Rohde & Schwarz SML Series Signal Generators - SML01 SML02 SML03 SMV03

[tszaboo]

4. This on is strange, very slow response to button presses. Often times I have to press a button for 2 seconds for it to register. The encoder is fast though, so not sure what's going on.

This is a known problem with older SML. From what I hear it is the contact pill in the elastomeric keypad that degrades. I know that some people have experimented with conductive silver, but that is apparently difficult to fix. Luckily, mine doesn't do that.

You can buy 'stickers' from Buttonworx which have a conductive pattern printed on them. They go between the original rubber key mat and the contact area. They work like a charm.

Interesting. I've google searched for this quickly and it also gave me an alibaba link as well.
Searching on AliExpress, " Diameter 3/4/5/6 MM 4 Sizes Black Conductive Rubber Pads Keypad Repair Kit" is an item for ~3 EUR.

This is interesting, when you mention the harmonics. They should not appear that high relative to the carrier. This is almost the exact same issue from my B3 pulse modulation option, where the switch MMIC died and one of the output amplifiers was also broken according to test points.

Not entirely sure what components were all broken on the B3, as I bypassed it and called it a day. I am hoping to swap the switches out at some point, though.

I also though so. I have low frequency output and modulation input, which might be an option (not shown on the picture). Anyway, I'll crack it open and see what's inside. And I really apricate the help.

Here is some brain dump from me:

From what I can see in the manual, the difference between the SML02 and SML03 is the number of filters after the step recovery diode.

Table 7-1 Frequency schedule of Output Unit 2 GHz / 3 GHz
n Input frequency Output frequency
2 605.25 to 909 MHz 1210.5 to 1818 MHz
3 606 to 885 MHz 1818 to 2655 MHz
4 663.75 to 825 MHz 2655 to 3300 MHz
The selection of the desired spectral line is made by means of a tunable, three-stage bandpass filter
which can be set to the desired frequency via a D/A converter using calibration data stored in the
instrument. This filter suppresses unwanted spurious, i.e. adjacent spectral lines from the step-recovery
diode, down to -60 dBc. The filter frequency ranges are specified as follows:
Table 7-2 Filter frequency ranges
Filter Frequency range
1 1210.5 to 2200 MHz
2 2200 to 2800 MHz
3 2800 to 3300 MHz

This seems to be the biggest difference between the SML02 and SML03, maybe the 02 doesn't have the 2nd and 3rd filter.
Anyway, first fixing.

I found on the KO4BB website an SME03 service kit, someone suggested on the groups.io to use that. It's written for windows 3.1 , there is no way to run that today unfortunately.
From what I see, there is some limited documentation on the diagnostics commands in the operation manual (SCPI commands), telling me this:

:DIAGnostic[:MEASure]:POINt?
The command triggers a measurement at a measuring point and returns the voltage measured.
The measuring point is specified by a numeric suffix (cf. service manual).
Example: :DIAG:MEAS:POIN? 2 Response: 11.56

Service manual tells me to run the software. 

[TylerPeppy]

4. This on is strange, very slow response to button presses. Often times I have to press a button for 2 seconds for it to register. The encoder is fast though, so not sure what's going on.

This is a known problem with older SML. From what I hear it is the contact pill in the elastomeric keypad that degrades. I know that some people have experimented with conductive silver, but that is apparently difficult to fix. Luckily, mine doesn't do that.

You can buy 'stickers' from Buttonworx which have a conductive pattern printed on them. They go between the original rubber key mat and the contact area. They work like a charm.

Interesting. I've google searched for this quickly and it also gave me an alibaba link as well.
Searching on AliExpress, " Diameter 3/4/5/6 MM 4 Sizes Black Conductive Rubber Pads Keypad Repair Kit" is an item for ~3 EUR.

This is interesting, when you mention the harmonics. They should not appear that high relative to the carrier. This is almost the exact same issue from my B3 pulse modulation option, where the switch MMIC died and one of the output amplifiers was also broken according to test points.

Not entirely sure what components were all broken on the B3, as I bypassed it and called it a day. I am hoping to swap the switches out at some point, though.

I also though so. I have low frequency output and modulation input, which might be an option (not shown on the picture). Anyway, I'll crack it open and see what's inside. And I really apricate the help.

Here is some brain dump from me:

From what I can see in the manual, the difference between the SML02 and SML03 is the number of filters after the step recovery diode.

