SMDU Commissioning continued
|
|
Continuing with
my repair of the SMDU, I'll recap on the progress so far.
Presently (1) a bad 15 volt
regulator and (2) the display has a strange fault. No doubt there
are several other problems but until these first two are resolved
those will have to wait.
(1) If the equipment is turned
on the display carries all zeroes because the 15 volt rail from
Pin 2 of the PSU is sitting at minus 0.75 volt. This voltage
was much the same when I swapped the two 7815 devices but if
left sometimes switches back to plus 15 volts. Or, if I connect
a bench supply (via a diode) to the 15 volt rail and slowly increase
the voltage, at 8 volts the internal PSU suddenly starts to work
correctly and sits at plus 15.02 volts. If I then remove the
bench supply this voltage remains steady, but switching on from
cold It's back to minus 0.75 volt.
Could there be something wrong
in the hardware fed by the 7815? Powering the rail alone from
a 15 volt bench supply it draws about 700mA and after say 20
minutes this drops to about 600mA. Disconnecting the set of lower
boards at ST27 results in a current of 200mA dropping to 100mA
so that anomaly is not in the frequency processing area (under
the cover underneath the chassis) but maybe in one of the set
of basic oscillators or in the display area. What I haven't established
is whether this 100mA difference in drain at the 15 volt rail
is associated with the bad supply voltage.
The display fault is odd. Firstly
there's an extra "decimal point" in most settings,
and if the frequency is tuned, say to 50MHz an oscilloscope will
show this frequency but the display shows something different.
For example at 50MHz on the
drum dial this is indeed the value read on a scope but the display
reads 34MHz and at 62.6MHz true output the display reads 25MHz.
Both with an extra decimal point. The fault is likely to be in
the TTL area driving the display. One point of note is I'm using
a pair of 5 volt regulators instead of the 5.2 volt original.
This could result in a low voltage, at one or more chips, but
I upped the voltage to 5.6 volts and the problem remained so
it's not the result of a low voltage within TTL circuitry.
My initial guess is maybe one
or more bad chips on circuit boards Y7, Y71, Y72, Y73 or Y75..
but it wasn't (it was due to a serious design weakness which
is endemic in R & S equipments).
|
|
 |
 |
|
First, I tackled board
Y75, whose circuit is shown above, because this handles the decimal
point circuitry and an extra DP was one of the errors. To remove
Y75 so I could test its components was a little puzzling because,
although I'd removed three small screws and several cables, it
was stuck due to its buttons protruding too far. I noticed right
next to the board was a paxolin tube carrying an M5 bolt that
was captive but sprung upwards when unscrewed. I noticed another
M5 screw opposite and a pair of grooves in the sides of the metalwork.
I unscrewed the leftmost bolt with the other in its sprung up
position and I found that the whole aluminium tray could be slid
backwards and upwards allowing Y75 to be lifted out.
I tested the six TTL chips and
the three diodes but it was only the three capacitors that gave
bad readings. I fitted two new 22uF and a new 47uF because the
old ones had high ESRs.
I refitted the board and checked
the voltages at ST2 against the truth table for the selector
switches and apart from noting that the optional UHF extension
made the table look a bit odd there appears to be an error with
Pins 2 & 4 not flipping from "HLL" to "LLH".
These pins are fed from the Range Switch board. Here I found
two "Tech Author" errors as in the truth table I believe
there should be 4 not 3 entries. Pin 5 (KHz/MHz) is missing.
That would most likely resolve the identical code for some entries.
Tracing connections I can see a drawing for ST79 (= ST3+ST4+ST5)
but as yet ST80 is a bit vague... probably including ST1, ST2
& ST114?
ST 80 is the large black connector
on top of the Counter Switching board Y75. Some of its wires
must originate on the Range Switch board Y10?
Also the entries for the truth
table in the drawing above do not exactly correspond with the
push buttons on Y75. The last two entries reflect the setting
of only one push button on Y75 and all the push buttons on the
Range Switch Y10 (the lowest switch being 0.14-50MHz and all
the others 50-1050MHz). The bad result seems to be a loss of
signal between the two boards as the Range Switch push buttons
are not reflected in the truth table results.
In my tests the extra decimal
points are clearly wrong.
