling between the power mains and the laptop ground.  This does not mean that the amount of energy is too low to damage equipment. It certainly is large enough to disrupt your data.

4. The maximum input rating (signal + common mode) (which probably includes the voltages to ground) for the USB-6212 is 10.4 V. If the voltage on the thermistor plus the "inter-ground" voltage exceeds that, malfunction or damage can occur.

What can you do about it?

The best solution to the ground issue would probably be to use an isolated device such as the USB-6216. Even that has 1200 ohms to AI GND when power is off.

When you have a situation where failure of part of the measurement system can do damage to other equpiment, you need to take extra precautions. The exact nature of those precautions depends on how fast the system must detect and react to a failure. If the times can be > ~100 ms, then a simple relay which disconnects the DAQ system from the other equipment in the event of a failure may be enough. For faster responses a much more carefully designed system will be required.

If you can tolerate the attenuation, connecting series 1 megohm resistors to both analog inputs and connecting a 200 kilohm resistor and a 100 nF capacitor across the input terminals might provide the power off impedance and some power line frequency filtering.

Lynn

[Henrik_Volkers]

Never touch a running system !

Seems that is VERY wise to use  two different thermistors 🙂

However, if you really need to hook to the single one.: Nail the USB (&notebook) to the same ground, add a differential buffer amplifier with seperate power (maybe the controller of that device provide some extra volt with some mA)  so that the power of the notebook & USB has no influence.

BTW: What controller is reading the thermistor?  Maybe you can read the temp via a digital connection (RS232, RS484, TCP ...) ?

[leo.su]

Lynn,

Thanks for the heads up. I didn't think the voltage would be superimposed over the data channels...

For the disturbance issue, I got another AC adapter for the laptop, this one with a ground prong. Now the normal (unspliced) USB cable works even when the laptop is plugged in. I'm reading zero VDC and 46mVAC between the equipment and 6212 grounds, much less than before. I'm using a differential channel on the 6212 for the thermistor readings, and it doesn't disturb the measurements.

As for the disconnect, I think the 6212 doesn't have enough current output to drive a relay. I'm thinking maybe a transistor of some sort... The equipment will be fine with reaction times of 15 or 30 seconds. It is a HVAC Chiller that produces cold water for cooling. An incorrect temperature reading might cause it to freeze the water in the heat exchangers, but that doesn't happen instantly.

Henrik,

Thanks for the tip. You are right that there are other control protocols for the equipment (Modbus and BACnet through IP and RS-485). These are already configured for use with the building's central system. For various reasons I can't access the data points through the central system.

Cheers,

Leo

[johnsold]

Leo,

The ULN200x famliy of devices make very nice interfaces between logic outputs and relay coils. They have high current transistors and protective diodes built in.   There are several versions for different drive voltages (TTL, CMOS, ...). I do not recall off the top of my head which one is appropriate for your system.

Lynn