NI 9215 Voltage measurement in a welder problems with noise

[CrownLouis]

Hello,

I am currently using a NI 9215 with screw terminals to measure voltage in a MIG welder, also I am using a voltage divider to reduce the magnitude of the voltage to fit in the measurement range of the NI 9215 (+/- 10V), as result I am getting wrong measurements, I tested the torch with a multimeter to obtain the voltage getting aroung 18V while the NI 9215 is giving me a measure of 40V and the wave form is not the want that I am expecting to be.

I already connected AI- to COM and also I connected a 1MOhms resistor, not at the same time,  without succes.

What am I missing here? Thanks in advance.

Louis

"Help me Obi-Wan Kenobi, you're my only hope"

[johnsold]

Louis,

Most likely you are seeing the results of differences in the way the instruments make measurements, plus possibly grounding problems.

Your multimeter probably does some processing in the analog domain and may use a dual slope digitizing process. Both of these tend to reduce the amount of noise appearing at the output. The NI 9215 takes samples at rates of 100000 samples per second or faster, depending on how many channels are being used. That is fast enough to capture much of the noise.

If you are using a battery operated multimeter which is not connected to the computer, grouding is not an issue. With the computer and DAQ device there is a significant possibility of ground loops.  Even though the NI 9219 has isolated inputs, you must use care to be sure that the input voltage limits are observed and that you do not create a ground loop due to the way things (like the voltage divider) are connected.

Lynn

[SteveD123]

What, EXACTLY, are you trying to measure?

Welding voltage is a very poor proxy for welding power. For a start you have to cope with the decrease in voltage when the arc is struck.

I know next to nothing about welders. However, I can be pretty sure that the waveform when welding will be horrendously spiky. If you are not trying to analyse the waveform harmonics, then a true RMS-to-DC converter module will do what your multimeter is doing. The first hit on google is : https://www.dataforth.com/model.view.aspx?modelid=20 .

However, the voltage is not likely to contain the information you want. The welding transformer adds a lot of reactance to the circuit, so the current waveform will be out of phase with the voltage. If what you are interested in is the amount of energy dissipated in the weld, (e.g. to look at power input vs welding speed and size of the heat affected zone) then you need to monitor the current as well. This can be done with a series resistor or a current transformer. If using a CT, make sure that it is a wideband type if you want to capture the complex welding harmonics.

Of course, electrode voltage might be exactly what you want, if the idea is to maintain a constant arc length. Here harmonic analysis of the electrode voltage may be useful as well as just the voltage. However, you may need to process out the high speed arc fluctuations.

Normal precautions apply: Use signal isolators (e.g. the dataforth link above)(but check bandwidth specifications), and have fuses in every instrumentation wire that connects to the welding circuit, as close to the pick-off point as possible. This is so that you do not get welding current (10's to 1000's of amps) going down wires that are only sized for a few mA of instrumentation current.