Rohde & Schwarz SMS 302.4012.26

The seller said that the on/off switch didn't latch so he held it in the ON position and something went wrong ("the fuse blew"). As you can see the signal generator looks to be in excellent condition and there shouldn't be too much trouble finding out the problem.

Soon after it arrived I decided to start investigations and realised it decidedly had NOT been designed for ease of servicing. For a start it's pretty heavy and the mechanical design doesn't offer simple handling (no handles or projecting parts). Working out how to dismantle it, to extract for example the power supply printed circuit board, was at first completey baffling, so I decided to just start removing all the screws that might allow parts to be detached. I unscrewed one end of three copper coax leads only to discover this wasn't necessary.

Above: The pair of RF modules stay screwed in place when detaching the rear panel but the flexible coax lead has to be unscrewed.

Below on the right you can see three of the four main smoothing capacitors and on their right the sluminium covers fitted over the two pairs of TO3 devices.

I replaced the connections except the one to the rear-mounted coax socket, and just carried on removing screws. Little by little the rear panel, on which is mounted the power supply, began to loosen. I gently pulled off several ribbon cables which used dual-in-line headers and ended up with just one puzzling problem.. the mains on/off switch operating shaft, but after donning my magnifying goggles I spotted the end of the shaft was split and could be eased off the switch.

At this point the entire rear panel with the PSU and a heavy metal-cased toroidal transformer lifted off.

I counted the screws and found I'd had to remove no less than 73 plus nearly as many washers to get this far.

I listed the operations here in more detail

Above: In the rear right corner is the faulty mains switch which I believe is a stock fault and easily fixed with a spot of WD40 and lots of waggling. Initially it was a real puzzle to gain access to the switch as the operating rod runs above a rectangular motherboard which looks impossible to extract without major dismantling. However, once the rear panel was partially freed the switch could be moved to the rear by an inch or so allowing access to detach the operating rod. But see below....

Upper left is the mains switch held by two M3 screws. As WD40 freed up the mechanism permitting it to latch and unlatch I didn't need to swap it. The switch design is similar to those used in CRT TV sets from the 70s and 80s and, as I still have a selection of these, swapping it may have been straightforward.

Above is the power supply mounted on the rear panel and below the circuit diagram which I had to stitch together from five pages.

Click on this mini-image to see it full size as a PDF. I'll be replacing this soon as I now have a much clearer copy.

The PSU uses a mains transformer with four secondary windings to provide +5V, -15V, +15V, +20V and +28V and all use similar components depending on the current draw. 2N3055 power transistors are used for +5V, +15V and +20V whilst uA723C, uA741C, LM320 chips are used for regulation. High current circuits use discrete silicon diodes whilst the low current 28V supply uses a full wave bridge.

I reached the stage where I wanted to remove the PSU board from the rear panel so that I could test the capacitors and found to my surprise that the German designers appeared to have lost their presence of mind. At first sight the board cannot be removed without unsoldering the three 2N3055 tansistors and the 4th device, a uA7915. Unsoldering is not easy because other components are obstructing access. An added complication are the insulating washers especially under the pair of stud diodes above the TO3 devices above.

Here's one example of the seemingly odd TO-3 pins. Ordinarily one would expect the case to be connected via one or both the outer holes in the case, but here we have three grouped solder joints.

The answer is not specially produced transistors carrying an extra pin but the addition of a plate with a pin pressed from the material (hence that slot which isn't a shadow).

The transistors are bolted to the rear heatsink via insulating bushes and the three pins mate with special sockets soldered to the board.

In the meantime check out the processing system below. This is essentially very simple, translating front panel push button presses into electrical control signals. To do this a program within the microprocessor scans the state of all the front panel controls and decides what to do. When things get too complicated, for example driving the front panel display to echo the RF output parameters, a second ROM carrying preset command signals can be brought into play.

Below you can see code numbers printed on the chips carrying ROM. During development lots of attempts would have been made to get everything working perfectly and the printed codes reflect the issue date of the chips used in the equipment. Before microprocessors this function was done by things like custom-designed yaxley switches and loads of wire. Changes would have involved wiremen and the manufacture of modified switches so the advent of microprocessors would have been a lot easier. A few years later, reprogrammable ROMs became available which would have saved scrapping bad chips carrying faulty code.

