filterinmg RF from analog signals

[DougFerguson]

I am writing the control software for a system that uses RF (running at 72MHz) as a major component. None of our signals have harwdare filters and for software I am sampling 100 samples at 1k Hz. I average those samples and update every 250ms.

Using language a ME can understand, can someone help me understand how RF can influence our signals and what options I have to filter that noise (harwdare and software). Or at least point me to a good resource.

[Buechsenschuetz]

To be able to give you some advice, you should give some more information. 

a) which is the frequency of the signal you want to measure/acquire

b) which is the role of the RF signal? Is the RF signal modulated with the signal to be measured, and if so, which type of modulation is used?

c) or do you want to measure the signal strength of the RF signal itself?

In any case, handling RF of 72MHz requires some knowledge of RF technique. In this frequency range there are some effects which are hardly noticeable when using low frequency (up to, say, 100kHz).

Also, if you have to block the RF from the DAQ system you will have to use a hardware filter. Digital (or software) filtering cannot replace hardware filtering completely. Think of the effect of aliasing. If you sample a RF signal with a sampling frequency of a fraction of the RF signal, the sampling will occur at certain points of the RF waveform and sampling will yield different results with each sampling. So sampling will yield periodically changing values (with the assumption that both the RF signal and sampling have stable frequencies) but the period (or frequency) and waveform of the sampling output have nothing to do with the input signal. 

[DougFerguson]

a) I am measuring normal analog signals that update on the 1 Hz scale. Nothing fancy.

b) I do nothing with the RF. Its a buy-in component and has its own controller and routine. However its housed in the same cabinet as the rest of our DAQ and control equipment. I am seeing unexpected results on feedback with the RF on and off.

c)No.

I assume the RF is on the feedback and control signals and cause a false offset? One of our other components has a bandstop filter which filters the 72MHz and runs it to ground. This compnent is a power supply which require accuracy to milliamps on a 0-110 amp scale. Should I repeat this on all our signals or is there a better method for a higher channel count?

[Jake_G]

Hey Doug,

What hardware are you using?  What device is this signal you are trying to measure going to?

Do you need the RF equipment near the signal and/or hardware?

[johnsold · Accepted solution]

Doug,

Here is the ME version of RF interference.  The radio transmitter/antenna combination creates electromagnetic waves in the space surrounding the antenna.  Think of dropping a stone into a pond of water.  The wave analogy is reasonable for basic understanding. A leaf floating on the pond will move up and down with the wave.  If you put a float on the water and connect it to solid ground through a gearbox, you can extract energy from the waves.

In your DAQ system two things are required to produce interference. 1. The RF wave must be coupled into the system. 2. The RF energy must reach a point where some non-linear device converts the very high frequency sine waves into some form which the low frequency DAQ system can measure. (2.a. There is one more complicated way interference can occur without a non-linearity - undersampling). 

How can coupling occur? The wavelength of an electromagnetic signal in free space is the speed of light divided by the frequency.  For 72 MHz the wavelength is 4.17 m.  An antenna 1/4 wavelength long can couple to the electromagnetic wave very effectively.  Are any of the wires in your system about 1 meter long?  USB cable? It is quite likely that you have some very nice antennas.  If the voltage coming out of any of these antennas reaches about 0.5 V, it will be enough to cause internal diode junctions in any semiconductor device (like amplifiers, multiplexers, and analog/digital converters) to conduct.  This conduction is non-linear and will produce a DC voltage out for the RF input.  This voltage gets summed (with unknown polarity) with the signals you want to measure, prodcuing errors.

What can you do about it?  Turn off the transmitter will solve the problem, but since it is needed for other purposes, that is not an option.  Since the signals of interest are very low frequency, it should be fairly easy to eliminate the interference.  Three general principles apply.  First is to eliminate the source, which has already been ruled out.  However, check to see if the transmitter power can be reduced without compromising the performance of the RF system. Second is to reduce the coupling.  If possible keep the transmitting antenna as far as possible from the source of your slow signals and the wires connecting to your DAQ system. Orientation of the wires and antenna can help also.  There are too many combinations to try to even give general directions on this without more information about the physical configuration of your system.  Shielding is a big part of the reduction of coupling principle and works well at 72 MHz.  All wires from your signal source(s) to the DAQ and from the DAQ to the computer should be shielded. The shields should be grounded.  Just make sure that you do not create ground loops in the process which can lead to other problems.  Connect everything to one ground using a "star" grounding system.  The third principle is to reduce the interferring signals by filtering. I would try putting a ceramic capacitor of about 100 nF from each signal line to ground at the input to the DAQ, keeping the capacitor leads as short as possible.  If your signal sources have very high impedances or are inductive, you may need to have an EE evaluate the appropriate filtering.  The advantages of capacitors are: Cheap!, simple to install, will reduce RF interference over a broad frequency range, require no tuning, and generally are pretty effective.

Depending on how severe the problems are, it may be quite straightforward to eliminate your interference or it could require a major redesign of the whole system.

Lynn

Addra Consulting, LLC

[DougFerguson]

Wow, great post. However it makes me feel quite dumb. I know nothing about RF sadly.

I think you have narrowed my solutions down very nicely. I will try the capacitors and verify my shielding and grounding.

I am using c-series modules with a 9188 until release then I'm moving to a 9074. There are three signals that are bothering me most. First is a simple mass flow controller using a 0-5 VDC control and feedback. Second is a 0-10 VDC pressure transducer, and third is 0-10uA current feedback which I am converting to 0-10VDC.

[johnsold]

Doug,

RF is a specialized field.  Think of it as taking an EE who was exposed to Newton's laws of motion as a freshman and asking him to design a turbojet engine. Nothing to be ashamed of, nor should you consider yourself dumb.  Quite a few EEs know nothing about RF.

It is possible that your mass flow controller and the pressure transducer are themselves susceptible to RF interference.  They may be producing invalid outputs when the transmitter is on.  Things to try: Put an additional capacitor on both the output lines and the power lines to the transducers as close to the transducers as possible.  If the connections are not accessible at the transducer (fairly likely, unfortunately for this situation), try wrapping the cable to the transducer around a ferrite torroid as close to the transducer as possible.  The ferrite increases the inductance of the wires which produces a higher impedance for the RF signal.  WIth a little luck this amy attenaute the RF getting into the transducers enough to eliminate the interference.

The current feedback may be tougher.  Are you bringing the low current back to the neighborhood of the computer and then converting to voltage there or is the current to voltage converter near the source of the current?  If the latter, try capacitors and ferrites as I suggested for the transducers.  Do you have an active converter (contatining an op amp) or just a 1000000 ohm resistor?

Lynn

[johnsold]

Doug,

Just curious - Have you made any progress in your battle against RFI (radio frequency interference)?

Lynn

[timhurdle]

Lynn, Just to let you know your words of wisdom concerning RF interference were not wasted on this ME, who in the past has had to discover the solutions you described so eloquently by painful trial and error. I have the theory too now thanks to you.  Tim Hurdle (Rolls Royce Marine)

[DougFerguson]

Sorry to not reply. I didn't see an email that you had replied asking my progress.

Turns out we had ground loop issues on the signals. As I mentioned, I inherited a lot of this system and almost every signal was grounded on both ends. After floating the signal end I started to see much more consistent results.

I am bringing the low current to the cDAQ and converting via a resistor very near the module. That cleared up once I floated at the source end ground and shorted the open channels so I wasn't reading open/floating signals. Interesting enough I get better results when I read the current as RSE versus Diff.