Table 7-1 Frequency schedule of Output Unit 2 GHz / 3 GHz
n Input frequency Output frequency
2 605.25 to 909 MHz 1210.5 to 1818 MHz
3 606 to 885 MHz 1818 to 2655 MHz
4 663.75 to 825 MHz 2655 to 3300 MHz
The selection of the desired spectral line is made by means of a tunable, three-stage bandpass filter
which can be set to the desired frequency via a D/A converter using calibration data stored in the
instrument. This filter suppresses unwanted spurious, i.e. adjacent spectral lines from the step-recovery
diode, down to -60 dBc. The filter frequency ranges are specified as follows:
Table 7-2 Filter frequency ranges
Filter Frequency range
1 1210.5 to 2200 MHz
2 2200 to 2800 MHz
3 2800 to 3300 MHz

This seems to be the biggest difference between the SML02 and SML03, maybe the 02 doesn't have the 2nd and 3rd filter.
Anyway, first fixing.

I found on the KO4BB website an SME03 service kit, someone suggested on the groups.io to use that. It's written for windows 3.1 , there is no way to run that today unfortunately.
From what I see, there is some limited documentation on the diagnostics commands in the operation manual (SCPI commands), telling me this:

:DIAGnostic[:MEASure]:POINt?
The command triggers a measurement at a measuring point and returns the voltage measured.
The measuring point is specified by a numeric suffix (cf. service manual).
Example: :DIAG:MEAS:POIN? 2 Response: 11.56

Service manual tells me to run the software. 

The test points are a bit hit or miss with the software tool, I have had decent luck just using the built-in test point utility on the SML to check them. Problem is, they aren't all available unfortunately.

[tszaboo]

OK, I made some time to open it up. The output coax assembly looked strange. There is an intentional split in the ground connection. I measured it with a VNA, things are even more strange now.
I'm not sure why RS wouldn't just connect the ground together in the first place.
But then, the S21 and S12 is down like 40dB or more. I even flipped it, because maybe they are doing something directional.
Did anyone measure this output assembly before?
Anyone wants to take bets before I connect the spectrum analyzer on the SMA directly, that this is the main issue for the signal loss??

[KungFuJosh]

OK, I made some time to open it up. The output coax assembly looked strange.

The cable shouldn't look like that. It appears somebody tried to repair it. It should be a solid/single semi-rigid cable with the cores on it directly.

The B19 output cable illustrates this more easily. The front panel cable is the same idea, but shorter.

Thanks,
Josh

[nctnico]

OK, I made some time to open it up. The output coax assembly looked strange.

The cable shouldn't look like that. It appears somebody tried to repair it. It should be a single semi-rigid cable with the cores on it directly.

I agree with that 100%. The output cable should not look like that. It needs to be replaced.

[tszaboo]

OK, I made some time to open it up. The output coax assembly looked strange.

The cable shouldn't look like that. It appears somebody tried to repair it. It should be a solid/single semi-rigid cable with the cores on it directly.

The B19 output cable illustrates this more easily. The front panel cable is the same idea, but shorter.

Thanks,
Josh

Thanks Josh, I also think so. But it looks very intentional on first sight.
But now I've put it under the microscope, and the split looks like it broke somehow? I've never seen a broken hardline coax where it was broken on the outside.

[KungFuJosh]

Thanks Josh, I also think so. But it looks very intentional on first sight.
But now I've put it under the microscope, and the split looks like it broke somehow? I've never seen a broken hardline coax where it was broken on the outside.

With the wrong tools, and a solid lack of skills, anything is possible.

[tszaboo]

Thanks Josh, I also think so. But it looks very intentional on first sight.
But now I've put it under the microscope, and the split looks like it broke somehow? I've never seen a broken hardline coax where it was broken on the outside.

With the wrong tools, and a solid lack of skills, anything is possible.

Actually, this could be shipping damage. I ordered it to a DHL service point. The guy handed over the package by slapping it on the counter on the shorter edge of the package. It took the better half of me not to shout with him. Fragile written on the package and all...
Other time I got a HP Power supply that was just tossed into my garbage bin (for paper) and a note delivered to me. I think these delivery companies could somehow break solid steel balls.
I'm not saying it was this. Maybe the seller also received it with this already. 