Here are the two processor boards, as distinct from the RF boards contained in a sealed enclosure. Above is the microprocessor board with an Intel P8049 which uses 8-bit architecture and embodies a built-in 2KByte ROM, 128Byte RAM and 27 I/O ports. There's also an HEF4738VP bus controller and a P8355 16Kbit ROM with I/O. I spotted a couple of SN54LS244 chips which are military spec amongst a group of common or garden 74LS and CMOS chips. Maybe they had sourcing problems?

Below, the supporting board which has a Signetics chip carrying two codes. "8010" probably the manufacturing date and "MP8243N" the chip code. I thought it might be a RAM device but I can't find a spec on the Internet. luckily I have a period data book from Signetics and their 8243 is described as an "8-bit Position Scaler". No doubt R & S discussed this chip with the Signetics rep who of course lived just down the road in Munich. In fact.. who knows.. the local rep might have arranged for this strange chip to be designed for R & S? Let's hope the chip doesn't ever fail as sourcing a new one might be tricky! It's function is to rearrange 8 data bits according to a 3 bit select code.

I mentioned that it was cumbersome moving parts of the equipment around because of its weight and the lack of handles etc. Most equipments can be stood face down with impunity but not this one. Whilst detaching the PSU board there was a loud crack and the spindle of the on/off switch flew across the room. I considered gluing it back but as the switch had already proved troublesome I raided my junk box and found a suitable "universal" replacement.. shown below. As it was designed to fit numerous different, now antiquated, TV sets it can be mounted in various positions and it fitted here perfectly. I'll refit the plastic safety cover later.

After the battle removing the PSU board so I could check the major parts I found nothing wrong with it. Maybe something's amiss on one of the plug-in boards, otherwise it may just have been a fluke that the fuse blew. The fuse appears to be rated at either 1.25A or only 125mA. I need to check this but it's possible that arcing took place at the faulty on/off switch.

Below is the result of testing. All the capacitors measured with an ESR of zero ohms except the 47uF which read 0.4 ohms. As I had to remove this because in-circuit tests were inconclusive, I fitted a new 47uF x 63v electrolytic. All the diodes and transistors were OK so I now need to reassemble everything. See details of this.

I decided the best plan is to reassemble the rear panel, refit and see if the fuse blows. It took a fair time and involved a long perusal of the manual's block diagram to identify where the last cable plugged in. Eventually I plugged it in and switched on.

The fuse didn't blow but the front panel displays were all garbled.

After a few minutes I sensed a burning smell and identified an 8.2 ohm resistor (R13) as the culprit. This resistor feeds the minus 15V supply to the fan, across which is a 470uF capacitor (C11). The circuit below has a 39 ohm resistor but that is for the SMS2 version. Maybe the original fan was unreliable?

I measured R13 and found it was 10 ohms (burning had increased its resistance a little) with -15 volts and -4.7 volts at the ends. This represents a resistor dissipation of a little over 10 watts. As the fan is stationary I guess that either it's failed, or there's a slim chance the 470uF capacitor has failed (probably not though as it previously tested at 498uF and zero ESR). I'll trace the fan wiring and see what this "voltage converter" is. It turned out to be a tiny sealed box but then I noticed there were 6 wires from the box, two of which (red/blue) went off one way whilst the other 4 disappeared behind the PSU board and reappeared half way down one side. As I pulled on them I noticed two of the wires were trapped between a metal lug on the pcb and the aluminium panel, screwed tightly and cutting the wire insulation. After freeing the wires the fan voltage rose to 10 volts and it spun nicely, but the display was still garbled with few recognisable characters.

I decided to do a quick check of all the supply voltages. These are wired to a set of 10 pins (labelled MP1-10 in the user guide).

Interestingly, opposite one of the PSU circuit diagrams, in European script, is pencilled a set of voltages.

I switched on the equipment just long enough to make the measurements. The results are tabled below and all are correct.

PIN	MP1	MP2	MP3	MP4	MP5	MP6	MP7	MP8	MP9	MP10
VOLTS	8.2	5.39	20.5	-15	15.2	25.5	25.5	20.1	43.6	28

Time for some serious fault-finding. Looking downwards on the front panel I could see at least one very large i/c so maybe this panel carries the display decoding? Either that or one or both the processing boards? First a look at the display panel schematic.

