Rohde & Schwarz SMS2
372.2019.28
|
|
I bought this rather sought-after
signal generator from a chap on Ebay where he'd described it
as "faulty". It was £25 to deliver and only £45
to purchase. Sure enough, as the seller had stated, the thing
came on and apparently passed its self-test but then failed to
respond to its keys. In order to let it run through the short
test however, the on/off switch needed to be held in the ON position
(a very common fault). See my first example
here.
Although the thing is decribed
in R & S literature as "easy to fault diagnose",
I don't find this particularly true as dozens and dozens and
dozens of screws need to be removed for virtually any fault-finding.
What's puzzling is how can the
thing pass its self test and be duff?
I suppose that the designers
just didn't bother to test everything, leaving out at least one
key parameter.
As always, looking back into
the initial evidence appears to pinpoint the likely fault because
during and after the initial few minutes, I should have noticed
the fan was dead. Maybe, in the SMS2 version the fan is designed
to run only when its hot enough to be needed? But I didn't believe
this to be the case (and it wasn't) because in my first repair
of an SMS equipment the fan (which should be always on) wasn't
running properly either.
The first thing to tackle is
the bad mains switch. Luckily these switches are very similar
to a standard (old) TV mains switch and I had one in my "switches"
junk box. Fitting it was a bit tricky but it was finally working.
Because most SMS or SMS2 equipments will need attention to their
switch I've put the following pictures below to give you an idea
of the work to either fix with WD40 or to fit a new one. |
|
This is a general view
of the underside of the SMS2 (the SMS is the same). The switch
usually turns on but fails to lock in position. My first example
responded to a squirt of WD40 but this one didn't so needed replacing.
A standard old TV switch does the job. Ignore the ancillary contact
set which were used for degaussing. |
 |
|
 |
|
|
|
|
Above you can see the
rear panel can be unscrewed enough to allow the switch to be
pulled back from its operating rod which just presses onto the
switch. Note that the existing switch bracket may have different
hole spacing to the replacement and so the new one will need
drilling to fit the mounting holes. Before pressing the bracket
tabs closed check the new bracket is facing forwards otherwise
it won't fit.
It's important to note that
if you use my recommended switch the solder tags are different
to the original. The original has two normally open contacts
across the switch body whilst mine has them front-to-back. If
you connect the mains wire wrongly there'll be a loud bang when
the switch is operated!
Below the old switch on the
left and right the new switch screwed in position. Because of
the position of the switch wiring right where one might grab
the chassis to turn it over it's essential to refit the protective
plastic cover. As the original no longer clipped into place I
superglued it using baking power to make the result more robust. |
|
|
|
I said previously the
job was "a bit tricky" and by that I meant that the
throw of the two switches was different. The rod needs to have
enough movement both to comfortably operate the engage mechanism
and to fully disengage it. My solution was to wind four turns
of thin bare wire though the rod's end slot. The wire blocks
the switch from sinking too deeply into the rod end and with
just the right number of turns of wire the engage/disengage works
fine. To stop the rod springing off when turning the switch off
I applied a little superglue to the inside of the rod end before
finally pushing it into place. |
|
Having got the signal
generator to turn on reliably I turned my attention to the main
fault of dead switches after self-test. I thought the most likely
problem might be the power supply (perhaps a missing or low voltage)
so I removed lots more screws and angled the chassis so I could
test bits of it.
The SMS2 is not easy to work
on as it's heavy and has no convenient handles etc to shift it
around. The circuit boards are mainly inside a screened box so
the box top needs to be removed if one of the boards needs to
be detached. Clues to the circuitry are given on the screened
circuit board box top and bottom covers, but one needs a copy
of the servicing manual to make sense of the markings. This example
still has its extender board screwed to the screening cover which
is nice. |
|
Turning first to the PSU
power regulator transistors as I'd decided to test these for
shorts or leakiness.
There are also two power rectifier
diodes sharing the heatsink (those two large nuts hold them in
place).
Those aluminium bushes support
a pair of protective metal covers because the transistor bodies
are at varying DC potentials. |
 |
|
The mechanical designers
had clearly liaised with the electrical engineers and devised
a rather odd arrangement of fitting the regulators (3 x 2N3055
+ LM7915)
This involves the use of a (third)
pin pressed from a plate which pressure-mates with the undersurface
of each TO3 device.
