It has been said, an informed consumer is the best customer. At Plexon Inc, our goal is for users to be as informed as possible, such that they can ask and answer interesting and unique research questions with equipment purchased by Plexon Inc. To meet that goal, we periodically generate content specific to a critical aspect of our core business. This document focuses on the “point of reference,” which is critical for electrophysiology recording systems. The grounding and referencing schemes utilized by electrophysiology labs are often unique to each lab. Even though the fundamentals of grounding and referencing are straightforward, each recording set-up is different, which means each grounding and referencing scheme is different. Additionally, how and why different grounding schemes “work” can lead to mythological understanding of what is and what is not essential to achieve clean, noise-free electrophysiology recordings. Simply put, there is a lot of misunderstanding of how neural data acquisitions work on a fundamental level, and a lot of voodoo that gets applied when setting up shielding, grounding, references, etc. At Plexon Inc, we often get questions about grounding and referencing best practices. While we try to provide advice and recommendations, a basic understanding of the techniques and a clear, concise definition of the terms, is needed. The purpose of this document is to:
- Outline and define how the ground and reference is used in a neural data acquisition system,
- Explain how to decide where within the brain to implant ground and reference electrodes,
- Describe how shielding differs from grounding and referencing, and
- Answer common questions about grounds and references.
If elements of this document are unclear, or if a specific question about a specific recording set-up is unanswered, please get in touch with Plexon Inc.
At its core, every neural data acquisition system operates in the same way: An electrical potential is measured between at least two points; a recording electrode—usually more than one, implanted in an area of interest in the brain—and a “point of reference” (POR)—usually distal to the recording area. It is the connection of the POR that is most critical to the recording, and often it is this connection, how it is used in the acquisition system, where it should be in the brain, etc. that is nebulous and confusing for users. Unfortunately, there is little commonality in the language used between different neural data acquisition systems. Such terms like “ground” and “reference” are used interchangeably in documentation and explanations, or worse, not understood by the user.
The information within this document will be specific to Plexon Inc hardware; however, the information is universal to neural data acquisition. Using examples, this document aims to explain how grounding and referencing works for most neural data acquisition systems, define key terms and variables, and outline some “best practices” when it comes to the ground, referencing, and shielding.
First, the “point of reference” (POR). It is the POR that is most critical to the recording. A poor POR (or worse, no POR) will bring trouble. Traditionally, the POR is a “wet connection” in the brain. “Wet connection” means contact with the brain or at least the cerebrospinal fluid (CSF). Often, the POR is a wire within the electrode implant soldered or connected to one or more skull screws on the animal. The screw penetrates the skull and makes a connection with the brain or CSF (I.e., “Wet”), but does not penetrate the brain. This is the point with which the potential of all the recording electrodes will be measured against—I.e., The “point of reference.” Because of this, the POR is the most critical and most important connection. Without a strong POR, noisy recordings at best, and unusable garbage data at worst, can be expected.
The POR is commonly known as the “ground” because it is mapped to the pin in the implant connector that mates with the ground input of the headstage. There must be a POR mapped to the ground input of the system. Period. Without a POR mapped to the appropriate input of the acquisition system, nothing will work as expected, and useable data will not be acquired. It is possible there could be more than one PORs in an animal or recording set-up, but there absolutely must be at least one.
The POR connection is often conflated with the reference connection in the brain. It is not true to say this is “wrong,” but clarity of the terms and what they mean is key. It is possible to have a reference connection in the brain and a POR. In fact, this is recommended for certain recording set-ups, especially those involving a freely moving animal. A reference connection (or reference electrode, or sometimes just “reference”) is a dedicated recording electrode within the array of recording electrodes. “Dedicated” means the recording electrode is dedicated to being a reference only, not one of the recording electrodes from which usable signal is expected to be acquired.
For example, a 32-channel array could have a 33rd electrode, which is dedicated as a reference electrode. However, for this 33rd electrode to be effective as a reference, two things must be true.
- The reference electrode must be mapped to the pin in the implant connector that mates with the reference input of the headstage and recording system. Typically, recording headstages have a certain number of recording inputs (E.g., 8-channel, 16-channel, 32-channel, 64-channel, etc), and some number of reference inputs and ground inputs. An effective reference electrode must first be mapped to the appropriate pin in the headstage (the reference input).
- An effective reference electrode is the same material as the recording electrodes and implanted in the same region of the brain as the recording electrodes, but typically lower impedance than the recording electrodes. The purpose is for the reference electrode to pick up the same artifacts as the recording electrode (I.e., It is the same material and implanted in the same location), but not the same signal of interest (I.e., It is low impedance compared to the recording electrodes).