Thankfully all the i/c's look standard off-the-shelf types with a few uncommon ones. basically a large P8279-5 driven from the microprocessor bus provides the drive to a large number of logic i/c's. The first step will be to check that the cable carrying the bus is the right way round as R & S didn't use polarised connectors, relying on either 16 or 24 pin DIL plugs. At first sight the entire display appears to rely on strobing dictated by an NE555 and obviously all the clock signals for the logic circuits have to be perfect. My last-but-one HP signal generator had a duff logic chip (74LS96N) that resulted in a display failure. In this R & S sig gen all the i/c's are in sockets and I have an i/c tester that handles 74 and CD 4000 series.

Here's a time exposure of the front panel powered up (which is exactly what one sees). I'd found a short duration picture showed only a few lit segments because the display is being continuously refreshed, although the single LEDs seem to be constant. I had some trouble identifying that the various flat cables were in their correct orientation.. for example the one connecting the microprocessor board to the front panel. The pins of the 16-pin DIL sockets are labelled on the pictures of the pcb track drawings so I believe the cables are right. The cable carries the 8-wire computer bus (pins 1 to 8) plus a number of control signals (pins 9-16). The microprocessor is presumably looking for commands resulting from front panel button presses, then working out what to send on the bus to illuminate the figures on the display. Because any button presses (apart from the POWER button) do pecisely nothing I guess there's a fault.

At first sight the pcb on the rear of the front panel looks mighty difficult to get at. Thinking logically (which may be a bad idea) pressing RF-OFF should change the state of its red LED. Also the AM, FM and EXT LEDs shouldn't all be on at the same time! Below a quick snap captures this, some of which does actually line up with the time delay results.

In fact the front panel pcb wasn't too difficult to extract, but prior to that I tested all the 74 and CD logic chips on the microprocessor and modulation boards (side by side on Motherboard 2.. the smaller backplane of the pair). All tested OK but I did find a short-circuit J111A FET on the modulation pcb. This wasn't connected with the display fault however.

Sadly the front panel chips were all soldered in place so testing is on hold. I decided to trace a simple anomaly viz. the simultaneous illumination of the AM and FM LEDs. These appear to be driven from B1 microprocessor which from Pins 27 and 28 drives the modulation indicator LEDs via logic chips plus driving the circuitry on the modulation pcb.

The process being.. (1) press AM, (2) the micro senses the switch activation, (3) the micro switches on the AM LED and at the same time turns on A/M mode.

As both AM and FM LEDs are both on clearly there's a problem. Below, the front panel pcb. To extract it the four screws securing the front panel assembly to the chassis are removed, the long solid copper coax cable is detached from the attenuator at the rear, the EXT modulation cable unplugged, the two flat cables (DIL16 connectors) unplugged and the dodgy bit.. the ON/OFF switch extension pulled off the switch. Once this is done the pcb can be pulled away after a set of M3 screws is removed.

I hadn't noticed until this point that a few push button covers are missing so IF I can get things working that will be on a list of jobs.

Communication between the various processing elements is via the CPU bus together with clock and control signals so it's possible that either the clock is bad, a control signal is bad or the CPU bus is being held up (to +5v) or down (to ground), either wholly or in part. The principle of the bus is based on tristate logic and it is relatively simple for a fault to jam the bus. Either a bad tristate chip or a bad gating signal allowing the bus to jam. This could be a simple a problem as a grounded or stuck enable signal.

Back in the 1970's when I worked in the defence industry we were encouraged to take out patents in Plessey's name. They said this was necessary to balance out patent possession within the UK's largest defence companies so a tit for tat arrangement would be possible. At the time I was managing a small team of engineers designing a multiplexer for air traffic control and we were using new chips almost straight from their designers' fag-packet sketches. One such chip used a tri-state bus and, as the chips were incredibly sensitive to static damage, we found shorted tri-state bus bits to be a frequent problem. I worked out a very simple solution which was to insert a resistor of only just enough value to help pinpoint a bus short without affecting noise immunity. This was duly patented and I received the princely sum of £30. Actually not a bad amount back then.

