scope trace graticule. The filter has a band
reject at 19 .1 MHz and a notch depth of
28 dB with a Q of 4. 3.
Figure 7 is a sweep of a 100 MHz
wideband amplifier I recently completed:
110 MHz center frequency; 60 MHz start
frequency; 160 MHz stop frequency; with a
total span of 100 MHz at 10 MHz per
major scope graticule. It just starts to roll off
at 108 MHz. Considering I needed to know
the performance here from “DC” to well
beyond 100 MHz, I had to begin with the
lowest RF band and just spin the RF band
switch right on through to the last position.
All were run at 100% sweep, and due to the
excellent leveling of the RF output, this
displayed a reliable presentation from band
to band. When checking any wide band
response of a DUT, use this method. All
lower bands here displayed a flat response.
Figure 8 is the same setup as in Figure
7, only this time with a detector probe (as
shown back in Figure 4) attached to the output. I built
two of these probes: One is constructed in a very small
metal enclosure with BNC connectors on each end and
the input terminated into a 50Ω impedance. The other
was built inside a felt tip marker housing with shielded
leads and terminated with a BNC connector. Although you
definitely want a Schottky type diode here, after
experimenting with several, the BAT62 was the winner.
This gives fairly linear performance down to 0 DBM levels
(about 600 mv p-p), and then enters its square law region
where response starts to drop off rapidly.
Some Words on Sweep
To begin, sweep testing is not an exacting science. In
an ideal world, as the scope’s horizontal display is
traveling along its 10 major divisions of length, the
corresponding generator frequency would be in an exact
step to those divisions in regards to linearity. This would
require that all “major players” in all these devices were
perfectly linear. Any scope manufactured in the last 40
years probably has a near perfect ramp driving its sweep
trace. The adapter’s Vt ramp is perfectly linear. Now
comes the weak link in the chain: the varactor diode.
Varactors do not have good linearity as far as their
capacitance value vs. the applied bias voltage (Vt). They
all tend to have a sag or bulge somewhere in their range
of Vt. There are different methods to partially compensate
for these errors in the Vt driving circuits, but none of them
can completely correct this shortcoming. There are other
methods to correct this by using servo control that
analyzes the non-linearity through a separate and parallel
circuit path, establishes an error signal, and feeds it back
to correction amplifiers in the basic Vt circuitry. Although
this method has superior performance, it is very complex
and expensive to build. Any further discussion of this
subject is beyond the scope of this article, and beyond the
degree of accuracy intended for a service grade
instrument. The best I could do on this design was to
stretch out the tuning bandwidth as far as possible, and
still maintain a respectable linearity for that given
bandwidth. Like most design work, it involves trade-offs.
After deciding on the optimum tuning range, I ran a
series of linearity tests. There are different ways to do this.
I chose to measure the frequency at each major scope
graticule division ( 10 in all) as the sweep trace crossed
them. I then computed the frequency error in percentage
as to what it should be in a perfectly linear sweep.
I ran tests on each band at the 10%, 30%, and 100%
sweep widths, with center trace frequencies always
referenced to the center frequency of that particular band.
I made measurements at each graticule division for the
sweep widths just mentioned on each frequency band —
240 test points in all. I then averaged these for an overall
specification. Here are my results:
100% Sweep (of total band): 1.2% average error and
2.4% maximum error.
30% Sweep (of total band): 0.2% average error and 0.4%
10% Sweep (of total band): 0.02% average error and
0.05% maximum error.
As can be seen from this data, the narrower the
sweep span, the greater the accuracy. Recall that where
the best accuracy is required is in narrow band sweeping;
wide band sweeping is not so critical — especially the
farther we move away from the center frequency. Due to
screen parallax and the thickness of the scope trace, it is
February 2017 37
■ FIGURE 6.
■ FIGURE 7. ■ FIGURE 8.
■ FIGURE 5.