Noise in system – A story about real life science

The events of this post are all true (as far as I can remember) and began on the 5th of June. 

Everything was going well. Too well. Nothing goes well in science, especially not as well as things were going prior to June 5th. I just finished piloting a bunch of new behavioral paradigms and I was finally getting some usable data. My master’s student was starting up his experiments over the semester break he had. My rats were all (almost all, there’s always one) behaving well and performing the tasks I spent months training them to do. And most importantly, their brain activity was coming in loud and clear. Until it wasn’t. 

On June 5th there was a series of power surges in our building and one floor lost power completely for a while. The building our lab is in is quite old and spliced onto a new, modern building (2010). Power surges happen from time to time due to the weird power situation between the two buildings so I didn’t think too much of it. Until all that wellness turned into chaos. 

 

 

And I don’t think I’m being too dramatic about using the word chaos. For these experiments I am recording the tiny fluctuations of electric activity (on the order of microvolts) in the brain as rats perform various tasks to understand what their brains are doing during these tasks. This discipline is called “electrophysiology”. The brain’s electrical activity is generally quite nice and ordered. You get beautiful waves of activity as neurons work together to figure out what to do and how to do it. These waves tend to be nice and regular and, in the grand scheme of things, are very very small. A single AA battery has ~1,000 more power than the largest brain waves. Between my student and I, we collected excellent quality brain waves from about 10 rats so far that day. I was on the second to last rat when the chaos began and I started to see huge spikes in the brain signals which couldn’t be real.

This isn’t my first rodeo though, half of electrophysiology is getting everything to work properly so you can do your experiments. Due to our equipment being sensitive enough to record brain activity, it can pick up all sorts of other activity. Weird artifacts in a real signal are generally referred to as “noise” and electrophysiologists are in a constant battle to get noise levels to a minimum. The bane of every electrophysiologist is the mains alternating current (AC) power. The power going to your house and through your walls is alternating at a fixed frequency (60 Hertz in the US, 50 Hz in NZ). However, these massive spikes I was all of a sudden getting were not mains power noise. They were too large in amplitude (how tall they are) and at irregular intervals (not a fixed 50 Hz). So I go through all the usual troubleshooting steps. 1) Make sure the cable is secure to the rat, check. 2) Restart software, no change. 3) Restart hardware, spikes still there. Okay, maybe this rat’s implant is getting bad. Overtime their brain implants can degrade, the immune system attacks the electrodes, components break. So I try a different rat yet the huge, chaotic spikes continue. 

 

 

Okay, fine. Keep going down the checklist. New cable that interfaces rat’s brain to the computer, mystery noise still there. Try a third cable? Noise still there. We use a commutator to avoid the cable getting twisted while rats run around, maybe this is dysfunctioning? Circumvent the commutator, no change. Okay, time to break out the signal generator. The signal generator can put out a controllable input signal at a specific frequency. It’s sort of like mimicking a rat’s brain activity. Aaaaaand drum roll please! A perfect sine wave signal with no huge spikes in it. Now the enigma deepens. With a signal generator putting a sine wave into one channel at a time, the signals come in fine with no noise. However, with a real, living rat plugged in, chaos. I further try some different rats, all the same, all have the mysterious noise. 

 

 

Rather than a signal generator, a second way to test an electrophysiology system is to use a “dummy headplug”. This is just a series of resistors, one for each electrode channel which all connect back to ground to complete a circuit. These resistors mimic having a rat plugged in. Furthermore, you can touch one of these resistors with your finger to create a load, pulling power and creating a fake signal. When I only place the resistors onto the cable which connects a rat to the computer, very similar noise could be seen. This didn’t make any sense to me. Tired and defeated, I walked home through the cold night hoping my problems would be gone in the morning.

 

 

Of course they were not. Thankfully, the University of Otago’s Psychology department has a team of amazing technicians, many of them trained in electrical engineering with decades of experience. After repeating my tests from the day before, I brought in the big guns. All day of the 6th we worked, we tested, we swapped equipment, we scratched our heads. Nothing was working. One of the technicians had the insight  that the noise was due to current flowing from one electrode to another, causing the spikes, rather than flowing to ground. We tested the ground connection and it had a relatively high resistance of 53 Ohms. It should be close to 0 Ohms. They rewired my ground connections and managed to get them down to 0.8 Ohms. And we were in business! No more huge weird spikes! Problem solved, or so I thought… 

I started to run my rats even though it was already a bit late in the day. The first rat’s brain activity was coming in great just as we saw in our tests. Then, they weren’t. After about 10 minutes into the session the spikes were back. Flabbergasted, I let the rat complete his task, placed him back home, and went to my own home. The next day (7th) I had a university conference all day. I showed everyone I could find the above video and no one had any idea what could cause such weird looking signals. After the conference my curiosity couldn’t keep me away. I went back around 5pm and checked things out. Everything looked pretty good, I only saw 2 spikes occur but otherwise the signals were clean. Weird. I recorded from a rat for a full half hour and everything went swimmingly. 