One important point: Not all recording set-ups require a reference electrode. Sometimes an additional dedicated electrode can be costly or not available from a manufacturer. Other times an additional electrode within the brain is not prudent from a histology point of view. A reference electrode will afford the user an option for when noise is a problem, but it is optional.
However, when noise is an issue, a reference electrode (I.e., Whatever is mapped to the reference input of the headstage) can be used as a reference for the other recording channels. Most neural acquisition systems include this option. This means, for a 32-channel array with a 33rd electrode used as a reference, the potential on channels 1-32 will first be measured against the POR (remember, the POR is always needed), and then the signal picked-up on the 33rd electrode is subtracted from channels 1-32. This means any noise, artifact, etc. picked up on all the channels (channels 1-32 and the 33rd electrode) will be removed from the recording channels (channels 1-32). This is especially useful for recording set-ups where a chronically implanted, freely moving animal is the subject.
The image above is the pinout of a standard 32 channel Omnetics connector Intan digital headstage, which is compatible with Plexon Inc’s OmniPlex neural data acquisition system, and most other modern acquisition systems. The left portion of the image is a schematic of the Omnetics connector on the input of the headstage, with the Omnetics pin number and Plexon Inc software channel number listed. E.g., Omnetics pin 12 corresponds to OmniPlex software channel 27. The mapping of the recording channels is immaterial to this discussion. Most sophisticated neural data acquisition systems have a channel remapping feature, so the layout of recording channels 1-32 is not important. What is important is the layout and location of the different ground inputs (in green) and the reference input (in red). The Omnetics connector on the headstage will mate with the Omnetics connector on the electrode array, so the positioning of the POR and reference electrode (if used) must match the layout of the input connector on the headstage. This means the POR of the array must be mapped to one of the pins that mates with Omnetics connector pins 1, 18, or 19 on the headstage. At least one of these pins must mate with the POR. As described in this document, one POR connection is essential, but more than one may be worthwhile. Additionally, the reference electrode (if used) must be mapped to pin 36 on the headstage connector.
The 32 channel Omnetics connector headstage was selected as an example because it is a popular headstage; however, the theory applies to most other headstages, regardless of channel count. At its core, each headstage should have a ground input, a reference input, and inputs for the recording channels. You can see this rule upheld in the following images of the pinouts of 8 and 16 channel Plexon Inc digital headstages.
Here are the pinouts for standard Omnetics connector 8 channel and 16 channel Intan digital headstages.
Notice for each headstage, there is a dedicated ground input and dedicated reference input. The ground input of the headstage should be mapped to the POR connection in the implant connector. If used, the reference input of the headstage should be mapped to the reference electrode input in the implant connector. Modern acquisition systems will have a way for the user to designate whether the electrode mapped to the ground pin, or the reference pin is used as the POR. In an OmniPlex system, this designation is set in software. An OmniPlex system that is used with digital headstages (I.e., A DHP OmniPlex system), this setting is found in the Plexon Digital Headstage Processor Device Setting in OmniPlex Server.
In OmniPlex Server, the pxs file (topology file) includes a “module” labeled Digital HST Processor. Right-clicking on this module and selecting Edit Device Options… brings up the following screen.
Within this screen, towards the bottom middle, is an option for the headstage “reference.” This is where the POR is selected. Choosing “grounded reference” means whatever electrode is mapped to the ground pin of the headstage is used as the POR. Choosing “true reference” means whatever electrode is mapped to the reference pin of the headstage is used as the POR. This is where the user makes the physical distinction of which electrode within the array is used as the POR. “Grounded reference” is the default, as the array must have something mapped to the ground pin.
Additional information specific to the OmniPlex system, including the location of the POR selection setting for other varieties of OmniPlex system, is available in the OmniPlex User Guide available on the Plexon website.
Is wrapping a ground wire around a skull screw enough for a good POR?
In short, yes; however, remember the purpose of the POR connection. A ground wire is typically the POR, and this is the most important connection of the entire system. Without a robust POR connection, the recording will be poor at best. Wrapping a ground wire around the skull screw should make a good enough connection, but the user will need to decide if “good enough” is truly good enough. Soldering the ground wire to the screw is better than simply wrapping it.
Is it necessary to have more than one POR?
