The situation, briefly, is that I'm doing science in a Faraday cage, but have a lot of electrical noise that just randomly started at some point. It happened before, and went away within a week or two, but recently it started again and after a few days I got fed up and started removing things from the cage. It turns out that the bifurcating gooseneck light that I use to illuminate the preparation is causing it. I don't know why that suddenly started, or what. I just got frustrated with removing it to make a recording and grounded both necks, and it's... much better, maybe even usable.
However, it's still somehow generating noise. I know this because when I touch it, the noise decreases even more, but springs back about a second after I let go. I don't think it's just me being grounded (it might be? Not an expert) because when I dip my finger into the grounded solution the prep is in, there is still noise.
Does anyone have experience with this, or ideas on how to fix it?
Posts
And inspect any cables with shields. If the cables get kinked in any way, the shield or drain wire can break, which can cause issues with noise obviously.
The problem I'm having is not mysterious this time. The two pumps, the intake pump to a small degree and the outtake pump to a much greater degree (probably to do with the type of pump, I'd assume?) add a small-but-damaging amount of noise. The tubing I'm using is just some aquarium tubing - polysomething or other. I can look it up if that matters - it probably doesn't. The ends of the tubing are connected to a much smaller piece of tubing, which I'm not sure the identity of (though I think it's polyurethane), which is what actually goes to the sample dish. The intake line physically enters the fluid, the outtake line does not generally, as it would completely drain the dish if it did. However, this means it's not touching the fluid with the ground, which leads to even stronger noise.
My question is this: is there a way to ground this properly? Since the materials are non-metallic, I can't just.. ground them. Does anyone have any ideas on ways around this, or some simple trick I don't know because I'm not a physicist?
I can probably fabricate small things if I need to, though since the dish is glass I can't make many alterations to it directly.
Anyway, this is my new setup and it has problems, and I'd appreciate any help.
My initial guess was going to be the liquid, but you said the outtake tube is empty most of the time...though if the pump is always running it may be doing something weird.
Since it sounds like your experiment is fairly sensitive, I would ground everything that crosses the cage boundary even if it isn't something that would be called "conductive". I'll say that "conductor" and "insulator" are more like engineering terms and not science terms, so everything is relative
Edit: Oh and also make sure the pump and data collection wire (?) are electrically separated outside of the cage as well. Pumps are hellish EM generators and you may be picking that up instead of something penetrating the cage.
Hmm the bolded is quite possible, they all come through the same area. Things are shielded, but still. That makes sense.
The cage is open, it's not like, completely sealed sensitive... but I'm doing neuroscience, so I'm trying to record things that are fairly small (millivoltsish) so even relatively small noise is an issue.
(worse, what I'm recording looks roughly like a single pulse of a sine wave, so it's really easy to lose it in the noise).
With debugging what the noise looks it might help if you could show us an example. Can you do a FFT (fast-fourier transform) of the noise to figure what dominant frequencies it occurs at?
1) The pumps might produce a whole bunch of noise while operating. Can you physically move the pumps as far from both your sample/cage and your data collection cables/instruments? Make sure they're plugged in on a different power circuit if you can. Does that help?
2) Can you feel any vibrations through the cables while the pumps are on? Physical vibrations can turn into electrical noise.
3) Do you still get noise while you pumps are adding/removing water from a vessel right next to your grounded sample (i.e. is the noise coupling directly through the liquid or just close proximity)?
4)How are you grounding your liquid? What else is plugged into the same power circuit as this ground?
I don't think this is a problem that you can't solve with proper grounding and shielding but at some point it's not worth spending a bunch of time solving problems that you can just buy a solution for.
I'm not at work yet, so I can't answer all of these, but the noise is roughly sinusoidal and I'm 90% sure it's 60 Hz (because it's always 60 Hz, I'm not un-used to noise of that type). I believe the pumps are on a different circuit from the data collection stuff, but I'll have to check. Grounding is to the same circuit as all the scientific tools, different circuit from the pumps and light, and is just a gold pin soldered to a wire that's placed in the sample dish.
I'll check the rest later today.
Edit: the problem seems to have been, mostly, that the tubing was coming into the cage via the same hole as the data recording equipment. I've taped all the tubing so that now it runs along the legs, towards the center of the open face where I interact with everything, then up over that, so it's not near any data collection cables (which run along the back and out the side). Trying to shield seems to make things worse. There's still some noise, but it's probably low enough for me to record the EPSPs I need. For everyone's information:
The outtake pump tubing is definitely vibrating a fair bit - it's a very, very cheap pump. So that might be contributing to residual noise (I dunno how to fix it, it runs along the entire cage and still vibrates). The intake pump tubing isn't, it's a super expensive peristaltic pump. The light and pumps are plugged into a power strip plugged into the wall, while a different outlet is powering a huge power strip that the data instruments and ground all come from. The power spectrum shows spikes at 60 and 120 Hz, though I'm not an expert at interpreting them (it's not trivial to get data from that computer to my laptop, and the computer's not networked). Since the noise at any given point is not quite as big as what I'm measuring, and I can see what I'm measuring even with some noise (since the noise moves up and down with the phenomenon, which is a bit slower than the noise), I've probably got this. Thanks again for all the help, I'm sure I'll need it again in the future (...hopefully not later today, though that's possible if I run the experiment and suddenly nothing works. Trials with petri dishes full of saline seem ok though).