A set of four 3-pin sockets
are fitted to the PSU circuit board and these allow the TO3 devices
to be pulled out so the circuit board can be detached from the
case. |
 |
|
Click
the picture to show enlarged power supply schematic
 |
|
I removed the four devices and
all seemed to be OK, however a little later I checked continuity
between the four sets of sockets and found to my surprise a dead
short between the sockets for the control pin and output pin
of the LM7915 with both grounded. The PSU circuit reveals the
control pin to be common to a set of four DIL chips and so these
connections plus the neg 15 volt rail (the LM7915 output pin)
is grounded.
The fault must be either within
the PSU board at one of the four DIL chips or, most likely, a
dead short on the neg 15 volt rail.
Of course the fan (which is
dead and should have pointed to the fault) is driven from the
neg 15 volt rail. The fan itself is not bad because it's driven
via a 39 ohm resistor and therefore is not responsible for the
short-circuit because the short has virtually zero ohms to ground
and not 39 ohms.
I guess I should now look to
see if the problem lies in the PSU board itself or one of the
many circuit boards (below) or any of the other parts within
the SMS2, Whichever is shorting the neg 15 volt supply to ground
should be identifiable by disconnecting each in turn to see which
removes the short.
Initially I attempted to remove
the first of the circuit boards but I found it to be so stiff
I left it in place. The reason being that any plastic parts such
as the coloured levering mechanisms for the circuit boards will
now be brittle with age (I think this example dates from 1984
making it 40 years old in 2024).
I'd broken many similar levers
in my Hewlett Packard signal generator so decided to check alternatives
first which turned out to be a stroke of luck.
See
later for more on this topic. |
 |
|
|
Before attempting
to extract the circuit boards I unplugged both the 24-pin (grey)
and 16-pin (K30 blue) DIL connectors from the PSU board.
Thankfully the short on the
neg 15 volt output proved to be present at one of the mating
plugs on the disconnected cables (meaning I could leave the blameless
PSU board in place). |
 |
|
|
I discovered the short
to be between Pin 8 and Pin 10 of the 16-way lead K30 which,
in the SMS2, terminates at a small metal box above the fan. The
box is labelled "SMSB2" and is the additional optional
part which provides the extended coverage from 520 to 1040MHz.
Looking in the manual I found the box contains two circuit board
assemblies.. a frequency doubler and a switch. These are coupled
by a few gold plated pins and are mounted together on the front
panel of the box. I removed the panel and separated the boards
as you can see in the pictures below. |
 |
|
|
The SMSB2 box is screwed
to a rear-mounted heatsink and secured by four screws to the
side of the chassis. To access two of these you'll need to detach
the metal plate fastened to the chassis by four countersunk screws
concealed by the end tongues of the earthing strips. The connectors
were detached with a 6mm spanner. A single grey wire also needed
cutting from an adjacent metal box mounted at the side. |
 |
|
|
 |
|
The twin circuit board
assembly was unscrewed (four tiny screws plus a coax plug) and
pulled from the box then the switch circuit board was detached
from the assembly. |
 |
|
|
Click
the picture to see the larger RF Switch schematic
 |
 |
 |
|
Once removed from
the panel and the boards separated I found the neg 15 volt rail
was connected to only the "switch" board where it's
wired to a set of op-amps and a relay driver. Also present across
the neg 15 volt rail were three small capacitors. Presumably
either one of the chips or one of the capacitors has failed.
The easiest way to discover the culprit was to apply neg 15 volts
(correctly) across the circuit and see what gets hot. It's important
to ensure the test voltage has correct polarity in case the short
disappears. |
|
|
|
I connected my bench PSU
set to a few volts to the neg 15 volt input on the Switch board
and cranked up the current limiter but 3A didn't really do very
much. The reason being a really dead short will have very little
resistance so any power will most likely be taken by tracking.
I had a couple of options. I could remove each component wired
to neg 15 volts in turn or just increase the current instead.
I chose the latter, connecting a 12 volt battery (in pretty poor
condition) instead of the current limited PSU and there was an
immediate smell of burning. None of the components was even warm,
but the very thin track from the neg 15 volt input pin had burnt
up. Mission accomplished though because the burnt track ended
at C5, a 1.5uF capacitor and after removing this I found it was
short-circuit.