[nctnico]

Thanks Josh, I also think so. But it looks very intentional on first sight.
But now I've put it under the microscope, and the split looks like it broke somehow? I've never seen a broken hardline coax where it was broken on the outside.

With the wrong tools, and a solid lack of skills, anything is possible.

Yep. Every now and then I lend a hand with test equipment repairs and you don't want to know what I've encountered so far.

Actually, this could be shipping damage. I ordered it to a DHL service point. The guy handed over the package by slapping it on the counter on the shorter edge of the package. It took the better half of me not to shout with him. Fragile written on the package and all...

That is your problem. Equipment should be packed so it survives falling off a flight of stairs. Couriers really don't care about fragile signs etc. Just make sure equipment is packed properly. If you don't then you should accept shipping damage. I've seen a fair amount of that...

Bubble wrap or instant foam are the best ways but with instant foam it is super important to place equipment in the middle so a foam layer can form around the equipment. Either way: there should be at least 8cm of bubble wrap, air cushioning or foam between the equipment and the card board.

[KungFuJosh]

It doesn't look like it happened in shipping. The photo looks like part of the cable is semi-transparent instead of grey. Somebody botched it.

[tszaboo]

It doesn't look like it happened in shipping. The photo looks like part of the cable is semi-transparent instead of grey. Somebody botched it.

You are right, this is someone's doing. Here it is compared to a high quality N to SMA connector. I hope to use this (and a semi-flex SMA-SMA cable) to fix the front connector. The center contact is a lot deeper on the SML's connector, I don't think it even makes contact at all.
On the good connector the center pin is 4.3mm deep, on the bad one it's 9.2mm. 

[KungFuJosh]

It doesn't look like it happened in shipping. The photo looks like part of the cable is semi-transparent instead of grey. Somebody botched it.

You are right, this is someone's doing. Here it is compared to a high quality N to SMA connector. I hope to use this (and a semi-flex SMA-SMA cable) to fix the front connector. The center contact is a lot deeper on the SML's connector, I don't think it even makes contact at all.
On the good connector the center pin is 4.3mm deep, on the bad one it's 9.2mm. 

If you can figure out how to reuse those cores, you'll be happy. If you can get a semi-rigid cable, and add a crimp connector, that would probably work.

[tszaboo]

Taking it apart further... Cleaning the keypad seemed to help a little, feels better, but it still doesn't register a lot of button presses so I will try the repair kit.
I measured the SMA connector directly, the level is still bad. I did the test

On the other hand, can you set up a slow sweep from 9kHz - 2.2GHz and leave trace persistence ON on the SA for the test? I am curious to see if the gain error is flat.

It is, that variation is more likely my bad quality coax.
 

I don't have OCXO or pulse modulation options.

I looked into the diagnostic menu. What jumped at me is the operation time is 900 hours, which is not a lot. And the "overload protection count" was 122 which sounds like a lot, and I'm not sure how that even happens so often to such an instrument. I'm not sure either if this is kept in the battery backed RAM and if this resets.

Harmonics seem to appear only below 750MHz.
I measured the output of the 2/3GHz output, which was 8.6dBm when the output was set to 0dBm. And it went all the way to 22dBm, when the output was set to maximum. So this module is working, the attenuator has the issue.

Some other things that I've noticed. I went through the built in testpoints (in the menu), and TP 404 and 407 was 0V. This is called MB 8VHF1, which makes me think this is supposed to be 8V, and not 0V. Can someone check it please? This is not covered in the manual.

[KungFuJosh]

I looked into the diagnostic menu. What jumped at me is the operation time is 900 hours, which is not a lot. And the "overload protection count" was 122 which sounds like a lot, and I'm not sure how that even happens so often to such an instrument. I'm not sure either if this is kept in the battery backed RAM and if this resets.

That's nothing. 100,000 would even be considered low. That said, the number could be inaccurate.

The overload thing probably doesn't matter. I think that's automatically counted when you set the output past a specific dB level.

Harmonics seem to appear only below 750MHz.
I measured the output of the 2/3GHz output, which was 8.6dBm when the output was set to 0dBm. And it went all the way to 22dBm, when the output was set to maximum. So this module is working, the attenuator has the issue.

That's (unfortunately) normal. The attenuator/output module is the most common failure point. You can use the switch conversion board I shared earlier in the thread if you need to replace them with something new.

