…Written by Frustrationi cum Noisare
This post, part one of a three-post series, is intended to be both handy as a trouble-shooting resource, and also as a primer for patch-clampers setting up new rigs. Dig deeper for a more complete overview in and Part 3 of the “Hunting for Electrical Noise” series…
Please talk to NeuroGig Ltd. about big savings and custom designs if this is you (via Open-Hardware Support and Community-developed Instruments)…
1. First step – Plan your hunt
After a couple of recent noise-hunting expeditions, I thought it might be useful to put a few ideas down which are better organized than my recent handwritten notes and text message exchanges with colleagues. It is easy to find noise troubleshooting advice online, but I don’t find any of these particularly helpful. In my opinion, these are either too technical (all of us don’t do electrical engineering in our spare time), not comprehensive enough (notes from manufacturers or online forum exchanges), not applicable to electrophysiology, or based on patch clamper superstitions (not that some of these superstitious recommendations have not proven helpful!).
So, here’s my attempt – I’ll start out with a bit of background to the concept in general. We’ll then look at general good practice when setting up your rig and how to maintain it. These sections should help with the conclusive remarks, where I offer a few tips for noise troubleshooting. An important part of this is to address the mindset I think is useful in approaching noise hunting. Much of this approach should be intuitive for scientists, but often the frustration accompanied with noise troubleshooting pushes this intuition to the background. I’ll therefore list some questions worth asking and how to approach the hunt more methodically.
1.1. General concepts – Motivation behind taking on the expedition
As far as electrophysiological recordings are concerned, electrical noise is interference which degrades or masks the recorded signal. Since we’re working in a forest of minuscule signals and high electrode impedance, combined with a complex environment (wet stuff, lots of electronics), the electrophysiology experiment has its unique set of challenges to address. This become very time-consuming and frustrating to resolve, even if you know what you’re doing.
Let’s categorize what we’re dealing with…Some noise types are inherent to the measuring instrument and the technique. When using the instruments in the prescribed way, nothing more can be done to improve this (The Axon Guide: Chapter 11 offers a good summary on this). We’ll call this “inherent” noise – still unwanted, but not much you can do about it. External noise is the kind of nonoptimized interference you get, typically from interaction between devices and the environment.
You can get away with less than optimum noise levels, depending on what you’re trying to measure. Your traces will usually have some noise presence (inherent plus some external). With your electrode in the bath, looking at your noise baseline, evaluate the peak-to-peak amplitude and RMS noise level readouts. Decide based on this whether it’s acceptable, or if you’ll have too much signal masked by the noise values (typically, you can multiply the RMS with six to get a snapshot of what 99% of your trace peak-to-peak values will fall within. For example, an RMS value of 3pA means 18pA peak-to-peak values on your traces not an ideal situation!).
1.2. The hunting grounds
To think systematically through the problem, it’s helpful to distinguish between the three actors when troubleshooting noise:
• the source (quite often several of them)
• the receiver [in our case usually the sample, electrode(s), preamplifier/ headstage, cable running to the amplifier, amplifier itself, cable to the Data Acquisition System or analogue/digital converter (referred to the DAQ hereafter), or the DAQ itself]
• and the coupling mechanism (so how does the noise get from the source to the receiver)
Remember that no system can be quieter than its noisiest link. This is evident from the RMS formula:
Taking the above calculation as an example, the dominant impact of the noisiest link is clear. One of the implications of this is that you may run into situations where the biggest noise source hides multiple other sources below its RMS behaviour.
Once the biggest issue is resolved, you may only get drop a few pA or mV in peak-to-peak noise, with the next biggest source now being the main culprit. Knowing this will help you to manage expectations when troubleshooting – you may need to dig deep into your patience reserves!
1.3. The ecosystem – typical coupling mechanisms
There are four ways noisy interference typically gets coupled from the source to the receiver:
• Electromagnetic interference – typically via free-space air; can be over great distances (e.g. radio frequencies)
• Conducted interference – also called “galvanic coupling”; this is when the interference is conducted via cable, chemical compound (e.g. salt residue) or liquid (e.g. pipette solution overflow into pipette holder). A special case here is “common impedance coupling”, where conductive cables are shared between two circuits, and activity on the one circuit affects performance on the other circuit. Common impedance coupling in the electrophysiology context is more typically an issue within instruments instead of between instruments
• Magnetic coupling – also called “inductive coupling”; typically, near-field activity, and often exacerbated by coiled cables in and around the setup
• Electric coupling – also called “capacative coupling” or “electrostatic coupling”. Also has an impact, but only over short distance. Stray potential differences cause electric fields to build up and interfere with nearby conductors
All four of these mechanisms are contributing factors to noise in electrophysiological setups, with electrical coupling probably being the most common.
1.4. Identity of the beast you’re hunting – Noise categories affecting electrophysiology recordings
How does noise typically present itself in an electrophysiology setup?
Once you have your pipette in the bath, your traces (displayed in gap-free mode) will often display a consistent noisy pattern. These patterns offer clues about the source and sometimes also the coupling mechanism. It’s therefore important to try to characterize the noise on your trace. Of course, noise sources tend to mix (remember the RMS implication above), but there will often be a single dominant pattern visible. The most common types are:
• Cycle noise – also called “line frequency noise” or “hum” (e.g. power supplies or light sources); This type of noise is probably the most common of all external noise sources and is picked up from mains power. Therefore, it has the characteristic 50Hz or 60Hz cycle (corresponding to the frequency of your mains AC power), or a multiple thereof.
• Digital noise – also called “white noise”; common devices on rigs like computers, some digital processor amplifiers, microprocessor-based devices, and digital or analogue cameras have high frequency clocks or oscillators – these are often not completely electrically isolated, which may generate stray signals. Because these are high frequency (often in the MHz range), they can be hard to troubleshoot, and pin-pointing the source may be more difficult
• On/Off switch noise – this is when you see spikes on your traces when devices switch on or off (or activate/deactivate). Switching devices on or off happens rarely mid-experiment but using TTLs for activating devices or changing the speed on perfusion pumps or valves is of course a common occurrence. Additionally, air conditioners and fridges/freezers in the lab often cause these on/off switch artifacts
• Other oscillations – Might be high or low frequency, and often caused by mechanical movements, pumps and interference between devices may cause oscillations. When troubleshooting, it’s important to look at your traces over short and longer time frames in order to identify these oscillations.
Once you have identified the type of noise, you can start to build your elimination strategy. A good start would be to identify potential issues in the way your setup is arranged.
Check out Part 2 of “Hunting for Electrical Noise”, where we’ll have a look at general good practice for low-noise recordings. Feel free to share your experience and troubleshooting tips in hunting for electrical noise on your own rig – every relevant entry will receive 30 NeuroGiga Seals
Pingback: Patch-clamp data analysis in Clampfit: action potentials | Spikes and bursts
Pingback: Patch-clamp data analysis in Clampfit: postsynaptic currents and potentials | Spikes and bursts
Pingback: Patch-clamp data analysis in Clampfit: passive membrane properties | Spikes and bursts
I don’t commonly comment but I gotta tell appreciate it for the post on this great one : D.