About this time in testing
I decided to sort out the intermittent operation of the 7815
feeding Pin 2 of the PSU circuit board which was really annoying
me. By experiment I found that adding a couple of diodes in series
and connecting the anode of these to Pin 3 (the output of the
working 7815) and the cathode to Pin 2 (the output of the bad
7815) kick-started the bad 7815.
|
 |
|
Picking up the correct
connections was a bit tricky but by checking the repair manual
which gave the track on the underside of the power supply board
and judicious use of my multimeter I was able to add the new
diodes (2 x 1N4005) as seen below. |
 |
 |
 |
|
Whilst on the subject of the
power supply. Turning back to the new 5 volt circuit which replaces
the faulty MC1569R-based circuit.
This is the final version of
the 5 volt regulators. The two 5 volt outputs are fed to the
two outgoing 5 volt rails with each of the two uA78H05SC regulators
(above right) supported by ballast resistors to accommodate some
of the heat losses. I arranged that each regulator has about
10 volts input under load by feeding them with an 8.2 ohm and
a pair of paralleled 6.8 ohm ballast resistors on an aluminium
bracket (shown opposite). One feed carries about 300mA and the
other about 1Amp making the ballast resistor dissipations around
750mW for the 8.2 ohm and 1700mW for each 6.8 ohm.
|
 |
|
Still having trouble identifying
the DP fault, I removed and examined Y10, checking diodes etc.
Everything seemed OK so I plugged back the board but left it
out of the chassis. Switching on I discovered the extra decimal
points had gone and all bar two of the ranges were working again.
All I could think of was the connectors at the end of Y10. These
use twin lines of pins/sockets and had I inadvertently fitted
a plug wrongly? That would perhaps explain the sudden appearance
of the extra DPs and frequency errors in the ranges. A serious
weakness in R & S equipments is that lots of connectors can
be plugged in wrongly as the plugs and sockets are not polarised.
I checked the pin voltages at
ST2 and the truth table values were now correct (as shown below).
|
SWITCH SETTING |
ST2-PIN 2 |
ST2 PIN 3 |
ST2 PIN 4 |
DEC POINT |
|
EXT 0.5-1GHz GOOD |
L |
H |
L |
NA |
|
MEASURED |
NOT APPLICABLE |
NOT APPLICABLE |
NOT APPLICABLE |
NA |
|
EXT 20-525MHz GOOD |
L |
L |
H |
DIGIT 4 |
|
MEASURED |
0.5V = GOOD |
0V = GOOD |
3.7V = GOOD |
DIGIT 4 |
|
EXT 15Hz-30MHz GOOD |
H |
L |
L |
DIGIT 3 |
|
MEASURED |
4.2V = GOOD |
0V = GOOD |
0.06V = GOOD |
DIGIT 3 |
|
INT 0.14-50MHz GOOD |
H |
L |
L |
DIGIT 3 |
|
MEASURED |
4.2V = GOOD |
0V = GOOD |
0.1V = GOOD |
DIGIT 3 |
|
INT 50-1050MHz GOOD |
L |
L |
H |
DIGIT 4 |
|
MEASURED |
0 = GOOD |
0V = GOOD |
3.7V = GOOD |
DIGIT 4 |
|
INT AF GOOD |
H |
L |
L |
DIGIT 6 |
|
MEASURED |
4.3V = GOOD |
0V = GOOD |
0.1V = GOOD |
DIGIT 6 |
|
|
|
Here's a couple
of Truth Tables (for Y73 & Y7 plus Y10) so that the logic
levels can be checked.
The second drawing can be clicked
to enlarge it.
|
 |
 |
|
Now that the DP fault
and the strange frequency readings are resolved I need to tackle
two non-working oscillators viz Range 7 (63.5-88MHz) = Y12 and
Range 5 (118-158MHz) = Y14. |
|
|
|
 |
My first plan was to see
if the working circuits differed from the two that had failed
by comparing Y12 & Y14 with the working Y11 opposite. All
use the same power supplies so we can rule out that.
Tuning arrangements are fundamentally
similar and components much the same so it looks like either
a fault common to Y12 and Y14 on the Range Switch board Y10,
or random component failures in both Y12 and Y14. When I'd first
tested the SMDU only one range had been bad.. now there were
two.
I checked Y10 because this was
currently accessible outside the chassis and it seemed OK.
One puzzling thing I'd seen
earlier was that pointing to one of the set of microswitches
around the tuning drum gear was a label marked "bad"
stuck on the chassis pointing to S11. Could others be bad?
These switches are used to feed
plus 15 volts, or a tuning voltage, to the 2N4416
transistors in the oscillator circuits ie. cutting off the
transistor or applying an oscillator tuning voltage to a varactor
diode.
|
|
|
Below is a drawing showing the
oscillator circuits.