Thanks,
Josh

[tszaboo]

I looked into the diagnostic menu. What jumped at me is the operation time is 900 hours, which is not a lot. And the "overload protection count" was 122 which sounds like a lot, and I'm not sure how that even happens so often to such an instrument. I'm not sure either if this is kept in the battery backed RAM and if this resets.

That's nothing. 100,000 would even be considered low. That said, the number could be inaccurate.

The overload thing probably doesn't matter. I think that's automatically counted when you set the output past a specific dB level.

Harmonics seem to appear only below 750MHz.
I measured the output of the 2/3GHz output, which was 8.6dBm when the output was set to 0dBm. And it went all the way to 22dBm, when the output was set to maximum. So this module is working, the attenuator has the issue.

That's (unfortunately) normal. The attenuator/output module is the most common failure point. You can use the switch conversion board I shared earlier in the thread if you need to replace them with something new.

Thanks,
Josh

Thanks, I was also expecting this to be the issue. I've taken it apart, and found 3 switches that had the issue, and a 4th may be on it's way out. There was a lot of voltage drop on the resistor series with the IC control pin.

Looks like my board is not made with the P35-4227-C06-300, like everyone else's but it's with SSW-124. The switches have a W1 marking like the datasheet.
https://www.alldatasheet.com/datasheet-pdf/view/157236/STANFORD/SSW-124.html
Standford Microdevices became Sirenza Microdevices
In 2007, Sirenza Microdevices was acquired by RF Micro Devices
Which is now Qorvo, who makes everything that you don't order a million or pay like the military obsolete.

[rf-messkopf]

Some other things that I've noticed. I went through the built in testpoints (in the menu), and TP 404 and 407 was 0V. This is called MB 8VHF1, which makes me think this is supposed to be 8V, and not 0V. Can someone check it please? This is not covered in the manual.

According to the service manual:

Diagnostic point	Description	min	max
404 D_8VHF1	Supply voltage for RF stages	7.55	8.35
407 D_8VHF2	Supply voltage for RF stages	7.55	8.35

Both test points belong to the output module. The manual says "Check 8-V voltage regulator N21, V146" and "Check 8-V voltage regulator N21, V234" when these are out of tolerance.

[bte]

I found on the KO4BB website an SME03 service kit, someone suggested on the groups.io to use that. It's written for windows 3.1 , there is no way to run that today unfortunately.

Out of curiosity, I tried that software on 32-bit Windows 7 and it seemed to start up fine. Maybe the software could be made to work along with a recent NI-VISA installation and a PCI-E (i.e. something recent enough that has a 32-bit Windows 7 driver) GPIB adapter.

[tszaboo]

Some other things that I've noticed. I went through the built in testpoints (in the menu), and TP 404 and 407 was 0V. This is called MB 8VHF1, which makes me think this is supposed to be 8V, and not 0V. Can someone check it please? This is not covered in the manual.

According to the service manual:

Diagnostic point	Description	min	max
404 D_8VHF1	Supply voltage for RF stages	7.55	8.35
407 D_8VHF2	Supply voltage for RF stages	7.55	8.35

Both test points belong to the output module. The manual says "Check 8-V voltage regulator N21, V146" and "Check 8-V voltage regulator N21, V234" when these are out of tolerance.

Thanks. I just found out that the service manual comes in two volumes and I only found volume 2 so far. But Volume 1 was on Elektrotanya.
I suppose the circuit diagram isn't available, or am I not looking hard enough?
Also worth reading, CCFL to LED conversion:
https://www.elektormagazine.com/labs/repair-of-a-rohdeschwarz-sml03-rf-signal-generator

[KungFuJosh]

I have maintenance, firmware, and service manuals attached to the first couple posts in this thread.

I think I shared them somewhere else before, but I just added service to the 2nd.

[tszaboo]

I've been looking at this attenuator for a few hours now, trying to make some sense of it.
The manual gives some idea what's going on, but I think it's somehow misleading on so many levels.
So we have a declaration for the attenuation levels:

The different attenuator pads of 5, 10, 15, 30, 40dB permit to set the attenuation in
the range from 0 to 125dB with 5-dB resolution. The attenuator pads are combined
in 5 groups and switched with GaAs switches.

