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

[TylerPeppy]

This reset the cal after updating, did it not?

[KungFuJosh]

I don't think fw update affected factory cal, but I've done the self cal a bunch of times after that anyway. No issues with mine other than 0dB might be slightly lower than 0 last I checked, but I don't need it to be perfect.

[TylerPeppy]

I don't think fw update affected factory cal, but I've done the self cal a bunch of times after that anyway. No issues with mine other than 0dB might be slightly lower than 0 last I checked, but I don't need it to be perfect.

I just read through the service docs again, and looks like N8 is the output amplifier stage at the end of the module. It is probably not an overly complex part (hopefully), but if not misbehaving, would explain the extra harmonics I see with it introduced into the RF path.

I did try the "tap the parts and see if it fixes itself" method to no avail.

- Tyler

[KungFuJosh]

Did you clean anything with IPA? I'd clean anything suspect and all the signal path connectors.

You might need to replace the SSW-124 (marked 4227T) in there. Those are not easy to find parts though. It's easier to find an R&S donor unit to scavenge parts from. You could also try reflowing and cleaning the one you suspect. It might be a good idea to clean the whole module anyway.

[TylerPeppy]

Did you clean anything with IPA? I'd clean anything suspect and all the signal path connectors.

You might need to replace the SSW-124 (marked 4227T) in there. Those are not easy to find parts though. It's easier to find an R&S donor unit to scavenge parts from. You could also try reflowing and cleaning the one you suspect. It might be a good idea to clean the whole module anyway.

On that thought, I actually used my near field probes to trace the input signal through the 4227Ts. I have actually seen no significant signal attenuation until the signal trace arrives at the B3 amp section (closest box to output connector, and the hottest part on a thermal camera).

Apart from that, all the connectors were cleaned, and re-torqued to 0.90N•m. The board was scrubbed, and I reflowed almost all the joints on the board top and bottom side.

A donor unit is likely going to cost multiple fold compared to what I spent already, so I will keep looking at either component-level repair or just bypassing the module.

- Tyler

[KungFuJosh]

Can you share some thermal images of it?

What do the 4227Ts voltages look like inside of B3?

[TylerPeppy]

Tested all 3 switches:

SW1:
- CTRL1:  1.522V
- CTRL2: -6.066V

SW2:
- CTRL1:  1.512V
- CTRL2: -6.052V

SW3:
- CTRL1: 0.7891V
- CTRL2: -6.003V

I am guessing SW3 might have some sort of problem. But, when I select the pulse gen modulation options in the menus, CTRL1 goes to 1.5 or so volts... UTSource has some of these switches, but they're USD$ 102.76 each plus USD$ 30.00 ship... No thanks!

Attached a thermal shot of the module.

- Tyler

[KungFuJosh]

If you're comfortable soldering/desoldering those, you can swap SW2 and SW3 and see if the voltage difference follows the chip or not...or if it magically fixes the problem.

It's been a while since I checked UTSource for those, but they were $6 at the time. 

[TylerPeppy]

If you're comfortable soldering/desoldering those, you can swap SW2 and SW3 and see if the voltage difference follows the chip or not...or if it magically fixes the problem.

It's been a while since I checked UTSource for those, but they were $6 at the time. 

Interestingly, my module seems to differ from your B3 in two small ways...

The inductor near the output connector is seemingly bodged on in mine, with a wire connecting one side of it to the cap near it.

Other one, is that I have two white switches, and one SOP-8 looking one.

Not sure if it's just a component shortage that would have mixed these, but the inductor thing was interesting. I will get a picture of it when I have it open again.

[KungFuJosh]

I definitely want to see some photos. I wonder if somebody previously attempted repair.

[TylerPeppy]

Here ya go... Photos are quite hi-res, but it doesn't seem to show well on the forum.

You can also see the main output module was RMA'd in '04

- Tyler

[KungFuJosh]

You can right-click and open the image in a new tab to view it full size.

Anyway, since your worst case scenario is to bypass B3 and not have any issues anymore...

How do you feel about swapping SW2 and SW3? I'm curious about that voltage.

[TylerPeppy]

Tried removing SW3, it's bolted down pretty good with these tabs on the side, my hot air has a difficult time getting it pulled up.

[KungFuJosh]

It might be easier with some flux and an iron, but you'll need to do a good job cleaning again.

[KungFuJosh]

I requested a quote for these: https://www.utsource.net/itm/p/12607896.html

They quoted $10 each for a quantity of 10 (used  ). I don't know if quantity effects it or not. If you requested what they show as 4227T, then it's probably the wrong part anyway.

[TylerPeppy]

These are the ones I was looking at: https://www.utsource.net/itm/p/9271510.html?srsltid=AfmBOopPQx8MNEpC7gQIxkpv9SFEWlaXGT0B78kVxJB8Olxne8BJL4qE

Probably were the wrong part then.

- Tyler

[rf-messkopf]

MASWSS0157 from MACOM and AW002R2-12 from Skyworks both come in SO8 packages and seem to have the same pinout as the P35-4227-3T. The control voltage is similar since they are also GaAS devices (i.e., depletion mode MESFETs), and both are rated to 2.5 GHz. Unfortunalety they are also obsolete.

Then there is the HMC234C8 in a SO-8 package but with different pinout. Obsolete as well. Maybe you can find a suitable GaAS SPDT switch in a SOT23-6 package and bodge that in if all else fails. This will likely distort amplitude flatness, but you should be able to calibrate that out if you can get your hands on the required cal software.

These old GaAs devices are prone to failure as the processes apparently weren't yet well under control back then.

Edit: I just realize that the P35-4227-3T is a terminated switch, i.e., the switch terminal that is not in the active path is terminated into 50 ohms. You should look out for that when searching for a replacement.

[KungFuJosh]

Be careful with part identification. It's a switch (DC - 6GHz High Isolation SPDT GaAs MMIC Switch); not an oscillator or a JFET or a 2 pin socket. All of which can easily be confused by the part numbers.

This is the correct replacement (the only source I know of, unfortunately questionable being sold as used): https://www.utsource.net/itm/p/12607896.html

Datasheet attached.

[TylerPeppy]

Be careful with part identification. It's a switch (DC - 6GHz High Isolation SPDT GaAs MMIC Switch); not an oscillator or a JFET or a 2 pin socket. All of which can easily be confused by the part numbers.

This is the correct replacement (the only source I know of, unfortunately questionable being sold as used): https://www.utsource.net/itm/p/12607896.html

Datasheet attached.

Might need many of your replacements there, in the event that they are already broken. Can't say I am fond of the idea of "used" chips, but might not have any options anyways.

[KungFuJosh]

If you can manage to swap SW2 and SW3, then you'll get a better idea after testing the voltages. If the voltage difference follows the chip from SW3, then you know it's the chip.

If you can't swap the chips you have, then ordering more of them won't help until you can swap them anyway.

Something like this might help:

[rf-messkopf]

Datasheet attached.

If that's the correct datasheet, the HMC347ALP3E from Analog Devices/Hittite and the MSWA2-50+ from Minicircuits are both available and perfect matches (nonreflective SPDT GaAs switch with negative control voltage and comparable isolation and loss) -- except for the package. Since both are 3x3mm QFNs you could make a small adapter board with castellated holes on the sides that fits on the SO8 footprint.

That's the pulse modulator assembly, right? I think the attenuator in the SML has similar switches and a software temperature compensation, therefore it is important to stick with GaAs MESFET devices, although I don't think there is temperature compensation in the pulse modulator.

[TylerPeppy]

Datasheet attached.

If that's the correct datasheet, the HMC347ALP3E from Analog Devices/Hittite and the MSWA2-50+ from Minicircuits are both available and perfect matches (nonreflective SPDT GaAs switch with negative control voltage and comparable isolation and loss) -- except for the package. Since both are 3x3mm QFNs you could make a small adapter board with castellated holes on the sides that fits on the SO8 footprint.

That's the pulse modulator assembly, right? I think the attenuator in the SML has similar switches and a software temperature compensation, therefore it is important to stick with GaAs MESFET devices, although I don't think there is temperature compensation in the pulse modulator.

This could also work, but will probably be magnitudes more expensive on a gamble with the used parts. Once I get a preheater for my soldering bench, I might have a bit more luck removing this switch and then replacing all of them.

[KungFuJosh]

Datasheet attached.

If that's the correct datasheet, the HMC347ALP3E from Analog Devices/Hittite and the MSWA2-50+ from Minicircuits are both available and perfect matches (nonreflective SPDT GaAs switch with negative control voltage and comparable isolation and loss) -- except for the package. Since both are 3x3mm QFNs you could make a small adapter board with castellated holes on the sides that fits on the SO8 footprint.

That's the pulse modulator assembly, right? I think the attenuator in the SML has similar switches and a software temperature compensation, therefore it is important to stick with GaAs MESFET devices, although I don't think there is temperature compensation in the pulse modulator.

The MSWA2-50+ certainly looks interesting at $19 a pop. The other one for $104 is less appealing.

I was already considering an adapter board, but my SML is working fine so there was no urgency for me.

There was another chip people were bodging in to replace these, but I don't remember what it was. It might also be worth considering an adapter for if the part number is found again and still available at a reasonable price.

Yes, the output module/attenuator has a bunch of these same switches.

Thanks,
Josh

ETA:
I'm still reviewing the datasheets, but it looks like HMC347ALP3E is not as a good of a choice compared to the MSWA2-50+.

Comparisons up to 3GHz:
SSW-124 (original):
Max insertion loss: 1.0dB
Typical isolation: 40dB
Minimum isolation: 30dB
Operating temp: -45C to +85C
Switching speed: 3nsec
Max control voltage: -10V
VSWR: 1.25
Current Price: OBSOLETE

MSWA2-50+:
Max insertion loss: 1.0dB
Typical isolation: 47dB
Minimum isolation: 29dB
Operating temp: -40C to +85C
Switching speed: 7nsec
Max control voltage: -8.5V
VSWR: 1.33
Current Price: $19

KSW-2-46+ (has been used previously for this purpose):
Max insertion loss: 1.8dB
Typical isolation: 40dB
Minimum isolation: 28dB
Operating temp: -55C to +100C
Switching speed: 4nsec
Max control voltage: -8V
VSWR: 1.3
Current Price: OBSOLETE

HMC347ALP3E:
Max insertion loss: 1.9dB
Typical isolation: 54dB
Minimum isolation: 50dB
Operating temp: -55C to +85C
Switching speed: 10nsec
Max control voltage: -7.5V
VSWR: 1.33
Current Price: $104

R&S has been known to push the voltages on these switches (which likely contributes to their failure rate), so HMC347ALP3E is not a good choice beyond the price.

Anyway, somebody smarter than me should compare further to make sure we didn't miss anything.

[rf-messkopf]

R&S has been known to push the voltages on these switches (which likely contributes to their failure rate), so HMC347ALP3E is not a good choice beyond the price.

Anyway, somebody smarter than me should compare further to make sure we didn't miss anything.

No idea. The HMC347ALP3E is spec'd to higher frequencies, so you can speculate that its FETs have shorter gates and hence a smaller Schottky breakdown voltage. Depending on which philosophy the company follows with their abs. maximum ratings and which margins they apply, this may be the reason for the -7.5V max control voltage. On the other hand, these devices might be manufactured in a more modern process than the original ones, which makes them much less prone to deterioration processes like gate sinking (i.e., the gate diffusing into the semiconductor material), which is dependent on gate-source voltage and temperature.

But that is all just reading tea leaves. For me, the deciding factor would be the price. 

That said, the old GaAs devices are notorious for failure, even if they are new old stock. I've had a SML where the signal level fluctuated wildly as soon as it warmed up. Presumably the switches in the attenuator. Since I bought it from a dealer, I simply sent it back (and received a better one as compensation). But I’ve replaced a few of the damn things in various devices. Most recently in an R&S EB200 wideband receiver.

[TylerPeppy]

R&S has been known to push the voltages on these switches (which likely contributes to their failure rate), so HMC347ALP3E is not a good choice beyond the price.

Anyway, somebody smarter than me should compare further to make sure we didn't miss anything.

No idea. The HMC347ALP3E is spec'd to higher frequencies, so you can speculate that its FETs have shorter gates and hence a smaller Schottky breakdown voltage. Depending on which philosophy the company follows with their abs. maximum ratings and which margins they apply, this may be the reason for the -7.5V max control voltage. On the other hand, these devices might be manufactured in a more modern process than the original ones, which makes them much less prone to deterioration processes like gate sinking (i.e., the gate diffusing into the semiconductor material), which is dependent on gate-source voltage and temperature.

But that is all just reading tea leaves. For me, the deciding factor would be the price. 

That said, the old GaAs devices are notorious for failure, even if they are new old stock. I've had a SML where the signal level fluctuated wildly as soon as it warmed up. Presumably the switches in the attenuator. Since I bought it from a dealer, I simply sent it back (and received a better one as compensation). But I’ve replaced a few of the damn things in various devices. Most recently in an R&S EB200 wideband receiver.

There are a few other options that would have probably been better avenues. For the price I simply couldn't give this guy up though. Siglent's SSG5040X would probably have been a much better decision.

eBay seller refuses to help me at all it seems, so I am kind of out in the weeds on this one hence why I started my own debug. He's been MIA for the past while. I guess I should have expected this though.

- Tyler