Using NI USB-6351 +5V as excitation for a full bridge load cell

[luissimm]

Hello,

I need to find a power supply for a specific load cell (120mV/V) without amplifier, and I have NI USB-6351 to do that. So I was considering using the +5V supply from pin 96 as load cell excitation. My question is, "Is this +5V stable enough?".

And if it's not, what about acquiring 2 channels (CH0 acquiring direct +5V and CH1 acquiring load cell output) and make a differentiation between them in order to cancel any noise in the +5V power supply.

Obs.: My acquisition need to be greater than 80kS/s and maximum filter in 5kHz, so any noise could screw up it.

Thanks

[johnsold]

I do not have access to that device or any other which might be similar to test but it is likely that the +5 V supply will have a lot of spike-type noise from the switching of logic ICs inside the device. My experience has been that this type of noise has so much high-frequency content that it is almost impossible to remove by differential techniques. 

Decoupling network might help but a clean supply is probably the best choice.

Lynn

[luissimm]

yeah.. I know that. Thinking on that, I just made some acquisitions at 100kS/s and analysed that. My noise dV is something about 7mV (pure direct signal) and 1,20mV filtered at 5 kHz IIR Butterworth 3rd order. 

In order to simulate a full bridge load cell, I tried to use a potentiometer as sensor (voltage divider), and than my dV is about 2,3mV, I'm actually hopeful it will works.

PS.: I supplied a load cell using function generator, ignoring the "quality" from that 5V, and it was ok. Later I decided to mesure this function generator noise I become surprised how bad this signal was. NI device provides much better signal, and because of this I believe it gonna work. Looks like is harder and more expensive finding a very good power supply than doing this. I accept sugestions for power supplies.

Thanks

[johnsold]

Look for a linear power supply, not a switching power supply. The noise comes from the switching process.  Since you only need to power the load cell, the power requirement will be small.  If you choose to use an external supply consider a split supply (for example +/-5 V). This makes the common mode voltage zero so you can often use a more sensitive AI range on the DAQ device.  A single supply can be used if its output is not connected to a common ground. A simple voltage divider with the center point connected to AI GND may provide the same benefit.

The exact details depend on the maximum power rating of the load cell and other factors.

Lynn

[luissimm]

I just made a simple test using a small linear power supply and the results were worse than I expected. I guess its because I'm acquiring at 100kS/s. Maybe I have a very poor power supply, it's possible too. I'll keep searching some way to solve my problem.

I'm thinking on build my own power supply (kind of home made), using LM7805 and some capacitors to make it a bit more stable. Linear power supplies usually has ripple voltage, I hope some capacitors can make it smaller.

[johnsold]

Yes, you will want a regulated linear power supply and probably additional bypass capacitors to keep the power supply output impedance low over the signal frequency range. The regulators produce a low ouput impedance at DC and power line frequencies but not at higher freqeuncies.

Lynn