The 5 groups makes no sense with the layout that we have.
Then we have a truth table for it:

If I read this, SWI1 2 and 3 are 5dB attenuators, SWI4-5 are 15dB and SWI6 and 7 are 40dB, total 125dB.
Below is a picture (from Josh) that I edited, numbering the switches my own way, and drawing out all the 0dB loss paths in blue.

What we actually have is this: SW 1 2 and 3 connects the input directly to the output with some limiters in between.

And then I started tracing my attenuator. I set it to the highest output level, and got almost 0dBm out of my generator. Judging by the IC's datasheet, negative control voltage turns the switch to that path.
And I was tracing the RF, by looking at which control voltage is present where.
The confusing part was, that the pins are swapped, the RF pin that's conducting is to the opposite of the negative control voltage. Anyway by the looks of it for me switch 1, 12 and 4 is damaged, they have almost 0V on the control pins, where the other side of the resistor has -7V. Ah the resistors. There is 100Ohm on the other side of the board in series with the resistor on this side, which is 6800Ohm. I placed a 1Kohm resistor parallel with the 6.8K to se if it makes a difference. Maybe just a bit of extra current brings back the switch. And it made a difference,  I got extra 6dB to the output signal. Which is still low, so this is not a solution.
11 and 12 switches in two different attenuation settings, where there are switches on both sides of the attenuators.
10 and 4 acts another way, where they also select two different attenuations, but their outputs are a single trace. What must be happening here is that they use the absorvative characteristics of the attenuator to make the non-selected RF path into a resistor.
Switch 5 and 3 can select whether switch 4 is in the path. Probably it's a way to increase isolation, otherwise it doesn't make sense to use 3 switches for 3 different gains.
Similarly, 6 and 7 selects whether 10 is in the path.

SW4 doesn't change until we reach -80dBm on the output. SW 5 doesn't change until -40dBm. So I thinkSW4 selects either a 40dB or a 80 dB attenuation.
There is also something going on with SW3. The driving of it is different, and depending on the output level, it has either -5V or -8V on the pins.

For me it's clear that my main issue is switch 1, which is always in the path. I don't really care much about the low ranges for now, plus it doesn't work anyway, I would much rather have a generator that can output 19dBm. So I'm going to swap SW1 and 5, or short out 5.

[tszaboo]

So I've replaced that input switch.
And I'm seeing something quite exceptional.
The driving pin of these MMIC switches work as a non volatile memristor. Memristor is a device that stores a resistance value, and non-volatile, because even after removing the power it stays that way. At least the broken ones, the good ones are fine.
Heating it up, it returns to about 2Mohm resistance between the gate and ground. This is the same for the good ones. And depending on their state when turned off, they might be in this 2MOhm state, or something as low a 2-3 KOhm. That's why they don't work, they pull the pin low when this happens, it makes a resistor divider with the 6.8KOhm resistor in front of the IC.

[tszaboo]

The IC I swapped to the input of the attenuator was also toast, so I went in an other direction with the repair, hacking in another switch IC. So this apparently works.
That trace is in max hold, that' why there is something every 100MHz. I got 17dBm max out of the 19 nominal. The difficulty of hacking in this IC is quite high, but I wanted to test it before committing more effort of making an adapter PCB. Sorry for the upside down generator in the picture, but it's this way. The generator was set to 5dBm.
The IC is MASW-007935.

[tszaboo]

While waiting for my adapter PCB, I turned my attention to the main RF generator board. Here I have a power supply failure, Some 8V is missing from somewhere. Below are my teardown pictures, I'm going to annotate it with some reverse engineering.

[tszaboo]

Below is my effort trying to understand this board.
My first big question mark was, where is the VCO? And then it hit me, the entire board is pretty much just a large multiplier, there is no VCO here..
I'm going to go back to the manual, trying to find some clues, to understand what this 8V actually does.

mod: And closing it back, because the testpoint is not on this board.

[tszaboo]

OK, another set of teardown photos. You can see the block diagram on the cover of the Mainboard.
My 8V power supply magically fixed itself after opening and turning it on. 
I was measuring 8V on that point. The fact they call it "Supply voltage for RF stages" makes me think this is the right point.
Looks like someone had a repair at this, the white part ( RF transistor? MMIC?) seems to have some extra solder and the PCB is doing something there.
I've also included some thermal images. Some of these parts run hot, 70-80 degrees, below the cover, so it's not an airflow issue. And I think these parts meant to run hot, they are reverse geometry resistors (0612) some of it in parallel.