Well, I looked at the tristate bus and found no anomalies... at least on the individual unpowered pcbs. Maybe it's time to call in my oscilloscope and see what's going on. I'm now banking on a missing clock. The fact that all the common 74 series i/cs tested OK so far makes me think all the i/c's are probably OK. The sig gen has been dropped during its life. Once in transit because the four plastic feet were broken off (that must have been something like a 4 foot drop on solid ground) and some time ago another bang resulting in a bent aluminium extrusion. Maybe the HC18U crystal was damaged... I just checked my list of crystals and I have a spare 6MHz one, but alas it wasn't present in the box so instead I tried one marked 6.125MHz. This failed to work so I decided to test the old crystal. This is easier than you might imagine.

Connect a signal generator to an oscilloscope using a pair of leads fitted with croc clips. Set the sig gen to say 100KHz under the marked crystal frequency (in this case 5.9MHz). Set the signal to the max level (say +10dBm). Open the croc clips and insert the crystal then slowly tune upwards and at the marked crystal frequency the displayed signal should dramatically rise in value before dropping back again. I tried this and the resonance was precisely 6MHz so the crystal isn't damaged after all.

Extracting, opposite, from the spec for the P8049 (click to see it), is the crystal circuitry. The pcb actually has a pair of 22pF capacitors equivalent to a load across the crystal of 11pF. The connecting leads to the crystal are parallel and about 3 inches long but obviously (used to) work.

The scope did show a repetitive signal at Pin 2 or Pin 3 but low in amplitude and with an indication of 600KHz. Maybe this is to be expected, bearing in mind I'm using an oscilloscope probe which might upset the crystal, so the next step is to monitor the various output pins to see if these are clocked?

I wondered if the 8049 might be faulty so searched through my collection of 40-pin chips for another or even something vaguely compatible. I found an 8042 whose pinning is just about the same. Obviously any ROM within the chip will be different but at least I can see if the front panel garbled display is the same or similar.

Swapping it had no effect so either both the 8049 and 8042 are both bad or something else is messing up the display. I'm not really happy about the 6MHz oscillator. Either the crystal or padding capacitors are not up to scratch or my attempt at measuring Pins 2 and 3 are stopping oscillation. According to the picture above the value of C2 (equals C1 and C3 in series) should be less than 8pF whereas the R & S designers chose 11pF (two 22pF in series). Clutching at straws could either of these 22pF be high in value? At least I can test the microprocessor on the bench.

Both of the 22pF capacitors were about their correct value and removing them had no effect, however during my experiments I noticed that when I inadvertently shorted Pins 3 and 4 with my scope probe the 6MHz crystal suddenly came to life with about a volt of signal. Pin 4 is the PC reset signal which is ground for reset and normally rests at 5v. Sure enough feeding Pin 3 from 5v via a 470 ohm resistor triggered the oscillator, however there was no activity at any other of the 40 pins and the display garbling was unchanged.

Could the microprocessor be damaged? How could this be? Well, at the rear is a mains voltage selector and I wonder what could happen if the 115 volt setting had been selected?

Well, I had a couple of other 40 pin chips, admittedly an 8042 and an M5L8042 which have the same pinning but obviously would not have suitable program code in their ROM but, I'd expect either to have some life by way of active output pins, but alas the display garbling was unchanged, so could there be a bad component, damaged or shorted tracks that's stopping the processor pcb from working?

Bearing in mind I've powered the processor pcb on the bench and it seemed lifeless so the problem, I should say is in the microprocessor pcb itself.

Other projects and lift repairs (before I retired a few months back) put the R & S box on the back burner, but recently I watched on YouTube a chap in the USA attempting to tackle his duff R & S SMS2 and during the process he'd noticed the pair of PSU power output leads to the motherboard and front panel were not polarised and could be fitted either way round. During my fault diagnosis I'm pretty sure I'd inadvertently reversed one or both of these power leads. In his case he'd noted that one of the power regulation transistors was running extremely hot and this no doubt was because a lead was reversed.

Any R & S electronics design engineer worth his salt would have recognised this shortcoming and planned for it. So, what about this design? What happens if a lead is reversed? I decided to tackle this question and see exactly what happens. One connection uses a 24-pin header plugging into a 24-pin i/c DIL socket and the other a 16-pin.