Around the drum dial is a stepped
ring which, as it rotates, engages or disengages with the microswitch
levers. Some (S11-S17) are wired to the oscillators and a few
(S18-S20) associated with the optional 1GHz frequency extender.
|
|
 |
 |
Once the aluminium
chassis carrying Y75 etc is slid back and fixed in its test position
the tuning drum assembly can be accessed and the upper plain
circular panel removed. Below this is the cover shown on the
left which can then be unscrewed.
Once this cover has been detached
you can see the interior of the drum (see below) with its set
of six oscillators. I noticed a red dot on the lid and a corresponding
red dot on the cover so you can identify the modules (Y11, Y12,
Y13, Y14, Y15 and Y16).
Left: view after removing six
screws and detaching a plain circular panel.
Note the markings for adjusting
the tuning ranges to match the drum frequency engravings, L(n),
C(n) & C(n+1)
Below, after removing a further
twelve screws and the engraved circular panel, you can see the
set of oscillators and the output circuit board Y20.
|
|
 |
|
Around the periphery of
the drum dial are lots of microswitches of which S11 to S16 supply
a 15 volt positive supply to the triangular circuit boards (note
pairs of solder blobs on the left edges). One blob carries the
+15 volts and as the drum is rotated the appropriate microswitch
engages disconnecting the positive voltage allowing the oscillator
to be activated if selected by the Range Switch. That +15 volts
is for cutting off the selected oscillator when the left or right
end of its tuning scale is exceeded. Unfortunately, if a microswitch
is stuck in its closed state, or has moved out of position relative
to the drum, it leaves the +15 volts in place so that particular
frequency range will be disabled (apparently giving a fixed reading
of 506MHz in my example).
My inoperative ranges Y12 &
Y14 were in this state. I proved this by monitoring the solder
blobs whilst rotating the drum. A squirt of switch cleaner on
the microswitch failed to sort this out because it's a mechanical
rather than an electrical issue.
|
 |
|
|
Switch cleaner failed
to fix the sticky microswitches because the problem was in the
hinge whose stiffness was greater than the return spring on the
lever.
Because the microswitches are
arranged in roughly equal steps some are pretty well inaccessible.
To access microswitch S12 I had to move** the whole of the tuning
assembly to the left. This enabled me to see S12's hinge. I dropped
some thin oil on this and it seems to have fixed it. S14 was
also stiff, but as this is covered by a circuit board, I had
to dribble oil onto its lever and hope it got to the hinge. Turning
on the SMDU proved all ranges were now working.
** To move the tuning chassis
there are four M4 screws fitted through shock mounts which need
removing. Then, after unscrewing the outer tuning knob (5/64
Allen key) and inner knob (three M3 screws) the chassis
can then be slid over by an inch or so which reveals more of
S12 enabling oiling of its hinge.
Below, everything necessary
reassembled ready for more comprehensive testing.
|
|
 |
|
I used my Tiny SA to test
the SMDU. This has a dry joint which makes its output all but
disappear but, after pressing the output cable a few times, the
output usually appears and I measured its frequency using the
EXT input to the SMDU counter. This worked perfectly down to
-35dBm across all frequencies that I tried, from the Tiny SA
Low output, from 1KHz to 350MHz. Below 30MHz I switched to the
15Hz-30MHz range.
I then checked the SMDU outputs
and found all the ranges now worked perfectly. I switched to
AM mode and the meter read perfectly, auto-switching as the A/M
depth was increased from zero to 100%.
|
|
|
Over the winter I've been
working in our conservatory because it's too cold in my workshop
and heating it would be too expensive. That meant the equipment
was in pieces in the conservatory but when Spring had got underway
I decided to move into my workshop. Once there I finished off
the mechanical part of the power supply changes and, after a
quick successful test, reassembled it and located it in my test
area below. That way I can use it and figure out if anything
needs to be done to get it back to a fully working condition..
Reassembly was fairly straightforward except the new rear heatsink
for mounting the 78H05 regulators was slightly too thick. I remade
the two clips holding it to the rear panel and this reduced the
protrusion sufficiently to fit the rear panel. One of the shock-mount
feet was broken so I removed all four because the maximum height
I could fit in the racking was 12 inches and the feet made the
height 12.5 inches.
The next
task is to continue testing my R308.
|
|
 |
|
|
|
|
|
|
