Testing the SMS2
|
Having fixed the power
supply short circuit and checked that the various push buttons
change the numbers on the display I need to see how the signal
generator works. For this exercise I used my Rigol spectrum analyser.
Very odd results! |
|
|
Output=1GHz level 0dBm |
Output=500MHz level 0dBm |
|
|
Output=250MHz level 0dBm |
Output=100MHz level 0dBm |
|
|
Output=10MHz level 0dBm |
Output=100KHz level 0dBm |
|
These results look very
strange and I'm pretty sure those relatively clean shapes at
100MHz and above would resolve with better definition into those
like the 10MHz and 100KHz outputs.
My initial guess is a phase
lock loop failure.
Some pretty major fault finding
is on the cards. Fortunately the equipment came with a board
extender and I have a fair copy of the 270 page manual. |
Output=1040MHz level 0dBm,
full spectrum display |
|
|
Most early equipments
designed in the 1970s that use discrete components in their phase
locked loop (PLL) circuits are quite complicated. My other current
project is a Plessey receiver of that vintage and currently is
suffering from a PLL problem, but at least one saving grace is
the absence of a specially designed chip whose failure might
bring about a grinding halt to any repairs.
I've decided to resort to looking
at the user manual to start fault investigations and below is
the first relevant page (thankfully in English). |
|
|
The colour coding relates
to the plastic levers used to extract or insert the printed circuit
boards.
My initial guess for the fault
is maybe a second bad tantalum capacitor.
From the chart it looks like
Y4/5/6/7 could be the source of the problem because the fundamental
output frequencies are present, at least in the VHF range.
I cheated at this point and
swapped four boards from my SMS, but with no real change I measured
the power supply voltages which I should have done initially.
This is covered in 5.3.1 of the manual. Oddly this is after
the fault-finding chart on the right which is 5.2.1. surely the
PSU should have been the first point of call? |
|
|
|
The
easiest point to make measurements was at the 16-pin DIL plug
where I'd originally discovered the neg 15 volt short circuit
inside the SMSB2 box.
PIN |
1 |
2 |
3 |
4 |
5 |
6 |
7 |
8 |
9 |
10 |
11 |
12 |
13 |
14 |
15 |
16 |
VOLTAGE |
5.27 |
5.27 |
15 |
25 |
26.3 |
0 |
0 |
-8.3 |
28 |
0 |
0 |
26.3 |
25 |
15 |
5.27 |
5.27 |
CORRECT |
5.10 |
5.10 |
15 |
20 |
25.5 |
|
|
-15.0 |
28 |
|
|
25.5 |
20 |
15 |
5.10 |
5.10 |
TOLERANCE +/- |
0.01 |
0.01 |
0.05 |
0.05 |
- |
- |
- |
0.05 |
0.01 |
- |
- |
- |
0.05 |
0.05 |
0.01 |
0.01 |
ADJUSTER |
Y11-R2 |
Y11-R2 |
Y11-R25 |
Y11-R36 |
- |
- |
- |
none |
Y11-R47 |
- |
- |
- |
Y11-R36 |
Y11-R25 |
Y11-R2 |
Y11-R2 |
|
|
Well, there's the reason
for the strange waveforms!
The neg 15 volts in miles too
low, and funnily enough I'd wondered why the fan (fed from neg
15 volts) was so quiet. That's twice I'd missed the same clue.
The easiest way to check was to add a 100uF capacitor to the
neg 15 volt rail, however the voltage remained at neg 8.3 volts
so I turned on my bench PSU, set it to 15 volts with 100mA current
limit and connected it to the neg 15 rail whilst monitoring what
was supposed to be a 100MHz signal.
The voltage initially fell so
I progressively turned up the current limit and at 500mA with
draw of 475mA the voltage stabilised at neg 15 volts and suddenly
the 100MHz signal went from over 50MHz wide to a narrow spike.
The SMS2 was working....
The manual tells me that the
current draw from the neg 15 volt regulator should be 600mA so
is the problem due to a bad PSU component or another bad tantalum
capacitor (drawing excessive current) on one of the circuit boards?
Alternatively those two control circuits (indicated on the right)
might have a bearing on the matter, or even a bad LM9815 which
is shown here as the equivalent LM320K-15. GL5/GL7 could also
be suspects.
