A to D question (differential mode vs single ended mode)
Bits
Posts: 414
My situation is as follows
A Sensor is embedded into a device about 3 meters away from a PCB that is connected with bundled wire (8 conductors, non shielded) where two of the conductors hold ~20 amps at 5Volts, so noise is an issue. Its working at the moment but now my boss wants to double the power.
So here goes.
Currently I am using an 20 bit ADC in single ended mode where the sensor is coupled to a simple voltage divider followed by a passive low pass filter.
I was wondering if I could make a better measurement using differential mode with a sensor in a Wheatstone bridge configuration followed with again a passive low pass filter?
Any comments are appreciated.
A Sensor is embedded into a device about 3 meters away from a PCB that is connected with bundled wire (8 conductors, non shielded) where two of the conductors hold ~20 amps at 5Volts, so noise is an issue. Its working at the moment but now my boss wants to double the power.
So here goes.
Currently I am using an 20 bit ADC in single ended mode where the sensor is coupled to a simple voltage divider followed by a passive low pass filter.
I was wondering if I could make a better measurement using differential mode with a sensor in a Wheatstone bridge configuration followed with again a passive low pass filter?
Any comments are appreciated.
Comments
Can you place the ADC on site?
A great read for me is keithley's free low level measurement handbook (there is also a downloadable pdf, and an android app, don't know apple side).
Lot of infos..
Massimo
Assuming the use of the 20-bit a/d was because the signal was worth measuring to that level, one would suppose that the all the details of high-precision measurement would be taken into account, too.
At the very least, providing a shielded cable (grounded only at the a/d end) for the signals over the 3 meters would not be too expensive to add. As well, the low pass elements could be optimized for any particular noise sources in the system.
Another good reference with practical examples is the Omega Engineering technical reference section.
Cheers,
Differential mode would be the way I would go to elimmate or reduce common mode noise. Also, would it be possible to either move the pcb closer to the sensor or maybe provide some conditioning at the sensor to reduce the effects of the noise.
Microchip has some good appnotes that may be of interest. Also, Bonnie Baker (an MCP engineer) has a great book out "A Bakers Dozen" on amazon and other sites. I don't recommend the kindle version though, to many good charts and graphs are hard to use. At least until someone comes out with a conversion program from the latest kindle secret squirrel format to PDF; then I will buy it for keeps. But this book is gold.
FF
Differential measurements can help AC rejection, as pickup tends to be common mode, and differential has equal impedance on each arm.
For strict DC measurements, it depends on the ground wiring.
If you have a separate return wire, with no current flow, (ie just for your sensor), then you are already decoupled from IR cross talk effects.
At 20A and 20 bits, I am guessing you already do that.
A reality check in such systems is to run a separate test power lead, physically spread, and see if your measurements improve.
I suppose I was more interested in the behavior of the different modes and employing a Wheatstone bridge or voltage divider. I am going to test these next week so I can be sure.
What sort of sensor is this and what is it measuring?