HIGH FREQUENCIES BETWEEN GROUNDS
REFERENCE > ADVANCED TECHNIQUES >

Lets imagine your computer is at ground G1, the instruNet device is at ground G2, and the device under test chasis is at ground G3. The AC and DC voltage between these 3 should be 0V, in theory, if they are all connected to earth ground through the 3rd prong of their power cords. However, when digital electronics is involved, this is seldom the case. This is primarily due to the >15,000,000 transisitors on the computer motherboard that switch >200,000,000 times per second with a <0.000000003 second rise time (nasty!). These, along with radio stations, create electromagnetic fields that radiate through the air (which are sometimes referred to as RFI, or "radio frequency interference") into a loop of wire (e.g. instruNet cable, sensor leads, power cord) and induce a high frequency low level voltage in that wire (e.g. <10mVrms, >100MHz). Another similar issue is the digital switching between two boxes (e.g. computer mother board and external instrunet hardwre) which cause current to flow on those wires when they switch (e.g. for 3nSec), which induces a voltage drop on both the common and digital signals wires during the switching duration, which means you get a high frequency between the grounds of these two devices.

How could this cause a problem if I am doing a differential voltage measurement which in theory should subtract out a similar high frequency on both input wires, or I turned on the analog low pass filter within the instruNet?
If your instruNet device does not contain an internal RFI filter (e.g. the i100 does not) which rejects > 100MHz signals, then those very high frequencies can leak into the instruNet electronics and cause erradic < ±100uV errors. This is due to internal amplifers that are based on internal feedback loops that process correctly the lower frequences (e.g. < 100MHz), yet not the higher frequencies.

How can I determine if this is causing a problem?
If you are working with low level signals (e.g. < 10mV) and the readings are eradic or incorrect, your problem might be related to high frequencies between grounds.

What to do about it ?
Below are several things to try:

• If working with a thermocoule, install an additional wire between Vin- and GND, as noted here; to reduce the AC signal between the instruNet GND and the instruNet Vin screw terminals.
• If working with a strain gage connected to the i100 (not with the i420/i430/i60x 12), add 2 capacitors, as noted here.
• Consider working with electrically isolated i601.
• Consider installing an i330 optical isolator, as noted here.
• Consider electrically isolating your sensor ground from your device-under-test (dut) ground, if possible, as noted here.
• Consider adding a capacitor to reduce rfi if working with the i100 (not with the i420/i430/i60x 12), as noted here.