LM35 temp. sensor stop working properly when 24 DC Power supply is turned on

[rodriferfe]

Hi all,

I am using a LM35 sensor to monitor the room temperature. I have a USB DAQ 6218 to read the voltage in the RSE mode and I power the sensor from one of the DAQ pins 5 V. I average 10 points and display the result which is smooth. The thing is that I am using the USB DAQ to control some pinch valves (digital outputs) and other equipment (analog inputs and outputs) which are powered with a 24 V DC power supply. The problem is when I turn on the power supply the temperarure sensor signal becomes very noise with spikes. Also when I add a channel to the DAQ assitant to measure the current as well as other stuff, the measures are wrong, the temperature is quite lower than it should be. I have attached and image in which the DC power supply was switched for like 20 s, it can be clearly seen when was switched on and off. Any idea what might be causing the problem and how to overcome it? 

Thank you very much

[Hooovahh]

I don't fully understand your setup, but one of your issues sounds like ghosting:

http://digital.ni.com/public.nsf/allkb/73CB0FB296814E2286256FFD00028DDF

Your hardware only has one analog to digital converter, and if you have a DAQ task that has more than one channel in it, it will switch or Mux that ADC on all the signals you want to sample really fast.  The issue with this is if the lines don't have matching impedances, or something to pull it quickly to the voltage level, then your signals will somewhat ghost one another, where the read value will be betwee the actual, and the signal you just read.  To help with this you can adjust your sampling rate, or read the same signal twice.  There are also things you can do to the circuit to have less of an impedance difference.  If possible you can also perform two seperate finite sample reads on different channels.

[rodriferfe]

Thanks for your answer Hooovahh,

No that is not my problem. My main problem is that just when I switch on the 24 PS you see what happens in the document attached. Bare in mind that in the document attached I was just measuring the temperature, I was not measuring any other thing so ghostling was not possible. It is like the 24 PS is interfering with my temperature measurements in some way. The relation between the 24 PS and the USB DAQ is just that I use such PS to power some devices that I control and measure with the USB DAQ. So I do not understand where is the problem, I am quite puzzled.

Thank you very much

[Henrik_Volkers]

Smells like a EMC problem 😉

Can you provide a schematic of your setup including all wires /connections  with length and the grounding/connections of the DUT, the DAQ and the PC? 

Groundloops, noisy power supplies and high impedances ...  try to avoid them.

[mtwar]

Hello!

Have you tried to change RSE to differential measurement? If not, please try it and show us the output.

The other thing is, can you use laptop to your measurements? If so, do it and unplug your notebook from the power supply - leave it on battery. What are the measurements in that case? It is possible (it has happened to me once) that power supply is interfering with the power line and it was interfering with my notebook's USB power supply. It is a guess but maybe?

Marcin

[Hooovahh]

---

@rodriferfe wrote:

Also when I add a channel to the DAQ assitant to measure the current as well as other stuff, the measures are wrong, the temperature is quite lower than it should be.

---

That sounds exactly like what NI described as ghosting in the document I linked to which is why I suggested it.

[Sam_Sharp]

Yeah - apart from the obvious ghosting as others have suggested (does the temperature signal follow your other signals?), I would definitely be looking at noise/grounding issues.

The low-cost USB devices share their ground with the USB bus, they are not isolated. If your USB ground is noisy, you will have noise issues - hence the suggestion of trying it on a laptop running off of a battery. As others have said - if you post a detailed schematic showing the connections, wire lengths etc. then we may be able to help you pinpoint the source of the noise.

[rodriferfe]

Hi guys,

I removed every single cable fron the USB DAQ except for the ones I need in order to measure the temperature (AI5, AI GND, 5V) but the problem persists when I turn on the 24 VDC PS.

The 24 VDC powers a breadboard where I have some valve drives and some other stuff. The 24 VDC is pluged to the grid. The USB DAQ is connected to a USB HUD as I have many other instruments that connect via USB (all of them are turned off right now) and they are also pluged to the grid. So basically the only connection between the USB DAQ and the 24 PS is trough the grid-USB HUD. I have made a measurement in which 1) 24 PS is on and everyhing is pluged to the grid, 2) the 24 PS is on and everything is unpluged and 3) the 24 PS is off and everything is unpluged (see file attached).

So it is clear that there is a ground problem. Obiously the fluctuations compensate each other more or less so averaging 100 data points instead of 10 helps which would be the fastest solution. In any case as I am out of deapth here I would like to have your feedback about if there is a better solution to solve the root of the problem.

Thank you very much to you all.

[Henrik_Volkers]

Again:

What exact type of DAQ do you have? BNC/Screw?

How have you exactly wired your setup? 

(Every wire!  A wire that carry currents will create a voltage!)

Differential measurements can help....    if applied correctly 😉

Do you have a scope ?

[rodriferfe]

Thanks for all your answers guys. Finally I reinstalled one of the unit were missing and the problem disminished, now I do not get such spikes. Using differential or RSE made no difference. Also when measuring 4 analog inputs at the same time I was getting a wrong temperature measure (15 instead of around 22). I believed that the latter problem might be due to ghosting suggested by Hoovahh above. The temperature was setup to be read the last of the 4 and I changed it to be the first one and it solved the problem.

Thank you very much