At any rate you can by conductive plastic tubing, which can be grounded. Not sure if that helps with anything. Also when you have so many grounds I would 100% check for currents/ground loops, or maybe through some capacitors on there.
I did not know conductive tubing was a thing, I'll look into it if I have more problems!
60Hz is just mains interference (assuming you're in a country with 60Hz AC power) and it'll also show up in harmonics of that frequency (120, 180 240 etc.) It's probably the cheap pump. Actually, it's probably everything, just with the cheap pump being the biggest contributor at the moment. What happens if you run with it turned off briefly?
Conductive tubing might help but if shielding it with foil or copper mesh doesn't help I'm not sure I'd expect that to. If it's 60Hz it's probably not vibrations from your peristaltic pump (which will probably appear at whatever RPM it's spinning at) though if you remove this noise you might start to see it.
Here is the current situation: I have improved my rig by adding three solenoid valves with various solutions and an automatic pump control. Everything is powered by a relay, which is controlled by an Arduino. Suddenly, I am getting nasty noise. I am fairly certain it is coming from the pump, as it only occurs when the pump is on. However, it also seems like it does not occur when the tubing connecting the solenoids to the pump is disconnected (the tubing is filled with a conductive solution, so it occurs to me that I may be inadvertently creating ground loops). Whether or not the solenoids are on doesn't appear to have any bearing on the noise. My basic setup is the same as it was upthread:
except now attached to the (other side of the) pump, there are a few solenoids, and they, along with the pump control, are wired into a relay being powered with a 12VDC adapter. And also there's an arduino controlling everything. The noise I get is kind of weird, top trace in the picture below:
It occurs in small bursts, it is in an increasing and then decreasing sinusoidal pattern, and it is 60 Hz. It's a bit harder to see in the picture above, but it generally comes into existence, hits a peak, then disappears, and is symmetrical across the tallest peak.
Does anyone have any suggestions, and also, does anyone recognize that noise pattern? It's really weird that it just starts and then stops.
I am assuming your pump controls are being run independently of your data collection computer?
I would try replicating your noise by just dry running your pumps with no experiment, but recording data. Leave the pumps running but unhooked and try physically separating them from your data line.
If it is radio noise being picked up in your data line you can try a few different things, depending on the frequency of the signal you want to measure. You can use a simple RC low pass filter, or using a ferrite clamp on your data line.
Edit: though, now that I think about it for a second, a ferrite clamp will probably not work for your frequency of noise.
You'll have to bear with me - I know very little of electrical engineering, so I might misinterpret things.
The data lines aren't really shielded (I think - I actually should probably look up the specs), other than starting in the faraday cage. I do know that normally noise isn't really an issue for them. The recording setup is completely independent of everything else. I have a digitizer and various amplifiers and such hooked up to an ancient computer, and then separately, an outflow pump plugged into a power strip, and then separately again, the inflow pump, relay, arduino, personal laptop powering the arduino, and so on. The pump turns on, then runs until I turn it off - this is normally a period of several hours, at least. It's a Thermo Scientific FH100 peristaltic pump, if you want specifics.
I am currently running a mock experiment like you suggest. I am not sure exactly what you mean by leaving the pumps running but unhooked, could you clarify?
It seems the noise happens when
1) the pump is on, and
2) the pump's inflow tubing is connected to the + junction with the solenoids (regardless of whether or not the solenoids are energized) and
3) there is an amount of fluid in the tubing between the pump and the dish, though it appears that it does not need to be totally continuous
It'd be pretty difficult, especially because the main reason for the pumps is to maintain a constant flow rate across the preparation. I can try to think of alternate methods, though.
From what I can tell there could be three ways you are getting noise in your measurements:
1) Vibrations caused by the pump being converted to an electrical system by whatever instrument you are using (this depends on the nature of your measurements. Pressure measurements or something of that sort could convert vibrations in this manner)
2) Electrical noise caused by conduction through your system
3) Radio waves emitted by external equipment (pumps etc.) that is being picked up by your signal wire (essentially acting as an antenna)
If you run your pumps without having them physically connected to the rest of your apparatus, you should be able to eliminate 1) and 2). Your noise from radio waves should fall off at 1/r^2 rate so moving them further away from your apparatus should demonstrate whether or not 3) is what is causing your noise.
This is assuming your data line is carrying an analog voltage and unshielded, in which case it is basically acting like an antenna.
