I have a noise/interference problem with an RC oscillator circuit connected to Arduino that only occurs when the Ardunino is plugged into the USB port on my computer.
The net result is a huge amount of jitter modulating the base frequency. (The capacitor is made from two insulated wires acting as a liquid level sensor.)
I don't have a ground loop in the normal sense as everything is running off batteries and I'm only using the USB for programming and monitoring.
Research indicates that this is a common problem affecting people working with audio signals and is caused by the USB power ground being connected to the chassis earth in the computer.
Apparently USB uses the power line/ground during initialisation, so disconnecting them is not an option.
The good news is that USB isolators do exist (http://www.bb-elec.com/product_family.asp?FamilyId=651), the bad news is that they cost ~$100.
Does anyone have any cheaper suggestions for filtering?
Try just disconnected the USB's 5V line and see if resolves the problem. As long as your board has a good 5V reference, this shouldn't be a problem and will minimize the ground-loop.
I have USB isolators for much less, check this -> http://www.circuitsathome.com/products-page/usb-interfaces/ All docs are posted as well, if you are good with building boards, you can spend even less by making your own PCB and ask ADI for a sample - they are generous folks. ADuM3160 will work, too with no modifications.
It is difficult to say without a schematic of what you have but it sounds like you have some very high impedances and this is causing pickup when connected to another ground. It is not the ground itself but the susceptibility of you circuit. Try to reduce any high impedances.
Grumpy_Mike:
It is difficult to say without a schematic of what you have but it sounds like you have some very high impedances and this is causing pickup when connected to another ground. It is not the ground itself but the susceptibility of you circuit. Try to reduce any high impedances.
The capacitive level sensor currently uses a 4060 osc/divider with a 390k charge/discharge resistor giving an oscillator frequency of about 5KHz.
This seemed like a good compromise between impedance, capacitor physical size and frequency as my oscilloscope only has a 1MHz sample rate which makes it hard to troubleshoot things running at RF frequencies.
(The divided output of the 4060 goes to a max485, through 45m of twisted pair to another max485 then the Arduino.)
According to An Introduction to Flow Meters capacitive sensors typically work at MHz frequencies so I guess there is nothing stopping me from running a much lower charge/discharge resistor to give something in the region of 500kHz. Assuming a linear relationship between interference and impedance then this should give a hundred fold decrease in noise which may be good enough for my needs.