Below is a table showing normal and reverersed connections for the 24-pin header. You can see that there isn't really a problem because in no case is there an excessive voltage sitting on the equipments power rails. There are several short-circuits but as the designers built a crowbar into each regulated power supply, apart from maybe a power transistor running a bit hot the circuitry is safe from damage.

Below is a second table showing the effect of reversing the 16-pin header. Again nothing untoward will happen. Although it first appears something bad will happen, for example connecting PSU pin 9 to the 5 volt rail, PSU Pin 10 carrying ground will present a short-circuit to this at Pin 2 which is connected to Pin 1.

The next stage might be to use my second R & S SMS2 (albeit also faulty) to compare various circuit signals and voltages. I carried out an analysis of the SMS and SMS2 equipments and it seems that nearly all their circuit boards are identical. The important difference is the microprocessor board which uses different programs in the chips.

I fitted the suspect SMS display panel in the SMS2 and it worked perfectly so either the microprocessor board is faulty or there's possibly something corrupting the tri-state bus.

Having had success with the second purchase (the SMS2) I decided to re-check the SMS example. What was worrying however was a discussion on the Net about an identical SMS equipment with the same fault as mine viz. a supposedly dead microprocessor. In that case the evidence suggested a previous owner to the writer had diagnosed a bad P8049 and had fitted a new device. Both the P8049 and the P8355 include ROM and unless one has a copy of the P8049 code together with a programmer a new P8049 will fail. In my case both the P8049 and P8355 have zero activity at their data buses and this would be the case if the chips were both bad. One slim possibility is the local oscillator has failed. This circuit is actually located at the empty 40-pin holder but I believe it is used for both the P8049 and the neighbouring P8355. A possible option is to remove the P8049 and P8355 and fit a replacement chip in either position. This would perhaps bring either or both chips to life if the oscillator starts up. In other words the task is to prove not the good or bad chips but the oscillator circuit. Another slim chance is failure of the reset/start circuit.

Fortuitously, when I suspected a bad display board in my SMS2 I tested the one from the SMS and it worked OK so at least I know the P8279 on the display board, together with the bus connections on it are in good order.

The microprogram board is Y11 and, in the SMS, is 302.7111. B1 is the P8049 and B2 is the P8355. There's also a third 40-pin chip which is B17 a HEF4738 used for interfacing with the IEC bus.

B1, the P8049 uses a 6MHz crystal or a clock signal from B6, a 74LS74 flip flop. B6 also feeds B17 and is used as an alternative to the crystal. Testing of V11 is carried out by changing the jumper positions, but unless you have a specific piece of test gear you can't do this.

Swap link from I to V

Swap links from the crystal to B6 (II & III)

Plug Start/Stop and clock into HP 5004A analyser

The P8355 doesn't use its clock input so must be run asychronously by commands from the P8049.

According to the manual the P8049 is driven by what are termed "interrupts". This is an Industry term which means that the microprocessor is continuously scanning for signals or commands which determine its next procedure. For example, if a front panel key is pressed that specific key will trigger the program to execute a particular action. Pressing a number key, say "1" would result in that value being stored. The next key would determine the action taken, so if the MHz key is pressed, the program would output "1MHz" to the display. This process would ensure the figures make logical sense.

It would make sense that the P8355 would have a program which perhaps activated the various functions within the equipment such as modulation type and attenuator settings. It could do this by monitoring the status of the address bus and producing a logical output to control the hardware on the various circuit boards.

I have scores of old computer chips but no example of the P8049 but I do have a P8742, which, if you examine the pinning on the left, you can see the clock, reset and power pins are the same. Naturally the replacement will not have the Rohde & Schwarz program, but at least I can check to see if it supports the crystal at pins 2 and 3.

Sure enough I can see a nice 6MHz sinewave at the P8742 but nothing for the P8049 so I guess that the former is faulty.

Both chips run at around the same temperature so is the P8049 beyond help?

With the P8742 in place the random front panel display clears but obviously whatever code is in the P8742 will not be valid.

