Shielding vs. filtering / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Shielding vs. filtering

Move Thread

LAN_403

Turkey9

Sun Mar 01 2009, 09:04PM

Registered Member #1451 Joined: Wed Apr 23 2008, 03:48AM
Location: Boulder, Co
Posts: 661

I need a way to eliminate interference on my scope trace. I have a signal from a piezo electric force sensor being fed via coax to my DSO about two feet away. There is a high voltage arc close to the sensor and the proximity can't be avoided. I get a good amount of interference from the high voltage and i'm wondering if there is a better way that i can shield the sensor feed. Also, the signal from the sensor is on the range of .14mV to about 50 mV. Since the high voltage is a fixed frequency, can i build a notch filter? Would it be easier to shield better? Do you think that the interference is traveling along the coax or is it penetrating the scope, in which case i could build a faraday cage? Sorry for all the questions. Any help is very appreciated!!!

Steve Conner

Sun Mar 01 2009, 10:12PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

You should use both shielding and filtering. A Faraday cage around everything, wrap the wire from the sensor through a ferrite ring, isolate the sensor in its cage from the test rig to stop ground loops, etc.

You could add a notch filter (would be called a "trap" in this case) but I expect the presence of an arc will generate a lot of broadband noise besides the original HV power frequency.

Turkey9

Mon Mar 02 2009, 04:25AM

Registered Member #1451 Joined: Wed Apr 23 2008, 03:48AM
Location: Boulder, Co
Posts: 661

So the sensor ground should be isolated from the rest of the circuitry? Here's another twist that could complicate filtering, the sensor outputs a very short pulse, around 1 ms that has to be differentiated from the noise of the high voltage. An idea i had: sync a transistor to the 555 oscillator for the high voltage transformer (flyback) so that it grounds the sensor output at the time that the high voltage pulses occur. Then i would just have to hope that sensor pulse occurs between the times when the output is grounded.

I'm not sure what the ferrite ring would do, but i bet it would smooth out pulses... am i right? would this damage the sensor output?

All i need is for my scope to be able to pick up the sensor pulse that is usually drowned out by the interference.

Proud Mary

Mon Mar 02 2009, 07:47AM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

Turkey9 wrote ...

I'm not sure what the ferrite ring would do, but i bet it would smooth out pulses... am i right? would this damage the sensor output?

It acts a 'lossy' choke in this case, and attenuates common mode current flow in the cable.

GeordieBoy

Mon Mar 02 2009, 11:58AM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

This sounds like a very difficult problem to solve.

The first thing I would suggest would be some amplification to the signal from the transducer, right at the transducer itself, or as close as you can get to it. Then send the amplified, and stiffened (low impedance output) down your coax cable to the scope or data aquisition card etc.

The ferrite core that Mr Conner mentioned is a very good idea. It acts as a common mode choke and prevents those currents running down the screen of the coax that you hinted at. Since the entire coaxial cable is wound around the ferrite core it does not alter the wanted signal in any way, but it is a well-established and effective way of preventing common-mode noise being conducted from the noisey end of a cable to the other end that you want to keep free of interference.

As for differentiating the small sensor output pulse from the spark gap hash, I think you might have problems. Do you need to process the results immediately in realtime? or can you run the experiment, gather the data and then analyse it later in something like Excel or Matlab?

If you can analyse the data later then you may be able to use digital filtering to remove the unwanted signal and/or boost the wanted signal to make it easier to see. Also if the wanted signal from the piezo has a particular known signiature shape, you can use things like correlation to pick it out from the forrest of noise. Likewise if it has a known frequency content then the FFT might prove a useful tool.

But, first stiffen up the signal, and shield and screen everything, pay attention to ground returns, and fit common mode chokes. Use the fancy data analysis stuff as a last resort if you've tried everything else and you're still stuck.

-Richie,

Turkey9

Tue Mar 03 2009, 02:13AM

Registered Member #1451 Joined: Wed Apr 23 2008, 03:48AM
Location: Boulder, Co
Posts: 661

I had already thought of the amplifier, but discarded it for reasons that don't come back to me. Would a simple high bandwidth op amp be good enough to boost the the signal and stiffen the output?

