Drivers | Sanatron | Example 1 | Example 2 | ||||||||||
Gas Devices | Puckle | Miller | Transitron |
Before we start it would be as well to define exactly what waveform we are after. First we have the region marked as 'S' in the diagram opposite which is the portion that scans the CRT. Here we need a linear ramp, preferably of quite high voltage swing for driving the CRT plates. Secondly, we have the flyback period 'F' whose time we strive to minimise. The third region 'I' is an idle period where we would like to wait for a trigger pulse to come along and start another ramp. |
At the heart of the design of almost all oscilloscope timebases is the principle of charging and discharging a capacitor. The charging waveform is illustrated above. As can be seen the ramp produced is pretty non-linear. However if we were to only operate the circuit over a small part of the ramp (as shown circled) we can get a significant improvement in the linearity. We also need some mechanism to rapidly partially discharge the capacitor at the end og the scan portion.
Enter the neon bulb.
These bulbs contain low presure neon gas with two electrodes inserted. Once
a certain voltage (the "striking" voltage, VS)
is reached, the neon gas ionises and the bulb conducts heavilly. Assuming the
current is limited in some way to avoid destruction of the bulb, the neon maintains
a constant voltage across it, VE, which is less than
the striking voltage. However, in order to remain in conduction, a certain minimum
current needs to be maintained otherwise the neon gas de-ionises and the bulb
ceases to conduct. If we place one of these bulbs accross our basic R-C network
then once the capacitor charges to VS the neon strikes,
discharging the capacitor to the voltage VE. If we make
resistor R large enough, the minimum current required for the neon to remain
conducting can not be maintained and the neon ceases to conduct. The circuit
then repeats the charging cycle and so we obtain an oscillation, as illustrated.
We now have a crude waveform suitable for use as a timebase. However it does
suffer from several shortcomings :-
a) | For acceptable linearity, the oscillation amplitude is necessarily small |
b) | The value of 'R' needs to be high in value to ensure the neon gets extinguished. In addition, 'C' needs to be kept much larger than any stray circuit capacitance. This places a limit on the frequency of operation. |
c) | The gas takes a finite time to ionise and de-ionise, limiting practical circuits to a maximum of 10-20KHz. |
The
thyratron is a gas filled triode. If the valve is held cut-off by a suitably
large -ve voltage applied to the grid, the thyratron behaves in a similar fashion
to the neon bulb. However we are not limited to just gasses since the heater
can cause the evaporation of other non-gaseous elemnets such as mercury. Mercury
has a quite low extiguishing voltage which allows a much larger amplitude of
oscillation compared to the neon bulb, although this must be at the sacrifice
of scan non-linearity. However, the device is also basically a triode. If the
voltage on the grid is raised above its cut-off voltage the triode will begin
to conduct as per normal triode action. If there is sufficient voltage on the
anode then this conduction can cause ionisation to occur even if the anode voltage
were below the striking voltage.
The cut-off voltage of a triode is not a constant value. Its value varies with
the applied anode voltage as shown in figure 5. Thus if we vary the grid voltage
can can cause the thyratron to rigger when its anode voltage has reason to some
value below its striking voltage. We can thus use the grid to control the amplitude
of the generated waveform.
The
effect is illustrated in figure 6. With the grid held very negative (V1),
the waveform reaches the striking voltage 'A' as if we were dealing with a simple
bulb. If we raise the grid voltage to V2 then when the anode voltage reaches
'B' the cut-off voltage becomes more negative than V2, so conduction starts
due to triode action triggering ionisation.
If we now add a capacitor (CT in figure 3) we can feed
synchronising pulses into the grid. These pulses are such that they raise the
grid potential and should this occur near what would have been the trigger point
then firing will occur synchronous with the trigger pulses. This is illustrated
by the pulses added to V2 in figure 6 ; the amplitude of the first two pulses
is insufficient to cause conduction as the cut-off voltage is not yet negative
enough (i.t. the anode voltage is too low). The third pulse is just sufficient
to cross the cut-off voltage and hence triggering occurs at point 'C' instead
of 'B' as would have been the case in the absence of the trigger pulses.
Triggering, good amplitude, what a shame about the non-linearity. What we need
to do is instead of driving the capacitor via a resistor we actually need to
drive it with a constant current.
Enter the pentode. The pentode has a very high slop resistance,
that means that its anode current is completely independant of the applied anode
voltage (OK, its not perfect and this only applies at higher anode voltages).
The anode current is controlled by the grid, so by appropriate choice of grid
bias voltage we have in fact a constant current sink.
If
we simply replace the timing resistor with this constant current generator then
the cathode would be connected to the capacitor and hence vary with the ramp
waveform. However, the bias voltages on Vg and on the
screen need to be set with respect to the cathode which would make biasing somewhat
complicated. We can get around this by turning the thyratron circuit "upside
down" so that the timing capacitor now connects to HT and the timing resistor
would have one end grounded. The result is shown in figure 8. Altering P1 varies
the grid bias of the pentode, hence varying the current and therefore speen
of the ramp. P2 generates the grid bias for the thyratron thus controlling the
size of the waveform, though it should be noted that the P2 resistor chain acts
as a load on the timing capacitor and hence influences the linearity.
We may have addressed the issue of linearity but there is still the issue of
limited speed. The fundamenal limit is the ionisation/deionisation time of the
gas, so clearly the gas must go !
NEXT PAGE : The "Puckle" Timebase
Copyright © J.Evans 2002 |
CLICK HERE TO RETURN TO THE OSCILLOSCOPE PAGE |
Last updated 15th ***** 2001 |