More than one POR connection can be a worthwhile bit of insurance. Often, multiple skull screws are used to anchor an implant to the skull, depending on the species being used and recording set-up. One “wet connection” within the brain is essential, but more than one POR connection could help prolong recordability from an implant and animal. Many things outside of the control of the researcher can compromise a recording: A skull screw could have not been screwed in far enough to make connection with the CSF; an immune response of the animal could encapsulate the screw and cut off the wet connection; the ground wire making connection with the skull screw could break. If any of these things were to happen, and if there was only one wet, POR connection between the animal and the ground input of the system, the likelihood of recording useable data from that animal and implant in the future is slim. This would be unfortunate too, as often animals are trained for weeks or months prior to implantation. A reasonable and quick step that can offer some preventative insurance from these problems is to use more than one POR connection. Most electrode manufacturers can build an array with multiple ground wires. Using more than one skull screw as the POR will offer some insurance against catastrophe.
What about a Faraday cage for reducing noise?
Depending on the recording set-up and environment, a Faraday cage can be useful. A Faraday cage can be as elaborate as a metal-enclosed recording room or as simple as a cone of aluminum foil around the implant. Regardless of how complicated the cage, how a Faraday cage is used (and more importantly, how to shield a Faraday cage) is unrelated to the POR connection in the animal. Often the source of shielding for the Faraday cage and the POR are conflated. Additionally, a “complete” Faraday cage will be most effective at shielding environmental noise from the recording set-up. “Complete” means a completely enclosed (or as close as possible to completely enclosed) cage. A useful metaphor when thinking of a Faraday cage is a shark cage used by divers. The shark cage is the Faraday cage, protecting the set-up (the diver) from the shark. A complete cage (I.e., One that completely encloses the diver) will be best.
What is most typical and traditional is for a Faraday cage to be shielded from the environment using the ground input of the acquisition system. This is different than the ground input of the headstage, which is the POR of the recording. The ground connection of the acquisition system is typically from the outlet in the wall, also called Earth Ground. Some acquisition systems have an isolated power supply, which can also be used as the source of shielding for the recording set-up. With Plexon Inc acquisition systems, it is possible to externalize the ground of the acquisition system via a green grounding wire on the amplifier or “blue box.” It is important to know this is not a ground input for the system, but instead a way for the user to externalize (I.e., Output) the ground for use as a source of shielding. Most traditionally, this source of shielding is connected to a Faraday cage. Alternatively, this externalized source of shielding can be used to ground potential sources of noise around the recording set-up. E.g., Stereotaxic frame or manipulator, guide tubes, surgical lights, etc.
What about digital referencing?
Digital referencing can be a source of confusion for many users. It is important to remember the difference between analog referencing and digital referencing. At its core, analog referencing typically occurs before acquisition. Everything discussed in this document so far has been related to analog referencing. Analog referencing (I.e., The subtraction of signal on recording electrodes from a dedicated reference electrode in the brain) occurs before the signal is acquired. This is partly why the POR and reference electrode set-up is so key. Once acquired, there is no changing how the signal was acquired, meaning there is no way to change the reference or ground inputs used offline. Digital referencing is an option for the user to route acquired signals through different referencing options. Sophisticated acquisition systems include the ability to do this digital referencing online, but it is possible to do this offline. There are three common digital referencing options:
- Channel-to-channel—With this method, the acquired signal on a channel can be used as a reference for other channels. This means a channel with artifacts and no spikes or signal of interest can be subtracted from other channels with similar artifacts that obfuscate spikes or signal of interest. The purpose of this type of referencing is to subtract away the artifacts without affecting the spikes or signal of interest. The channel-to-channel digital referencing set-up can be complicated, as multiple digital references can be used to rid artifacts from different groups of channels. This is especially useful when recording from multiple brain regions, as an artifact common to one group of channels can be referenced away with a digital reference independent of another group of channels with a different artifact.
- Common Average Referencing (CAR)—With the CAR referencing method, an average signal from a group of acquired channels is used as the digital reference. Sophisticated acquisition systems can do CAR (and CRM, described below) online, but this method can also be applied offline. With CAR, the user selects a group of channels with a common artifact or noise. From this group of channels an average signal is calculated, and it is this average signal that is subtracted from all the channels. Typically, the more channels that can be included in the average calculation, the more effective the technique.
- Common Median Referencing (CRM)—This method is similar to CAR: the user selects a group of channels with a common artifact or noise, from which a median signal is calculated, and it is this median that is subtracted from all the channels.
Both CAR and CMR are also useful for recording set-ups where groups of channels may be affected by different artifacts, but all channels have spikes or signals of interest. One downside of the channel-to-channel method is whichever channel is selected as the digital reference cannot have spikes or signal of interest to be effective. Otherwise, a channel selected as a digital reference that has signal of interest will add the signal to the channels it is referenced to. This could lead to inverted spikes or signal of interest from the reference channel on the other recording channels. Effectively, the reference would be adding spikes to other channels, not just removing artifacts. This is precisely the reason the dedicated analog reference in the brain should be lower impedance than the recording channels. A reference electrode that is picking up spikes or signals of interest is bad and should not be used as a reference.
What about field potentials?
Field potentials can be tricky and should be carefully considered when choosing a reference option. An important question the researcher should ask is—What do I value most? Traditionally, recording spikes from a handful of channels is much more difficult, and therefore much more valuable, compared to ensemble multi-unit activity or local field potential (LFP). Typically, a lab interested in LFP or multi-unit activity will implant the reference electrode in the brain in an area different than the recording electrodes, as to not pick up the oscillations or signals of interest. The most important details of the reference electrode do not change. It should be placed in an area likely to pick up artifacts similar to the recording sites and it should have lower impedance than the recording sites, such that it does not also pick up spikes.
What about recording from multiple brain areas?
Recording from multiple brain regions is becoming more common. With multiple groups of recording electrodes, each within a different area in the brain, a reference electrode can be more difficult to place. Some acquisition systems offer more than one reference input. This could be a solution, as reference input #1 could be implanted in recording area #1 and reference input #2 could be implanted into recording area #2. The user would then need to carefully select which channels should be referenced to reference #1 and which to reference #2. But some headstages only have one reference input. In a situation where only one reference input is available, and a dedicated reference input is truly required, a reasonable compromise could be to implant the reference electrode in one of the two areas. This is suboptimal, as the artifacts common to recording area #1 may not be common to recording area #2. However, a reference electrode implanted into one of the two areas would offer the user some reference options. Another alternative is to implant the reference electrode in a third area, potentially nearby or adjacent to the two recording areas. This would offer a reference within the brain nearby the recording (I.e., To pick-up artifacts common to the recording areas) but should not pick-up spikes or signals of interested (I.e., Because the reference electrode should be lower impedance compared to the recording electrodes).
What if my electrode only has a reference electrode?
It can be tricky and misleading to think of the ground and reference “channels” of the array without considering to which pins in the implant connector the channels are mapped. Instead, it is best to think of which electrodes in the array are mapped to the ground pin(s) and which are mapped to the reference pin(s). Most headstages have at least one ground input and one reference input. Likewise, the connector on the implant will have a pin for the reference input and a different pin for the ground input. It is unlikely an electrode array would be manufactured with an electrode mapped to the reference input pin but not also have an electrode or wire mapped to the ground input pin. Since the POR, which is always the ground input, is required, recording from an array without a ground input will be impossible. What is more likely to happen is the terminology of the different electrodes will be mixed up. It is common to conflate “reference” and “ground.” Sometimes, users say “the ground is the reference” or “I only have a reference, I do not need a ground.” The loose use of terminology is a part of the reason why this document was created and shared. Instead of referring to the reference or ground abstractly, it is better to refer to which pin the electrode is mapped to. This way there is no uncertainty about the set-up. So, in the case where an electrode only has a reference channel, what is likely the case is the manufacturer of that electrode has a wire mapped to the pin of the implant connector that mates with the POR input of the headstage, which they refer to as the “reference.”
What about tying the ground and reference together?
As discussed, each electrode array must have a POR. However, since most headstages have a ground (POR) input and a reference input, there can be confusion about what to do with the reference input when not used. It is not required that all recording set-ups use the reference input. It could be adding a dedicated reference electrode to the array is too costly, not possible from a manufacturing standpoint, etc. Unused, open channels within a headstage can be bad. “Open” means the input is left unpopulated, or sometimes called “floating.” Effectively, the input has nothing attached, and therefore is open to the environment. Floating channels within a headstage can be a source of noise. These floating inputs can pick up environmental noise, which will be added to the recorded datafile. It is best practice and strongly recommended that unused channels are tied to ground. This means an unused reference input (I.e., No dedicated reference electrode in the implant) should be tied to ground. This can be done by simply soldering a wire within the implant array between one of the ground inputs and the reference input. Most electrode array manufacturers offer this option.
There are many ways ground and reference electrodes can be set-up in a recording. Many factors contribute to the placement of ground connections, whether to use a reference electrode, the type of shielding needed, etc. At the most basic level, however, grounding is simple: the potential is measured between an electrode implanted in an area of interest in the brain and some point of reference. Approaching the problem of grounding and referencing from this simple idea is best. Shielding and referencing set-ups can be complicated. But at the core, remember, it is the POR that is most important.
Comments and questions should be directed to Andrew Klein (firstname.lastname@example.org). If you do not find answers to a question you have about grounding, referencing, shielding, noise, etc., please reach out.
Written by Andrew Klein