But... NO and yes.. |
|
|
The three 2N3055 power
transistors and the neg 15 volt regulator are mounted on the
rear panel and my first idea was to check the latter as I had
a spare fitted to the earlier SMS (below). It was then I noticed
it was diagonally opposite to the one on my SMS2. This would
explain the wrong voltages on both the neg 15 volt (=8.3V) and
the 20 volt (=25V) rails.
I corrected the mistake. I turned
on the equipment and a number of odd things happened. It started
up correctly but no signs of the RF output. I then noticed a
smell of burning from somewhere within the power supply although
nothing was especially warm so I started measuring voltages.
The neg 15 supply was still sitting at about neg 8 volts and
the stabilised 20 volt supply was now also low at around 8 volts
instead of the previous 25 volts. I detached the only connector
that was at hand, being the 16-pin DIL plug above the SMSB2 box
and I'm pretty sure the 20 volt supply returned to normal. I
checked the resistances around the SMSB2 connector and convinced
myself the RF output module had failed short-circuit. Before
detaching this for checking though, I looked at my first SMS
example (which fortuiously also has the SMSB2 option fitted)
and found precisely the same resistance readings so decided (with
relief) that the RF output module was blameless and again looked
at the 20 volt supply. Below is a schematic of this and it's
worth pointing out that this is basically the same circuit used
for the 5-volt and the 15-volt supplies both of which were working
properly. First though, I tackled the neg 15 volt supply which
now had the correct LM7915 in place of the wrong 2N3055. |
|
|
This is what I found by
making checks with my multimeter.
The burning smell had not originated
from the missing track to pin 1 of the LM7915 because the open
circuit must have happened when the SMS2 had been powered up
with the 2N3055 and LM7915 reversed (some time ago).
This swap had somehow resulted
in an almost correct 20 volt supply of 25 volts (via the incorrect
LM7915) and a low neg 15 supply from the wrong 2N3055.
Incidentally PSU fault-finding
is very difficult because access to the circuit board is very
restricted. Removing it is not easy
because the mains transformer is bolted to the chassis and its
wires soldered to the board. Some screws can be removed however
allowing the rear panel to be pulled outwards as shown below.
After a little investigative
work I found an open circuit "pseudo" ground connection
to the LM7915 corrected with that red wire you can see below.
The neg 15 supply is now working normally. Note I'm using the
term "pseudo" to refer to an alternative ground connection
used for overcurrent detection and crowbar action.
See more on this later. |
|
|
|
Above is a basic voltage regulator
circuit using the uA723C |
|
Although the SMS2 had
actually been working at the point I'd used an external supply
to correct the neg 15 volt rail, I had noticed that the 20 volt
supply had been a bit high and, thinking back, this may have
been due to some damage that had occured with the 2N3055 and
LM7915 reversed. This damage had somehow resulted in an almost
correct 20 volt supply from the (wrong) LM7915 but was now reducing
the rail to around 7 to 8 volts from the (correct) 2N3055.
The regulator circuits are slightly
complicated because the designers used an overcurrent technique
which comes into play if one rail is say shorted to ground. This
involves the use of a small 2.2 ohm resistor connecting the crowbar
control wire to ground. An excessive current draw through this
resistor will feed back to the uA723C devices (via a uA741C)
which control the 2N3055 series pass power transistors. The uA723C
is a voltage stabiliser set to something slightly higher than
the required output from the corresponding 2N3055. For example
this will be say 20.6 volts for the stabilsed 20 volt rail. Also
incorporated in each of these three circuits (viz. 5 volts, 15
volts and 20 volts) is a potentiometer for accurately setting
the voltage. These are noted in the table
up this page.
Below is the schematic for the
20 volt regulator. |
|
|
I started looking for
the problem with the PSU board screwed to the rear panel and
found the voltage at pin 2 of the uA723C was zero. As these chips
are used for the 5 volt and 15 volt regulators and fitted in
sockets I switched the latter and that for the 20 volt regulator
but found only a slight change. The uA723 output voltage was
slightly higher but only about 8.5 volts and this was reflected
in the low 20 volt supply. Although the uA723Cs are in sockets
the uA741 op amps are soldered in. It's now looking like B7,
the uA741, is bad as I found R29 and GL14 seem to be OK, as were
R30, R32 and R33. Before tackling removal of the circuit board
I'll check R37 and R35 and their connections to the pseudo ground
supply and the 20 volt rail. The voltage at R36 is around 2.6
volts, which using the 8.5 volt output makes this about right.
