Important note to all Plexon customers regarding (i) a warning for users of the Plexon HST/8o50-G20 headstage, (ii) filter-induced frequency-dependent phase shifts in field potential signals, and (iii) the Plexon FPAlign Utility which corrects for most of the filter-induced phase shifts.

A paper that was co-authored by two Plexon staff members was recently published in the Journal of Neuroscience Methods ⁽¹⁾ which addresses the issue of frequency-dependent delays of LFP signals through band-pass filters, as well as the phase shifts caused by one particular Plexon headstage (the HST/8o50-G20). We would like to bring these issues to the attention of all Plexon customers, and to notify you as to how you can (i) correct for most of the frequency-dependent timing delays in the LFP band introduced by the filters using the Plexon FPAlign Utility, and (ii) possibly avoid some of these frequency-dependent time delays in the future.

Warning for users of the Plexon HST/8o50-G20 headstage

The “HST/8o50-G20 headstage” (AC-coupled, gain 20x, 8 channel headstage with a 10-pin omnetics connector) has considerably lower input impedance (~47MOhms) at low-frequencies than the other Plexon headstages. This may result in a “voltage-divider” effect, which can introduce frequency-dependent phase shifts in the LFP frequency band. The size of the phase shift at any frequency is difficult to predict, since it is dependent on the impedance of the electrode at that frequency. We have tested the headstage with several different kinds of electrodes and found mixed results, ranging from less than a few degrees of phase shift for some electrodes up to 60 degrees of shift for others. The upper limit of the phase shift caused by the voltage-divider effect is 90 degrees of phase angle.

Therefore if you are using the Plexon HST/8o50-G20 headstage (or the HST/16o50-G20 headstage) and your experimental analyses require precise LFP phase information with respect to spike timing and other external events, we recommend that you switch to our HST/8o50-G1 gain 1x headstage. You may also need to adjust the gain of your preamplifier boards. Plexon will make the headstage exchange and change the preamp gain at no charge. The HST/8o50-G20 headstage is the only headstage in the Plexon product line that has this potentially problematic “voltage-divider” effect due to the lower input impedance at low-frequencies. Please contact Plexon by email (support@plexoninc.com) or telephone (214-369-4957) if you wish to initiate the exchange of your headstage or have any questions. We expect to have new high input-impedance 20x gain 8o50 headstages this summer if you do not wish to change the gain on your preamp boards and can wait a few more months. We at Plexon apologize in advance for any inconvenience this may cause.

A technical explanation of the voltage-divider effect is provided here.

Frequency-dependent phase shifts in the LFP band caused by filters

Plexon local field potential preamplifier boards incorporate per-channel bandpass filters for removing spectral content outside the frequency range of interest. Low-cut filters are used in order to recover signal recording capability after movement artifacts or other voltage transients, whereas high-cut filters are used to remove spike signals from the LFP band and allow digitization of the LFP signals at a reasonable sampling rate (typically 1kHz) without aliasing. In addition, the gain 20x Plexon headstages (HST/8o50-G20, HST/16V-G20, and HST/32V-G20) also contain low-cut filters (< 1Hz cutoff), to remove any DC potential that might accumulate at the electrode metal-brain interface.

These filters, like most filters, inherently produce frequency-dependent phase shifts in the signal. The phase shifts are equivalent to time delays (known as group delays) which likewise vary with frequency. One way of largely avoiding this problem is to make sure the filter cutoff values (the -3 dB values) are factor of ten or more away from the field potential frequency of interest. In the case of the common 0.7-300 Hz band pass LFP filter, for instance, the filter will cause signals between 7-30 Hz to be delayed by less than 16 degrees for any frequency component. Outside this frequency range, however, the delays can be as large as 180 degrees of the period (i.e. 50% of the period).

That’s the bad news. The good news is that it is possible to post-process field potential recordings using the Plexon FPAlign utility so that the timing delays are greatly reduced. After running your .plx file through the FPAlign utility, the timing errors in the frequency components of the LFP signal all the way up to the -3 dB cutoff frequencies will be greatly reduced, and can then be used in subsequent analyses that are sensitive to LFP phase information.

Using the FPAlign Utility to correct for the timing delays caused by filters in Plexon headstages and LFP preamp boards

FPAlign is a utility program that greatly reduces the filter-induced timing delays in field potential channels ("slow continuous channels") in Plexon .plx data files that are caused by the filters in the LFP preamp boards and the gain 20x headstages. The approach that is used is to apply a digital model of the analog filters that were used in the preamp (and headstage, if applicable) for the original recording, but applying the filter to the recorded signal in time-reverse order, thereby cancelling the time delays caused by the original filters. This also has the effect of applying the magnitude (amplitude) characteristic of the filter again, effectively making the cutoff slopes twice as steep. For many applications this is an advantage in that undesirable out-of-band signals such as spikes and very low frequency artifacts are further attenuated.

It is very important to note that FPAlign can only work correctly when the preamp and headstage filtering that was used at the time of recording is known, so that the correct reverse filtering can be applied. Plexon PBX preamps usually have a sticker on the underside of the blue aluminum enclosure, indicating the filter characteristics for the spike and field potential boards. The individual preamp boards are also labeled with their filter characteristics, and if there is any question of boards having been swapped or replaced, these board labels should be examined. See the PBX documentation for instructions on how to disassemble the preamp and separate the boards, or if in doubt, contact Plexon technical support for assistance. You must also determine whether the headstage that was used in recording has 20x gain, and therefore contains a low-cut filter. If you are unsure whether your headstage has 20x gain, please refer to the Neural Headstages page on the Plexon website (www.plexoninc.com/products/headstage/high-impedance_headstage.html) or contact Plexon technical support for assistance.

The values of the resistor and capacitor components that make up the filters in the preamps and headstages inherently have certain “tolerances” – i.e. guaranteed to within a certain percentage (1%, 5%, etc.) of the stated value. The actual filter cutoff values therefore vary slightly from channel to channel. Because the FPAlign utility adjusts for the phase shifts caused by the average (mean) values of the filter cutoffs, there will be some residual error in the temporal phase of each frequency in the LFP signal after the data is processed by FPAlign. Fortunately residual errors are small relative to the periods of the signals. The residual group delays after FPAlign processing is less than 25 degrees (7% of the period) for most frequencies, and never exceeds 50 degrees.

The FPAlign utility can be downloaded for free from the Plexon website (www.plexoninc.com/support/softwaredownloads.html). The download includes a document that provides the LFP-band phase delays as function of frequency for some standard Plexon preamplifier and headstage filters, and the residual phase delays as a function of frequency after FPAlign processing.

Please contact Plexon by email at support@plexoninc.com or telephone (214-369-4957) if you have any questions on how to use the FPAlign program.

⁽¹⁾ Nelson MJ, Pouget P, Nilsen EA, Patten CD, and Schall JD. (2008) Review of signal distortion through metal microelectrode recording circuits and filters. Journal of Neuroscience Methods 169: 141-57.