Also, the whole setup with the sensor and arc is inside of an old and modified propane tank. The tank isn't used as anything dangerous, just a containment vessel. It is grounded, and i'm wandering if this would act like a faraday cage.

Would I put the chokes at the scope end and the sensor end of the coax? Does it matter what kind of ferrite I use? I have a bunch of toroids that I got off of the flybacks in computer monitors, a couple used as chokes, and a couple from the output side of ATX PSUs (the ones used for SSTCs).

I don't need to see the data right away, but my scope is an old crt style one so how would I output the data to my computer? The pulse i'm expecting will be a very short and sharp spike, but unfortunately this is very similar to the interference. I might try to trigger the pulse at a point where it will occur between the interference spikes if the shielding and filtering doesn't work.

Dr. Slack

Tue Mar 03 2009, 02:01PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

I'd take a difficult signal transfer problem like this a little bit at a time. Don't go from one step to the next until you have sorted whatever is going wrong.

So, with the interference running, and your scope input shorted _at the scope_, is the trace clean?
Turn the scope sensitivity up until it could read the sensor if connected, still OK?
Now connect a coax to the scope input, short it centre to coax at the far end. Still clean?
Connect the sensor between centre and coax only, don't "ground" the scope coax at the measurement end.

I can't suggest any more without seeing results of each experiemnt, but you get the general idea.

An amplifier on the sensor is only useful if the scope does not have the sensitivity to display the sensor, OR the coax you have picks up interference, so benefits from a bigger wanted signal in the coax OR the output impedance of the sensor is too high to drive the coax and scope adequiately. There may be common mode being induced onto the coax, this is where a ferrite over the coax will come in handy. The screen may be sparse, get a beterr quality cable. The ground connection of the cable to scope may be poor through dirt, corrosion or a poorly made connector.

If you are picking up the interference direclty on the sensor leads, then it doesn't matter whether you connect sensor to coax or to amplifier, it's much the same problem. You have to minimise the inductive area of the lead connection and electrically screen the conductors. Both happen automatically in coax, so think coax when making the connection.

good luck, and suspect everything.

GeordieBoy

Tue Mar 03 2009, 02:02PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

For common-mode chokes you want high-permeability Mn/Zn ferrite. You can use the common-mode choke toroids from PC PSUs but NOT the yellow/white iron-powder toroids from the Low Voltage output side.

You can use two common-mode chokes, one at the sensor end (ideally right where the sensor cable goes through the wall of the faraday shield,) ...and another one right at the input to the scope.

You could use a data acquisition card to sample data to a PC and process off line.

BTW, What is the application? I'm sure others are wondering too

-Richie,

Turkey9

Wed Mar 04 2009, 03:30AM

Registered Member #1451 Joined: Wed Apr 23 2008, 03:48AM
Location: Boulder, Co
Posts: 661

I'll do the test's, shorting the input and coax, and let you know what I get.

The project is actually for the science fair. I've built and improved thruster (as in the type used on satalites, ie ion drives) that is based on a Pulsed Inductive Thruster that NASA worked on a while ago. My design involves creating a plasma between two concentric rings on the face of the thruster. This is where the interference is coming from. The propane tank is used as a vacuum chamber, getting to around an estimated 100 microns. The force sensor is at the end of a lever/stand that the thruster is mounted on. It is around two inches away from the plasma. The high voltage is DC, and adding a smoothing cap helps, but I don't have enough to do much. I don't want to give too much info on the thruster because I'm not 100 percent sure that everything is working right at the time and I don't want to give results based on false data.

Update: I was able to find a temporary fix to my problem. Luckily for me the time when the pulse from the sensor will occur there is also a large current sent to a coil. I would a sense coil around one of the high current leads and use the voltage spike on that to trigger the scope. I was able to find the pulse from the sensor this way after a few tries to get the sensor and interference pulses to happen at different times. BTW, I won over-all at the science fair today, and am going, all expense paid, to Reno Nevada for the international fair.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob