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

[CrazyTiger]

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.

In the past as an experiment, i tried making an  tiny carrier board to replace the existing switches marked W1 on the attenuator to something modern, it worked sort off, since I had the adjustment program, NRVD and NRV-Z51, even adjusted the unit. Problem was I could not get the frequency level response linear above 2GHz, main reason likely the carrier board was FR4, 1.6mm, with limited space, re-routing rf traces becomes a hasle and noisy, could have performed better with a better pcb material and thinner board. But it was a weekend experiment.

[KungFuJosh]

I had the adjustment program

Which one? SME03 or something else? Do you still have it?

Thanks,
Josh

[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.

In the past as an experiment, i tried making an  tiny carrier board to replace the existing switches marked W1 on the attenuator to something modern, it worked sort off, since I had the adjustment program, NRVD and NRV-Z51, even adjusted the unit. Problem was I could not get the frequency level response linear above 2GHz, main reason likely the carrier board was FR4, 1.6mm, with limited space, re-routing rf traces becomes a hasle and noisy, could have performed better with a better pcb material and thinner board. But it was a weekend experiment.

Interesting. I have an SML02, not 03. So the frequency is limited.*
I couldn't really attempt to make the trace 50 Ohm on this adapter board. There is a complicated transition from an SO8 pad to a castellated edge, then a PCB, then the pins of the SOT23-6 switch. The trace is only 1.5mm long, and it's less than the half of the total distance RF needs to travel.

*Though from what I can see maybe there isn't any hardware difference between these two types.

[rf-messkopf]

Problem was I could not get the frequency level response linear above 2GHz, main reason likely the carrier board was FR4, 1.6mm, with limited space, re-routing rf traces becomes a hasle and noisy, could have performed better with a better pcb material and thinner board. But it was a weekend experiment.

Another reason could be that the adapter board compromises the impedance matching between switch and transmission line, leading to reflections which cause ripple. Also, the electrical length of the lines may have been slightly changed. If the circuit relies on phase matching, this might also lead to ripple.

Edit: The layout might also compensate for the SO-8 pin of the original switch by matching elements in the layout (e.g., trace thickness). This may be upset by the adapter board, even if you try to stick to 50 ohms impedance as close as you can.

[KungFuJosh]

*Though from what I can see maybe there isn't any hardware difference between these two types.

There might be something as stupid as jumpers that set the top frequency in the 2G/3G module. I never tried them though, since all the units I had were 3GHz. They use the same module for both units, so I dunno if they bothered to change anything else or not.

Admittedly, it's a dumb guess, but what do those jumpers do? 🤔

[tszaboo]

*Though from what I can see maybe there isn't any hardware difference between these two types.

There might be something as stupid as jumpers that set the top frequency in the 2G/3G module. I never tried them though, since all the units I had were 3GHz. They use the same module for both units, so I dunno if they bothered to change anything else or not.

Admittedly, it's a dumb guess, but what do those jumpers do? 🤔

It's a great guess, but it's only part of it.
I checked my jumpers and it's in a different position. When I moved it to the same position as on this picture, I get an error while starting it up. EEPROM data missing, calibration lost. And the generator only goes to 1.1 GHz.
So maybe rewriting the EEPROM with content from an SML03 and then manually calibrating it, might just give it that extra speed. Which I wouldn't mind, testing 2.45GHz stuff would be interesting. I have to see into the service menu as well, maybe there is something to "reset" the EEPROM content, but I will attempt any of this after I made a backup from it.

[KungFuJosh]

Does running the self calibration help?

[TylerPeppy]

*Though from what I can see maybe there isn't any hardware difference between these two types.

There might be something as stupid as jumpers that set the top frequency in the 2G/3G module. I never tried them though, since all the units I had were 3GHz. They use the same module for both units, so I dunno if they bothered to change anything else or not.

Admittedly, it's a dumb guess, but what do those jumpers do? 🤔

You might be onto something. Here's the inside of my SML03. Notice the jumper closest the ribbon cable is in the upper position, rather than the lower one on the 2G/3G module...

I wonder if this is how one gets the 3GHz option, and then the unit is subsequently coded to work in the extended range.

Tyler

[KungFuJosh]

I dunno if there's any coding involved. This series seems to me to be "slap it together and it works."

My guess is that you flip the jumpers, and either run the self cal, or the external cal script to accommodate the new range. I'm hoping that the self cal will be sufficient.

[tszaboo]

I don't have the module in front of me. As you can see the jumper is connected to pin 14 of that connector.
According to the service manual, Pin 14 is EEPROM I2C DATA. I don't see on my previous image where the other jumper would go, but I'm going to guess it's the other I2C pin.
This, on the bottom right, looks like a pullup resistor to me, that gets selected by the jumper.
Anyway, it should be trivial to read out this EEPROM, I'll make some time this weekend to do that.

[tszaboo]

See below the EEPROM's content of my 2/3GHz output unit.
There is one text on it, OPU3, which is the name of this unit.
The rest of the EEPROM is FF.
I get the error messages 313 and 315 when I change the jumpers, 313 is lost calibration, 315 is lost configuration. While it uses the

Code: [Select]

0000: 05 00 2C 00 CE 47 10 00 00 20 00 00 C7 61 A6 00  |..,..G... ...a..|
0010: 02 00 95 01 CA 7E F9 05 BC 2A 00 00 CA 00 4F 50  |....~...*....OP|
0020: 55 33 00 00 00 00 00 00 00 00 00 00 64 00 1C 00  |U3..........d...|
0030: 80 FF 00 00 00 00 00 00 00 00 00 00 00 00 00 00  |................|
0040: 00 00 00 00 00 00 00 00 FF FF 00 00 FF FF FF FF  |................|The jumper X2 seems to set the middle pin either to the SDA, or GND. Considering that SDA would be pulled up most of the time, I guess this is just to select 3.3V or GND on something.

[tszaboo]

And here is the attenuator's EEPROM content while at it.
It's a lot more data. It has ".Attenuator2.2.d" in it as ASCII.

[KungFuJosh]

Have you tried running the self calibration after changing the jumpers?

[tszaboo]

Have you tried running the self calibration after changing the jumpers?

I'll fix the attenuator first, before doing any sort of calibration. I think half the steps aren't working properly. Then I'll do one without changing the jumpers to make sure my procedure is right.

[CrazyTiger]

Problem was I could not get the frequency level response linear above 2GHz, main reason likely the carrier board was FR4, 1.6mm, with limited space, re-routing rf traces becomes a hasle and noisy, could have performed better with a better pcb material and thinner board. But it was a weekend experiment.

Another reason could be that the adapter board compromises the impedance matching between switch and transmission line, leading to reflections which cause ripple. Also, the electrical length of the lines may have been slightly changed. If the circuit relies on phase matching, this might also lead to ripple.

Edit: The layout might also compensate for the SO-8 pin of the original switch by matching elements in the layout (e.g., trace thickness). This may be upset by the adapter board, even if you try to stick to 50 ohms impedance as close as you can.

Agreed, It was extremely small space with RF traces going in a X layout, and nearly imposible to create a proper 50 ohm trace. I was happy that it worked to certain extent. The SML is with 70 y.o man now, he is happy testing his homebrew radios using it.

[CrazyTiger]

I had the adjustment program

Which one? SME03 or something else? Do you still have it?

Thanks,
Josh

Everyone can download from my google drive here

https://drive.google.com/file/d/1KbRV6YfIuk0Jys9V2ygdzytKlReskcqF/view?usp=sharing

[KungFuJosh]

Thanks for sharing that!

[tszaboo]

Well, I've just tested this, and at 100MHz, it gave me the correct levels from 15dBm to -120dBm. I call that a success.
I think I replaced unnecessarily maybe 2 switches because I made a short under the repair PCB.
If anyone wants one of these boards, PM me.

Now, back to that keyboard repair (which I forgot to order conductive pads) and see what I can do with that LCD lines.
And testing, if the frequency response was affected by the switches. Maybe a calibration.

BTW, I don't like these GAAS switches. When not powered, they are like 2 Ohm to ground. Makes searches for short circuit practically impossible.

[KungFuJosh]

Do you mind sharing the gerbers? I shared gerbers previously for a different option as well.

[tszaboo]

Do you mind sharing the gerbers? I shared gerbers previously for a different option as well.

Sure, I wanted to test is a bit to make sure it works. I'll gather the files and make a github repo for it when I have time.
Looks good up to 2GHz. That variation on the amplitude is not unlike the original parts. The picture is a mx hold trace with 10dBm nominal signal.
I see a bit of 2nd harmonic, but by the looks of it the specifications are -30dBc? Might be even in spec.
Now, onto calibrating it. I guess I jut terminate the output and send it. It's doing that now.
...
So that run just fine. No real change on the output level, because the attenuator is an external calibration.
 By the looks of it, it's 30 something attenuation steps at around 40 something frequency steps.
So counting only 30 seconds each step, it's still 10 hours to do 
I was able to run the service program, but I don't have the right equipment to do that cal. So I guess next would be
1) Figuring out if I can control my SA and the SML02 over SCPI
2) Figuring out to change calibration values over SCPI
3) Getting the right RBW/VBW values
4) Wishing that I had a power meter
5) calibrate the RF cable
6) Writing python script to calibrate it, or vibe code it
Sounds easy 

[KungFuJosh]

1) Figuring out if I can control my SA and the SML02 over SCPI
2) Figuring out to change calibration values over SCPI
3) Getting the right RBW/VBW values
4) Wishing that I had a power meter
5) calibrate the RF cable
6) Writing python script to calibrate it, or vibe code it
Sounds easy 

I made a definition for the SML series to work with TestController. I also made one for the V3500A. TestController has a built in Parameter Sweeper, so I've done tests sweeping across the full range of the SML03 (and the SMIQ06B and others).

The SCPI commands are pretty simple for running the tests and the manual has commands for calibration. I don't know if those commands are any different than the front panel self-cal though.

[tszaboo]

I didn't really had time to work on this, other than some software claude coded for calibration.
When doing a verification, it turned out that in some ranges the signal was 5dB down, so I have to go back to solder before I can attempt to calibrate it. At least everything now is controlled by SCPI.

[tszaboo]

So I fixed that 5dB error, yet another switch needed to be replaced.
I get this on my calibration script, everything looks good on lower frequencies. At higher frequencies, and higher attenuation, the level gets wrong, about 900MHz and 80dBm is where it goes wrong. I have to look into this further, maybe the matching is different. There are a lot of L/C parts manually soldered onto the switches. I placed these back, but maybe they need to be selected specifically. See an example output below.

Code: [Select]

1100.0000 MHz ===
 step   nominal  measured   error  drift  off-ctr   ref    rbw     ppm
----------------------------------------------------------------------------------
    0      5.50      5.15   -0.35   0.00     0.0%    20    30k   -2.31
    5      0.50      1.26   +0.76   0.00     0.0%    10    30k   -2.31
   10     -4.50     -3.40   +1.10   0.00     0.0%    10    30k   -2.31
   15     -9.50     -8.39   +1.11   0.00     0.0%     0    30k   -2.31
   20    -14.50    -13.64   +0.86   0.00     0.0%     0    30k   -2.31
   25    -19.50    -18.40   +1.10   0.00     0.0%   -10    30k   -2.31
   30    -24.50    -23.11   +1.39   0.00     0.0%   -10    30k   -2.31
   35    -29.50    -29.59   -0.09   0.00     0.0%   -20     3k   -2.31
   40    -34.50    -34.18   +0.32   0.00     0.0%   -20     3k   -2.31
   45    -39.50    -38.41   +1.09   0.00     0.0%   -30     3k   -2.31
   50    -44.50    -43.23   +1.27   0.00     0.0%   -30     3k   -2.31
   55    -49.50    -48.16   +1.34   0.00     0.0%   -40     3k   -2.31
   60    -54.50    -53.20   +1.30   0.00     0.0%   -40     3k   -2.31
   65    -59.50    -58.01   +1.49   0.00     0.2%   -50    300   -2.31
   70    -64.50    -62.70   +1.80   0.00     0.2%   -50    300   -2.31
   75    -69.50    -69.22   +0.28   0.00     0.2%   -60    300   -2.31
   80    -74.50    -74.01   +0.49   0.00     0.2%   -60    300   -2.31
   85    -79.50    -88.46   -8.96   0.00     0.2%   -70    300   -2.31
   90    -84.50    -93.38   -8.88   0.00     0.2%   -70    300   -2.31
   95    -89.50    -98.39   -8.89   0.00     0.5%   -80    100   -2.31
  100    -94.50   -103.78   -9.28   0.00     0.5%   -80    100   -2.31
  105    -99.50   -107.97   -8.47   0.00     0.5%   -90    100   -2.31
  110   -104.50   -112.88   -8.38   0.00     0.5%   -90    100   -2.31
  115   -109.50   -119.60  -10.10   0.00     2.0%  -100     30   -2.31
  120   -114.50   -123.63   -9.13   0.00     5.1%  -100     30   -2.31
  125   -119.50   -124.02   -4.52   0.00    49.5%  -110     30   -2.31



Code: [Select]

50.0000 MHz ===
 step   nominal  measured   error  drift  off-ctr   ref    rbw     ppm
----------------------------------------------------------------------------------
    0      5.50      5.58   +0.08   0.00     0.0%    20    30k   +0.00
    5      0.50      1.48   +0.98   0.00     0.0%    10    30k   +0.00
   10     -4.50     -3.16   +1.34   0.00     0.0%    10    30k   +0.00
   15     -9.50     -8.27   +1.23   0.00     0.0%     0    30k   +0.00
   20    -14.50    -13.23   +1.27   0.00     0.0%     0    30k   +0.00
   25    -19.50    -17.94   +1.56   0.00     0.0%   -10    30k   +0.00
   30    -24.50    -22.96   +1.54   0.00     0.0%   -10    30k   +0.00
   35    -29.50    -28.49   +1.01   0.00     0.0%   -20     3k   +0.00
   40    -34.50    -33.10   +1.40   0.00     0.0%   -20     3k   +0.00
   45    -39.50    -38.19   +1.31   0.00     0.0%   -30     3k   +0.00
   50    -44.50    -42.79   +1.71   0.00     0.0%   -30     3k   +0.00
   55    -49.50    -47.89   +1.61   0.00     0.0%   -40     3k   +0.00
   60    -54.50    -52.87   +1.63   0.00     0.0%   -40     3k   +0.00
   65    -59.50    -57.58   +1.92   0.00     0.2%   -50    300   +0.00
   70    -64.50    -62.61   +1.89   0.00     0.2%   -50    300   +0.00
   75    -69.50    -68.17   +1.33   0.00     0.2%   -60    300   +0.00
   80    -74.50    -72.87   +1.63   0.00     0.2%   -60    300   +0.00
   85    -79.50    -78.47   +1.03   0.00     0.2%   -70    300   +0.00
   90    -84.50    -83.11   +1.39   0.00     0.2%   -70    300   +0.00
   95    -89.50    -88.19   +1.31   0.00     0.5%   -80    100   +0.00
  100    -94.50    -93.14   +1.36   0.00     0.5%   -80    100   +0.00
  105    -99.50    -98.03   +1.47   0.00     0.5%   -90    100   +0.00
  110   -104.50   -102.92   +1.58   0.00     0.5%   -90    100   +0.00
  115   -109.50   -107.89   +1.61   0.00     2.4%  -100     30   +0.00
  120   -114.50   -112.62   +1.88   0.00     2.4%  -100     30   +0.00
  125   -119.50   -119.16   +0.34   0.00     2.4%  -110     30   +0.00

[KungFuJosh]

I dunno how easy this would be to test, but I'm curious if you have any issues with the 2G/3G module bypassed.

[TylerPeppy]

So I fixed that 5dB error, yet another switch needed to be replaced.
I get this on my calibration script, everything looks good on lower frequencies. At higher frequencies, and higher attenuation, the level gets wrong, about 900MHz and 80dBm is where it goes wrong. I have to look into this further, maybe the matching is different. There are a lot of L/C parts manually soldered onto the switches. I placed these back, but maybe they need to be selected specifically. See an example output below.

Code: [Select]

1100.0000 MHz ===
 step   nominal  measured   error  drift  off-ctr   ref    rbw     ppm
----------------------------------------------------------------------------------
    0      5.50      5.15   -0.35   0.00     0.0%    20    30k   -2.31
    5      0.50      1.26   +0.76   0.00     0.0%    10    30k   -2.31
   10     -4.50     -3.40   +1.10   0.00     0.0%    10    30k   -2.31
   15     -9.50     -8.39   +1.11   0.00     0.0%     0    30k   -2.31
   20    -14.50    -13.64   +0.86   0.00     0.0%     0    30k   -2.31
   25    -19.50    -18.40   +1.10   0.00     0.0%   -10    30k   -2.31
   30    -24.50    -23.11   +1.39   0.00     0.0%   -10    30k   -2.31
   35    -29.50    -29.59   -0.09   0.00     0.0%   -20     3k   -2.31
   40    -34.50    -34.18   +0.32   0.00     0.0%   -20     3k   -2.31
   45    -39.50    -38.41   +1.09   0.00     0.0%   -30     3k   -2.31
   50    -44.50    -43.23   +1.27   0.00     0.0%   -30     3k   -2.31
   55    -49.50    -48.16   +1.34   0.00     0.0%   -40     3k   -2.31
   60    -54.50    -53.20   +1.30   0.00     0.0%   -40     3k   -2.31
   65    -59.50    -58.01   +1.49   0.00     0.2%   -50    300   -2.31
   70    -64.50    -62.70   +1.80   0.00     0.2%   -50    300   -2.31
   75    -69.50    -69.22   +0.28   0.00     0.2%   -60    300   -2.31
   80    -74.50    -74.01   +0.49   0.00     0.2%   -60    300   -2.31
   85    -79.50    -88.46   -8.96   0.00     0.2%   -70    300   -2.31
   90    -84.50    -93.38   -8.88   0.00     0.2%   -70    300   -2.31
   95    -89.50    -98.39   -8.89   0.00     0.5%   -80    100   -2.31
  100    -94.50   -103.78   -9.28   0.00     0.5%   -80    100   -2.31
  105    -99.50   -107.97   -8.47   0.00     0.5%   -90    100   -2.31
  110   -104.50   -112.88   -8.38   0.00     0.5%   -90    100   -2.31
  115   -109.50   -119.60  -10.10   0.00     2.0%  -100     30   -2.31
  120   -114.50   -123.63   -9.13   0.00     5.1%  -100     30   -2.31
  125   -119.50   -124.02   -4.52   0.00    49.5%  -110     30   -2.31



Code: [Select]

50.0000 MHz ===
 step   nominal  measured   error  drift  off-ctr   ref    rbw     ppm
----------------------------------------------------------------------------------
    0      5.50      5.58   +0.08   0.00     0.0%    20    30k   +0.00
    5      0.50      1.48   +0.98   0.00     0.0%    10    30k   +0.00
   10     -4.50     -3.16   +1.34   0.00     0.0%    10    30k   +0.00
   15     -9.50     -8.27   +1.23   0.00     0.0%     0    30k   +0.00
   20    -14.50    -13.23   +1.27   0.00     0.0%     0    30k   +0.00
   25    -19.50    -17.94   +1.56   0.00     0.0%   -10    30k   +0.00
   30    -24.50    -22.96   +1.54   0.00     0.0%   -10    30k   +0.00
   35    -29.50    -28.49   +1.01   0.00     0.0%   -20     3k   +0.00
   40    -34.50    -33.10   +1.40   0.00     0.0%   -20     3k   +0.00
   45    -39.50    -38.19   +1.31   0.00     0.0%   -30     3k   +0.00
   50    -44.50    -42.79   +1.71   0.00     0.0%   -30     3k   +0.00
   55    -49.50    -47.89   +1.61   0.00     0.0%   -40     3k   +0.00
   60    -54.50    -52.87   +1.63   0.00     0.0%   -40     3k   +0.00
   65    -59.50    -57.58   +1.92   0.00     0.2%   -50    300   +0.00
   70    -64.50    -62.61   +1.89   0.00     0.2%   -50    300   +0.00
   75    -69.50    -68.17   +1.33   0.00     0.2%   -60    300   +0.00
   80    -74.50    -72.87   +1.63   0.00     0.2%   -60    300   +0.00
   85    -79.50    -78.47   +1.03   0.00     0.2%   -70    300   +0.00
   90    -84.50    -83.11   +1.39   0.00     0.2%   -70    300   +0.00
   95    -89.50    -88.19   +1.31   0.00     0.5%   -80    100   +0.00
  100    -94.50    -93.14   +1.36   0.00     0.5%   -80    100   +0.00
  105    -99.50    -98.03   +1.47   0.00     0.5%   -90    100   +0.00
  110   -104.50   -102.92   +1.58   0.00     0.5%   -90    100   +0.00
  115   -109.50   -107.89   +1.61   0.00     2.4%  -100     30   +0.00
  120   -114.50   -112.62   +1.88   0.00     2.4%  -100     30   +0.00
  125   -119.50   -119.16   +0.34   0.00     2.4%  -110     30   +0.00

I think Josh might be onto something. I am looking at the block diagram for the SML02/03 2G/3G output module. Your error at 1.100GHz seem to increase around 1.185GHz if I am not mistaken. Maybe I am missing something with the test, but in 250MHz - 1200MHz range, a section of the multiplier circuitry is completely bypassed. Passing the frequency range, and you are now switching it back in, which is potentially where the loss is originating from?

Tyler