Output Gain & Filtering
Comprehensive facilities for amplifying and filtering the current signals (or voltage signals in current clamp mode) are included. The gain is switchable from x1 to x1,000 in steps following a x1 x2 x5 sequence. The incoming current signals have a value of 10V at full scale on the two ranges of 1nA and 100nA, giving sensitivities of 100pA/V and 10nA/V respectively at x1 gain. The incoming voltage signals in the current clamp modes have already been amplified by ten. We doubt whether the highest gains will be of practical value for patch clamping, but we included them because the capacity was there and because they might be useful in other applications.
It can be particularly useful to have a record of the gain setting, so we have included the popular facility of an output voltage that can be recorded by the data-capture system to provide the gain information. We provide similar voltage values to those used on other commercial equipment, and the telegraph output also takes the gain of the headstage into account. The output at a gain of 1mv/pA in voltage clamp mode is 2.5V, rising in 0.5V steps for each gain increment, so it varies between 4V and 9.5V over the gain range of 1 to1000 in patch mode (10mV/pA to10V/pA), and between 1V and 6.5V (0.1mV/pA to100mV/pA) in big cell mode. The telegraph output in the small cell mode varies according to the current range selected for this mode on the rear panel switch, which can be either of these. In current clamp, a voltage output of 1mV/mV (i.e. unity gain) would also give 2.5V, but since the voltage is already at x10 gain before being sent to the gain selector, the total gain varies between 10 and 10,000, giving a telegraph voltage range of 4V to 9.5V.
It is also worthwhile if not essential to include an adjustable low-pass filter after the gain stage, since patch currents are inevitably noisy, and the noise depends on the signal bandwidth. Therefore a low-pass filter is used to remove noise that is at higher frequencies than those that are of interest for measurement of the patch currents. Unfortunately, filters do not have an ideal (step) frequency response. In the simplest, or single-pole, type, which just consists of a single capacitor and resistor, the gain halves with each doubling of frequency above a given "corner" or "cut-off" frequency, but steeper responses can be obtained by cascading several such sections together. In common with normal practice for patch clamping, we use a four-pole filter. In filters having more than one pole, the shape of the response around the corner frequency can be modified by feedback to give a more or less sharp transition between a flat response at lower frequencies and a smoothly falling response at higher frequencies. A sharp transition may seem better, but such filters distort the waveforms of transient signals to a greater extent, and again in common with other patch clamps, we use a Bessel characteristic in order to achieve the best compromise.
Changing the cut-off frequency of a four-pole filter requires four components (usually resistors) to be switched, and for a larger change the other four frequency-determining components (i.e. the capacitors) may have to be switched as well, which makes for a messy operation in practical terms. We therefore opted for an alternative method, which also has the advantage of allowing the cut-off frequency to be continuously variable. We had already devised a variable filter circuit for our fluorescence equipment, using transconductance amplifiers - the output of which depend on a control current as well as the incoming signal - to simulate variable resistors, and this turned out to be an ideal application for it. The other possible technique is the switched-capacitor filter, and some very smart integrated circuits of this type are available, but this method requires a digital control frequency, and we were therefore concerned about the possibility of digital interference reaching the sensitive parts of the patch clamp electronics.
Although the transconductance technique requires a larger number of devices than other methods, it gives excellent performance without causing any interference risks. We use a switched control voltage to vary the frequency over a tenfold range, and we then switch the four capacitors to change the range by further factors of ten. As mentioned in the introduction, the use of analogue switches here, as elsewhere in the Optopatch, allows us to perform four-pole switching of the analogue signals by only single-pole switching of the digital control voltage, which feeds all four analogue switches together. It also removes the need to send any of the signals all the way out to the front panel and back, which can be a problem, especially at higher frequencies, because of the unavoidably greater stray capacitance associated with the wiring.
The other advantage of electronic control is that it lets us offer an accessory module (which will fit into one of the expansion slots), to provide an additional four poles of filtering, giving eight-pole Bessel filtering altogether. Although four-pole filtering gives good results under most conditions, eight poles do seem to offer a worthwhile improvement, but that also seems to be as far as it is worth going in practice. For us, standardising with four poles and offering another four as an option seemed the best compromise. Our module connects to the existing control signals, so is easy to incorporate into the system. All that is needed, apart from inserting it into the signal pathway, is to change the corner frequencies and damping settings of the existing four-pole filter to make them appropriate for being part of an eight-pole one, and that is easily accomplished by changing some jumper settings on the main circuit board.
The base frequencies of the filter are 1Hz, 10Hz, 100Hz, 1KHz and 10KHz, set by the freq range selector, and they can be varied over the range x1 to x10 by the freq value selector. We had originally intended to make the freq value selection by a continuously variable control, but consultations with experienced patch-clampers suggested that it would be clearer to provide a selection of fixed values instead. However, we have provided a wider choice of values than is usual (x1, x1.5, x2, x3, x5 x7 and x10), and the possibility of providing a variable control voltage via the computer connector, when the system is under computer control, has been retained.
The lowest frequency ranges are not appropriate for patch clamping, but they may be appropriate for other applications. In particular, they may be useful for filtering the signals generated by the capacitance measurement facility, and a switch is provided to connect the capacitance signals directly to the filter. To do this, the switch should be in the "cap" position, when the filter takes its input from the imaginary phase output, otherwise it should be set to the "current" position.
In addition to the voltage output representing the gain, several signal outputs are available on the rear panel. These are:
- The voltage or current output from the headstage, at the x1 gains specified above (including the effects of any leakage current compensation in voltage clamp mode).
- The same signal, after amplification by the gain stage.
- Another output from the gain stage, but filtered with a three-pole Bessel response at 25KHz, which can be useful for general monitoring purposes.
- The output from the four-pole variable filter.