Edit: the more I think on it the I think it is probably unlikely to be radio waves that low of frequency. What rpm are you running the pump at?
The solenoids switching will have a transient response which could look like your signal (a damped oscillation, essentially). Have you tried just switching your solenoid valves in isolation from the rest of your system and seeing if there is a response?
At your sensitivity levels, I would definitely make sure the data wire is shielded (with the shield grounded properly to either the cage or earth ground in your electrical system) after it leaves the faraday cage.
1) when the tubing is no longer in contact with the (grounded) saline in the dish, noise gets very high, and is very spiky (not like the relatively calm waves I posted earlier). However, it looks like it has a similar thing with respect to brief bursts of bigger, then smaller, activity, with what appears to be the same envelope as the noise I'm having issues with.
2) as the tubing is pulled further away, the noise becomes smaller.
3) the tubing generates noise when not in contact with the saline even if the pump is off, though it's much smaller in size than if the pump were on.
Here is the data acquisition bit. Mostly this is irrelevant, I think. There is a computer to the right, and the cables on the left are coming out of the faraday cage. The one that is passing the signal that is most affected goes up and into the second device from the top.
Here is the setup inside of the cage. The blue wire going off to the right is the one with the funky signal. It connects to the device in the silver holder on the bottom. The dish is right there. The yellow wire is the ground. The pump is connected to the tubing in the lower right.
This is the pump, there isn't much to see, really. There is a problem when it is on (sometimes. It's super sporadic). The yellow wires go into the relay; when they're shorted, the pump turns itself on.
This is the under part of the cart, where the solenoids and drugs live.
Edit: Additional info: If I disconnect the 12VDC connection, the one that powers the relay, the noise stops. If I short the pump wires manually, causing the pump to go, the noise appears and disappears completely when the adapter touches and is separated from the +/- splitter I have on the end.
Ketbra mentioned it earlier but there's 3 obvious sources of noise.
1) Anything that runs of wall power is gonna put out 60Hz rf. Any long unshielded wires will act as an antenna for this and pick up noise.
2) Any mechanical vibration might shake your whole sample slightly. This will also introduce noise. This includes making and breaking new water connections. Each "drop/section" of water is not necessarily at the same potential so adding more will almost always show up in your electrical signal briefly.
3) Anything that's got a direct electrical connection will also introduce noise into your sample. Sometimes sharing a ground will do that too.
Your data line looks unshielded which is likely picking up your noise source. Actually everything is, but your noise source is the only one you look at with enough detail to notice. It might be possible to eliminate this source but then you're just going to be seeing the next smallest noise source. The fix is to shield your data line better.
I see two ways to do this. The easy (and shitty) way or the good (but makes measurements more annoying) way.
Easy way
All your instruments have BNC coaxial connectors. These are shielded connectors. At some point this is turned into a unshielded banana plug cable and then runs some distance to your sample. This gives you a large distance to pick up noise.
You need something like this (https://www.showmecables.com/bnc-female-to-dual-banana-plug-adapter). Plug the top part (signal) into you dish, the bottom part you connect to ground and you use a long BNC cable to go from your measurement instrument direct to this adaptor. Now the only unshielded part of your line is directly in the dish, a much smaller amount of wire to pick up interference from.
Hard way
All of your measurements take place in a die-cast aluminium box.
Here's one I prepared earlier. (https://photos.app.goo.gl/mxoAAYkayoUoEwjb2) Shielded cables connected with the side of this box and are then split off into unshielded lines once inside. There's a big hole in one end with a copper mesh sleeve. You stick you microscope objective inside this hole and then wrap the copper mesh around it. I use a TIRF objective that needs to be really close to my sample, but if that's not the case for you you could probably use a transparent, conductive window instead. Your pump and syringe lines would probably need dedicated feedthroughs. This is probably going to be the biggest noise source in this case but it would still be shielded far better than what you have now.
P.S None of this addresses source 2) of noise but it's hard to do much about that other than moving the source further away. Your pump isn't on the same table and the source is intermittent which makes me think this source isn't likely at the moment.
Edit: I'm not sure the box is an option, unfortunately, since I need to be able to manipulate the sample (or, well, jab it with 100micrometer precision). The Faraday cage DOES close, though it doesn't seem to make a huge difference.
It's not pretty, it's not the best, but it's quick and cheap.
What a coincidence, I have a huge thing of it lying right by my rig :P
Even if the connection on your data line is something else, the same principle applies. You need to shield your data line for the longest possible distance between the sample and your measurement instruments.
Based upon your observations, my guess is the biggest noise source is the transformers that supply 12V DC to the relay. A better transformer might well reduce a lot of that. If you haven't already you should also try making sure it's plugged in to a different circuit than your instruments. I think the sample/instruments/faraday cage should be on one ground with no loops. Everything else should use a different ground or be floated.
Looking for a good AC adapter is ongoing, but sometimes the solution is to abandon all good practice.
Ground everything.
Everything?
EVERYTHING!