I noticed on an Internet forum someone suggested putting one's bad processor in the freezer to cool it down, but that didn't work, neither did increasing the 5-volt supply although the highest it would go was only around 5.7 volts. I did however find by fiddling with a probe that just touching some pins made a relay chatter and this blanked some of the display contents to the extent that with several touches all would be blanked. I also found that by shorting Pins 3 and 4 (Clock & Reset) that the oscillator started working. The reset pin is sitting at 5-volts (with a ground pulse for reset) so maybe adding a resistor from pin 3 to 5 volts might get the oscillator running? Looking back at my previous experiments I see I'd already tried hooking up Pin 3 to 5-volts via a 470 ohm resistor to no avail, but I'll try again.

Sure enough, with Pin 3 tied to the 5-volt supply via a 470 ohm resistor the crystal oscillated at its correct value of 6MHz but still no bus activity. Removing the other two 40-pin chips failed to fix the problem of getting the microprocessor to work, and I noticed that touching Pin 11 with the scope probe caused a relay in the circuitry to drop in and out. The fact that the increase in voltage at Pin 3 got the crystal to run I'm wondering if an internal voltage regulator has failed reducing an on-board supply to drop below a critical level? My final test might be to add a 10uF capacitor across various pins to ground as this might increase the on-board supply sufficiently to get the program to run. This resulted in a lot of relay chattering but no display change.

Below is a table of voltage readings measured with a DC meter at the P8049 pins (click to see the spec)

Marked with a cross are possible bad readings because their active rest state looks wrong. You'd think that if the processor is looking for a ground signal the rest state should be at Vcc level such as Reset? For example Pin 5 is set at ground for "single step" operation. TTL (guaranteed level) has a logic "1" defined as a voltage greater than 2.0 volts and a logic "0" as equal or less than 0.8 volts.

The "xx" indicates a problem, because at the "ALE" pin, should be a clock signal reflecting each microprogram cycle. This can be used to protect a device from damage, for example, if it's not present prior to the internal ROM being programmed. Typically a DC measurement at a pin carrying a series of pulses would normally show up as a DC voltage and zero volts (as seen at pin 11 below) would indicate that no pulses are present. Some "DB" pins have a sawtooth waveform present. To confuse things a tr-state output might have an indeterminate voltage (ie. not a true TTL level). Such a pin is only valid during a program action and at this point the level is standard TTL.

You'll note that very few pins have a proper TTL level present so I need to check for the presence of pulses at these pins.

PIN	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40
NAME	TO	X1	X2	RE	SS	INT	EA	RD	PS	WR	ALE	DB0	DB1	DB2	DB3	DB4	DB5	DB6	DB7	VSS	P20	P21	P22	P23	PR	VDD	P10	P11	P12	P13	P14	P15	P16	P17	P24	P25	P26	P27	T1	VCC
VOLTS	5.5	2.2	4.2	5.5	1.1	5.5	0	0	0	0.5	0	0.8	0.6	0.7	0.8	1.1	0.8	0.5	0.3	0	1	1	1	1	1.2	5.5	3	3	1	0	0	1.2	5.5	5.5	5.5	5.5	5.5	5.5	0	5.5
NOTES					x			x	x	x	xx
P8742	5.4	1.8	1.4	5.4	2.5	5.4	0	1.3	0	0.2	1.0	0.03	0.03	0.03	0.03	0.03	0.03	0.03	0	5.4	5.4	5.4	5.4	5.4	4.5	5.4	0.03	0.03	0.04	0.02	0.02	0.02	0.09	0.09	0.09	0.09	0.09	0.09	0.09	5.4

What next? Is it possible that the microprocessor has failed completely? To help confirm this because of course another chip may be responsible as, on a bus system, any device might be responsible for strange voltages. To come to a conclusion I found in my junk box another relatively compatible device (a P8742)and swapped this for the P8049. The row in the table above lists the voltages at the 40 device pins. Of course the P8742 will have different ROM contents which may confuse things because I'm measuring DC voltage rather than checking for pulses. As there are no pulses present at all on the P8049 pins then the pin voltages should be consistent with valid TTL levels and as they're not the device is definitely faulty even when forced to oscillate by fudging the voltages at pins 2 and 3. All that remains is to pack the SMS away.

The next project will be my CJD receiver.

pending

Rohde & Schwarz SMS 302.4012.26

Return to Reception