It's not possible to access Pins 2 and 3 of B7, but my guess
is that B7 has failed. |
After
removing the three 2N3055 transistors and the LM7915 followed
by the heatsink then removing the board securing screws and the
fan (for easier access) I found to my surprise the reason for
the final power supply fault.
The power devices with their
insulating washers are secured with long M3 screws. These have
12 to 13mm thread lengths (probably half inch) but two had 15mm
threads and the result is shown below.
It seems the problem wasn't
the result of the swapped power devices but the 15mm fixing screws.
I now need to figure out exactly
what happened. |
|
|
Looking at the back of
the circuit board there are several tracks that have clearly
been carrying excessive current, but only one has melted open
circuit.. that being the one I'd diagnosed previously which was
responsible for the loss of the neg 15 volt output.
Taking the cut connection at
B8 first. This has disconnected R41and most likely grounded Pin
5 of the uA731 and upset the reference operation of the regulator.
The output voltage was a bit high at 25 volts but this was not
too serious.
Looking at B6. This of course
is the 20 volt regulator and Pin 13 of the uA731 will have been
short-circuited to ground. Although this pin is used for "frequency
compensation" grounding it will force the current limiting
circuit to turn on. I'm hoping that this will explain the zero
voltage on Pin 2 (also associated with current limiting). Pin
12 has also been intermittently grounded ( I spotted the results
below before reassembly).
If I'm lucky all I need to do
is make good the open circuit track in the 15 volt supply (and
remove the red wire that I'd added). Make good the open circuit
at R41 and, after final inspection, to ensure I've not missed
anything, reassemble and test.
Fortunately, I checked the components
side of the board before reassembly and found the burnt resistor
below. This is a zener load resistor providing via the two zener
diodes a working voltage for the uA723 regulator. This voltage
(about 36 volts) is at Pin 12 which was also grounded by the
long screw. The resistor was reading about 280 ohms and had not
failed completely. From the evidence I think the short on pin
12 had been intermittent. |
|
|
After fixing the various
faults on the PSU board I started to fit it back on the rear
panel and realised an error I'd made. Several of the blue plastic
TO3 insulators had broken from being brittle and I'd replaced
these with some similar black ones but I hadn't realised that
their precise physical size mattered. In fact the flat tops were
thinner than the originals. Not that much, and with the 12mm
screws they'd been OK, but with the slightly longer screws with
15mm thread length the tolerances had been exceeded and the tips
of the screws had just hit the rear of the circuit board.
Everything went back together,
although plugging in the 24-pin connector was difficult due to
the restricted access and refitting the power devices was frustrating
trying to get these to plug in successfully. I checked them for
chassis shorts and all was well, so next I'll turn the thing
on and see if it works properly.
It works! Below a signal at
100MHz
Next I adjusted all the voltages
to match those in the manual. All adjusted nicely as shown in
the table below. |
|
PIN at front Test points |
2 |
4 |
5 |
8 |
10 |
VOLTAGE |
5.10 |
-15.1 |
15 |
20 |
28 |
TOLERANCE +/- volts |
0.01 |
|
0.05 |
0.05 |
0.01 |
ADJUSTER |
Y11-R2 |
none |
Y11-R25 |
Y11-R25 |
Y11-R47 |
|
Now a 600MHz signal at -50dBm
extending into the SMSB2 option.
The next task will be to repeat
the tests shown at the top of this page.
|
Below are pictures of
the SMS2 outputs from 100KHz to 1040MHz
The output was set at CW -20dBm
apart from the last which was set at 0dBm to check for harmonics
Rigol settings are as indicated.
Note RBW is 100Hz (span 100KHz) apart from the two full span
pictures at 1MHz.
A 2m RF BNC to BNC cable was
used fitted with suitable adaptors
Apart from accurately setting
the power supply voltages none of the internal pots etc had been
touched. |
|
|
SMS 2 set to 100KHz -20dBm |
|
SMS 2 set to 1MHz -20dBm |
|
SMS 2 set to 100MHz -20dBm |
|
SMS 2 set to 500MHz -20dBm |
|
SMS 2 set to 1040MHz -20dBm |
|
SMS 2 set to 100MHz -20dBm
Full span 0-1500MHz
150Mz per horiz division
No harmonics are obvious (<40dB) |
|
SMS 2 set to 100MHz 0dBm
Full span 0-1500MHz
150Mz per horiz division
Second harmonic just visible |
|
Here it is, complete with all
its front panel buttons. |
|
|