Testing the SMS2

See the repair

 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.

 

 Return to Reception