When the rat finished his task I stopped recording his brain’s activity but I could still see what was going on in there on my computer screen. We run our rats in the dark so we can track little LEDs on the cable and know where the rats are. I turned the lights back on to the room and there they were. Those huge, chaotic spikes. THE LIGHTS! I turned them on and off about a dozen times and sure enough, when the fluorescent lights turned on the same spikey signals which have been plaguing me were there. But only briefly while the lights flickered on, then the spikes stopped. I ran around turning on all the fluorescent lights I could find. Our equipment was picking up lights in the office next door and in the hall way. However, all those lights were on and working fine. Though the hallway did have two blown lights. There had to be a flickering light to cause the spikes I had been seeing. So I ran through the building checking for flicking lights. None on my floor. Above me is just a roof. I checked the floor below me, there weren’t even any lights on; everyone had left hours ago. I turned them all on, no flickering. Well I had a hypothesis but no smoking gun. On the other hand, the noise was gone and there was no flickering lights. There were now two fully dead bulbs in the hallway, maybe one had been flickering the two days before. Either way, the noise issue was over for a second time.  

 

 

Then it was back for a third time. My master’s student ran his rats without any issues. I was doing some things and didn’t get to my rats until later, around 4pm. First rat’s brain was looking great, signals were clear smooth waves. Then, the spikes. UGH! I went with my last resort test. I went and got my old recording system from my PhD and set it up in my room. I wanted to see if I’d get the same noise on that. I got a rat, plugged him into the plagued system and, of course, the spiky noise was gone and I couldn’t run my test. I returned all the equipment, tested two more rats (no spikes), and went home.  

 

 

The next day (9th) was even more eventful. The cable which interfaces our rat’s brain implant to the computer broke and stopped transmitting any signals at all. But that’s what backups are for. New cable swapped in everything looked good and we collected a bunch of great data. But then it started to get late. These spiking problems always seem to be worse at night. I plug in my last rat around 5pm and, yes you guessed it, the spikes were back. I dropped everything and re-setup my old system in the cursed room. This time the spikes persisted. I checked the signal on my old system and they were fine; clean and absent of chaotic spikes. Back to the new system, spikes still present. Something about the new equipment is picking up or causing this noise. The noise persisted till 9:30 pm when I tapped out. 

Our building has two ways out, an internal stair well and an outdoor stair well. I’d been avoiding the outdoor route since it’s winter now and delaying going in the cold was ideal. But I was tired and at this point didn’t care. I walked into the frigged air and was greeted by a massive, flickering fluorescent light. You’ve got to be kidding me. Is it possible this could be picked up from so far away and through two walls? It was a huge security light high up and unless I found a sledgehammer wouldn’t be able to turn it off. The lights are automatically controlled by brightness levels and come on whenever it gets too dark. Tomorrow the tests would continue. 

The following day I sped along and ran my rats as quick as I could, trying to get them all finished before sunset (~5 pm). I finished the second to last at 4:31 pm, no spikes. I had to rearrange some things and make more sucrose reward. I plugged in the last rat at 4:56 pm. Spikes. I run outside. The distinctive pops and pings and flickering light hits my senses. I’ve got you now. 

Today, on the final day of this story, I did manage to run all my rats before 4:30 pm. I then started a recording with the dummy plug and sat and stared at that cursed light. Every 3 minutes I checked the signals. Good, good, good, good. Light comes on at 4:44 pm. Spikes. I sprint down the stairs and ask the kind technicians, the same ones trying the week prior to help, to get the bulb out as they were packing up for the day. After tracking down a ladder and a wrench, the dying light was taken out for good. Spikes gone, hopefully once and for all. 

 

---

The leading theory is that the hundreds to thousands of volts required to start up a fluorescent bulb and which pulse through it when it’s flicking out were being picked up by the new equipment. The power surges may have led some of these lights to an early grave which he hadn’t seen/thought of earlier in this fiasco (5th and 6th hallway lights dying, 7th all good, 8th outside light starts to die). The signal generator has massive resistors between channels, stopping the noise from leaking across channels. The high ground resistance may have been its own problem and also exacerbated the lights. The headstage which ended up breaking may have also contributed to the earlier day spikes. 

We are asking the university to install LEDs to replace the fluoros. 

---

 

This work by Blake Porter is licensed under a Creative Commons Attribution-Non Commercial-ShareAlike 4.0 International License