There seems to be rather a lot
of these in the SMS2 so let's hope the failure was an isolated
incident! Sometimes one comes across a really stressed tantalum
connected back-to-front (eg in my Wavetek 2407).
I fitted a 22uF miniature electrolytic
(I'd misread the figure as 15uF but it's only a decoupler so
will be OK!) and reassembled everything.This was no mean feat
and involved securing many dozens of screws as well as refitting
lots of sub-miniature RF connectors. Note these take a 6mm spanner. |
 |
|
|
After completing the reassembly
I powered the SMS2 and was rewarded with a longer self-test followed
by a new display including an indicated (correct) RF output level
of -137dBm. All the front panel buttons seemed to work (I set
the frequency to the 2m band and checked the modulation figures
lit up). The next step is to check the RF output etc and fit
a couple of new plastic button caps. |
 |
|
Earlier
I mentioned brittle plastic affecting the SMS2.
When I was refitting the three
2N3055 transistors and the LM7915 voltage regulator I found it
a bit awkward. Once the insulating pad and the metal plate carrying
the pin (connected to the device case) are in place you'll need
to twist them so that their holes line up perfectly with the
three motherboard sockets before inserting the device. You should
be able to feel some resistance as the two device pins engage
with the motherboard sockets. When you're confident the devices
are plugged in you need to carefully insert the blue plastic
insulating tubes. I found that about a third of these blue tubes
broke into pieces. Once the final fixing screws were tightened
one or two cracked as the screws were tightened fully. Fortunately
I have plenty of spares and I was able to finish the job.
Once all four devices are tightened
up check continuity between each screw head (=ground) and the
device case to ensure there isn't a short to chassis. In addition,
the case of my particular LM7915 had a different shape to the
standard TO3 case and the fixing screw heads were exceedingly
close to the case. To increase the spacing I added a small washer
under each screw head. The washer diameter has to be no bigger
than the top ring of the tube otherwise this can short to the
case especially if the plastic tube degrades and squashes. |
 |
 |
|
One of the repaired brittle
plastic feet which had suffered in shipping. |
 |
|
|
 |
Here's a tip to avoid
something that might be a problem if parts are not reassembled
properly.
In order to repair the switch
board the SMSB2 box needs to be detached from the chassis.
At the rear is a large finned
heatsink which mates with the separate RF module screwed to the
back of the assembly. The heatsink is secured by two screws and
of course needs to be removed in order to detach the SMSB2 from
the chassis.
You can see on the side of the
case (top of picture) four fixing screws, two of which were hidden
by the aluminium strip sitting to one side.
The correct re-assembly procedure
should be to tighten the four screws only after fitting the heatsink
and its conductive paste. Once the heatsink is fully tightened,
and only then, should the four fixing screws be tightened otherwise
the finned heatsink may not be in proper contact with the RF
module.
At the same time as the box
is refitted to mate with its heatsink the coax plug (ST1) must
fit snugly under the lip of the chassis. |
|
|
This instruction
sheet was included |
 |
|
Next, I need to compare my older
SMS front panel board with that from the SMS2 to confirm they're
identical so I could test it. They're both marked "302-7911"
and look identical and to do this the front panel has to be removed
to detach the display board.
The procedure isn't too tricky.
The four countersunk screws are removed and the plastic front
comes off. Then the on/off switch operating rod has to be pulled
off the switch. Next the solid coax lead to the rear of the equipment
has to be freed leaving it attached to the front panel together
with the flexible coax lead which just pulls off as it uses a
two pin push-on connector. There are four small screws fastening
the circuit board to the backing plate and a few plastic locating
clips which respond to squeezing with pliers then two flat cables
to detach.
Each cable is terminated with
a 16-pin DIL header. One just pulls off but the other requires
the end circuit board to be pulled out an inch or so. Note the
orientation of the two cables otherwise either could inadvertently
be re-inserted back-to-front and also note the way the pair of
extraction levers work so the board can be refitted.
The SMS signal
generator with the bad display which has been parked in the workshop
waiting for inspiration has either a microprocessor problem or
a duff display panel. Now I
know it's the former because the SMS display panel worked perfectly
in the SMS2.
Both panels had missing push
buttons caps but as these were different to those on the SMS2
panel, I just fitted the "2" and "RCL" caps
to the SMS2 board. |
|
|
|
|
|
|
|
pending
|
|
|
|
|
