This unit was built solely to do something useful with a 5MHz t.c.x.o. which
I rescued from a scrapped board.
It provides time-marks which can be used to check oscilloscope calibration.
I added the voltage source so as to check vertical sensitivity.
The transistor following the tcxo is a doubler. The coil consists of 20 turns
of 0.2mm enamelled wire, close wound on a 5.3mm dia former with a ferrite slug.
I tried putting an aluminium can over it (for convention’s sake) but it completely
destroyed the Q. The coil was tuned by adjusting the slug, on its way out, to
a point past maximum amplitude which gives more or less equal amplitudes on
alternate cycles (by looking at the divider input with a high-Z ‘scope probe.)
They tended to be unequal because the output of the tcxo wasn’t symmetrical
between its positive and negative peaks.
The diode clamps the positive excursions to the divider supply rail.
The chips in the main divider are dual decades.
You can use either of the two independent counters "first" and feed its output (Q4) to the other depending on your board layout. And, although pins 1 and 9 are shown on the data sheet as "clock", in which case the enable pins, 2 and 10, should be held high, you can instead clock pins 2 and/or 10 and hold pins 1 and/or 9 low. (Again for board layout considerations.)
The selected output goes to the divide by 5 input of the 7490 first. The divide
by 2 is last in order to obtain a 50% duty cycle at the output. The differentiating
capacitor at the input of the BFX48 is switched, by another wafer, in order
to provide a more or less constant percentage pulse width in each of the switch
positions.
The measured rise-times of the outputs are 48ns for the 50% and 5.5ns for the
narrow outputs respectively. These were made with a 10M//11pF probe on a 'scope
with a 3.5ns rise-time. That of the 50% output in particular could be improved
by reducing the value of the collector resistor if you can afford the extra
current.
If you look at the output without using a high-Z probe it cocks the rise-time
up, especially with just a co-ax lead. Remember that RG-58 cable has a capacitance
of around 30pF/ft which appears in parallel with the 'scope input capacitance,
all of which has to charge up.
The 8th position of S1 disconnects the supply from all the above and applies
it to a 10V reference chip. The pot at its output is a 10-turn one with 0.25%
linearity. Note that the accuracy of this output will depend on the linearity
of the pot and not its absolute value. The 9k in series should be selected to
be exactly 9 times the measured value of the pot, the absolute value having
no importance.
A cheaper alternative would be a switched attenuator as suggested in the diagram.
Because this output has significant, and variable, source impedance it is suitable
only for checking devices with a high input impedance such as scopes and digital
(and some analogue) meters.
In retrospect, if I were to build it again I would buffer the pot’s output with
a voltage follower using something like the OP-07 or a more recent similar device.
They have very low offset and very low offset drift.