differential measurement and manually grounding a floating signal

[qwerty257]

Hi All,

99% of power supplies these days offer floating outputs. DAQ cards offer the user the option of measuring floating or grounded signals. With so much flexibility i am overloaded with choice as to how to connect my circuit.

As a result i have some questions

1) If channel count is not an issue does differential mode offer the best performance/benefits of all the types of mode?.

2) Is it peoples preference just to manually ground the signal source (connect it to earth ground) rather than faff about with bias resistors when using differential mode? This would seem the easiest soloution to me. I cant think of any situation where this wouldnt work.

3) To establish if an outout is floating can i simply take a multimeter and measure for a short between earth ground? If no connection exists it is safe to say the output is floating?

E.g

Switch mode PSU with floating output powers a sensor of some sort. The signal output from sensor is not connected to earth ground. Connect the common of the signal to earth ground and measure using differential mode. What problems might i run into ?

thanks

[johnsold]

1) Differential mode works best with floating signals. It offers no advantage with grounded signals. (One input will be grounded.)

2) Earth ground is usually a safety ground and may or may not be a good signal reference.  It is better to connect your ungrounded signal source (includiing flaoting power supply)  to AIGround at the DAQ device.  The computer power supply (except for laptops) is probably connected to earth ground fro safety.

3) NO. If you measure the voltage, you may find tens to hundresd of millivolts, both DC and AC, between two points whach are connected to earth ground.  Electric Codes allow up to some volts under fault conditions.  If you try to measure resistance, any voltage across the two points or any currents flowing through the meter may produce very strange readings.  I have seen negative resistances reported when trying this.  You could also destroy your meter.  If the power is off on all the equipment, a resistance measurement might tell you something.

If you have only one sensor, you probably will get it to work. As soon as you add a second sensor, you need to start thinking about ground loops and other signal integrity issues.

Lynn

[crossrulz]

Most systems have a common ground.  We typically just tie the negative of the power supply to the common ground.  You can then simply use the single ended terminal.  It is by far the simplest method and the most realistic real-world simulation as well.

[qwerty257]

I have been looking at the NI paper on analogue signals and noise and it shows two methods of using differential signals. One using bias resistors and one where the signal source is tied to earth ground.

Am i correct to think regardless of what method is used there must be a refrence to earth ground? either on the signal source or DAQ device.

[johnsold]

There must be a connection to AI ground which is probably connected to earth ground somewhere.  As I pointed out above I dislike the term "earth ground" because it tends to mean different things in different contexts. For an old radio operator earth ground brings up images of 8 foot long copper rods driven onto the earth at the base of an antenna.

Remember that currents flow in loops. So even if the current at the input of the DAQ device is picoamperes, a complete loop needs to occur somewhere. By connecting everything in the system to a common reference point (which may or may not be earth ground), and preferably connecting each component to that reference point by only one conductor, the possibility of stray or unintentional current due to stray impednaces is minimized.

Lynn

[qwerty257]

I think this may be where i am going wrong. AIGND is connected to earth on my system. The difference between A+ / A- is referenced to AIGND. Therefore if i connect my signal source to earth am i creating an earth loop. Unless i can somehow remove the connection to earth from the DAQ side?

[qwerty257]

attached diagram

[johnsold]

This looks like a configuration where you need the bias resistors. Use equal value resistors. One from +Sig to AIGnd and one from -Sig to AIGnd.  Typical values might be 1 megohm.

Lynn