E4433B RF sig gen repair

[microbug]

I put the synth board back in with the lid loosely resting on top, and the PLL was still losing lock. However, when I screwed the lid down tightly, the PLL stopped losing lock -- so I will ignore this error for now.

I had a look at the VCO output, which is well marked on the PCB with a socket for an RF probe. It's all over the place -- I expected to see 0.5-1GHz, but it was a mess of peaks that were almost never in the right place. This testpoint was on the output of RF amp U16, just after a 4dB pad. I also checked directly on the output of U16 and saw the same.

Next I checked the input to U16. I swept the sig gen from 2GHz-4GHz, and lo and behold, a clean peak moving across the spectrum! Power level is consistently -16dBm, which is about right if we assume the coax I soldered on forms a splitter that drops another 6dB. The coax is quite thin and lossy too.

Then I googled 'Agilent "A06" repair'. What do I find but another thread pointing the finger at the same part number. Very suspicious...

I will probe around a bit more and check the other Avago MMIC (A03/A06/A07) parts on the board.

Edit: will take a better picture of the problem area later, but U16 is visible on the zoomed picture of the synth board a few posts ago.

[R-1125F]

It looks like you found the problem. This should not be too dificult to fix.  MSA-0686 is available on ebay but who knows if you get real of fake.  The TSP repair video has a similar repair with a part that needed a different bias.   

[vaualbus]

Cool!
Seems you found at least something "wrong"! I would first check the BIAS T of that amplifier maybe some the resistor/inductor got open? 
Also I would check for soldering cracks, I had a failing 8643A generator failing self test, I track it down to the output amplifier than I try resoldering the main output RF transistor amplifier and be hold afterwards it works and passed self test, I guess there was a cracked join in the soldering! The HP 864x series were the first RF generator made by SMD components that HP designed, very beautiful machine!

[microbug]

The Minicircuits MAR-6SM+ looks like a promising replacement that is available from Mouser.

Thanks vaualbus, I'll check that too. I don't know if a broken U16 explains all the symptoms -- why would the PLL still lock? Maybe it is less broken than it looks when it's loaded correctly.

[vaualbus]

I think is because the lock is before the divider!
As you see the schematic block F of that page, we have the VCO output that is 3db splitted into two paths: one is than feed to U15 and go to the prescaler and than eventually yo the phase detector locking the loop, the other side is feeding U16 and than the dividers/filters and eventually generate the Frac-N module output signal.

As it can be seen, U16 that seem suspicious is on the other side of the splitter so the instrument is "deaf" of the problem. I guess they can only figure out it by having the ALC detect an invalid level of signal?   

Never mind I confused U15 with U16  U16 is feeding the lock, U15 the dividers and output.
Still dunno what happen if one that amplifiers are bad and could create a problem on the others.

[microbug]

Good news: I had another U16 (MSA-0686) on an old PCB from an HP 8922M GSM test set.

Bad news: I swapped it in and the symptoms are exactly the same. The input and output of U16 look identical to before.

The bias tee DC voltage is around 3.5V at U16, which looks about right (datasheet). The 9.1V rail measures 8.6V. I checked each resistor and both capacitors in the bias tee, and they all look ok. Inductor is a short so I am assuming it's fine.

Edit: I checked U15's output and it looks a lot like U16, i.e., seems like it might be broken. I will now have a look at U12, which is MSA-0386 that drives both U15 and U16.

[microbug]

I have found that by pressing the synth board down into its RF gasket (as it would be when assembled fully), the PLL unlock error goes away AND the U12 output switches between a single, stable peak at the correct frequency, and a mess of peaks all over the place.

Sadly, pressing the synth board down doesn't change the output of the whole instrument, which remains a fixed 2.72GHz at -3dBm. So it seems I've just solved a problem I created myself  -- and a problem I identified earlier (the PLL losing lock)

I put the synth board back in its shielding block and tightened the screws. Now I get no frac-N unlocked error and the coherent carrier output looks perfect -- one peak at the desired frequency, and some harmonics visible (which should be filtered out later on the output board). Main peak between 1-3dBm. The problem must have started when I opened the synth board RF shields.

So I'm back to suspecting the I/Q modulator chip. I'll have to poke around the output board again, and see if I can find the source of the 2.72GHz signal -- it's not coming from the synth board. Perhaps the modulator chip is self-oscillating.

Edit: screenshot of the coherent carrier output attached.

[vaualbus]

It is still strange to me that the only fault could be the I/Q modulator....
I guess looking, at the schematics, you could bypass the IQ modulator chip?
The output is AC coupled to next amplifier I am expect the chip to have some loss so if you have a 3db attenuator maybe you could insert that there to avoid demage the next stage chip;
Also I see that the IQ modulator output is differential and is converted back to a single ended signa by two transformer (T50, T51) maybe one is open?
I still do not get why the output is that magic 2.72Ghz signal at all.

[R-1125F]

You didn't mention what options you have.   Is it possible that the digital generator boards are somehow interferring with the IQ modulator?

[vaualbus]

Doesn't really make sense thought anyway?
The IQ is at IF frequency, how they ended up at a so high frequency?
If the IQ modulator is a multiplier of some sort, it would produce some spurious armonics at multiple of the carrier frequency, so why you are stuck at 2.7Ghz?
So, let's said it failed how the hell the output is stuck at 2.7Ghz, if the input signal is at 4Ghz as it was shown in the first pictures of this thread?
It is still puzzling me, either the IQ is oscillating, and we see the result of it or who knows.
What I guess is possible to do is to bypass it temporary with a coax cable, the output is DC coupled into the next stage MIMC amplifier so I guess it should "work"?
Or maybe better if you have another sig generator you can try feeding a signal after the IQ modulator and see if you get any output at all!

[microbug]

There aren't any options installed except the high-accuracy timebase (OCXO).

I took some more measurements today. I started with the coherent carrier, and noticed the 2.7GHz signal there as well as the correct signal. This made me suspicious, and I tried the same measurement with the vector output board shield properly installed. The 2.7GHz signal disappeared. If it's coupling into this part of the circuit, it could be coupling into other parts too.

I measured before the I/Q modulator, immediately after, and after the next IC (U80, a 1GG3-4205 RF amp). Screenshots attached.

The following stands out to me:

- Across the I/Q modulator, the signal at the set output frequency loses 30-35dB!
- The gain block U80 appears to be effective only at the set output frequency. The 2.7GHz signal is relatively constant on either side of the gain block.

These measurements were all taken with a piece of coax connected straight to an RF trace, so they won't be very accurate in absolute levels. The very large I/Q modulator losses do seem like a smoking gun suggesting this part needs replacing.

Unfortunately I've damaged a DC block cap again, so need to leave it here for today until I can replace it. I will try bypassing the modulator IC with some coax and a 10dB pad when I've replaced the cap.

[microbug]

I tried bypassing the mixer chip with a 10dB pad. Instrument output looks exactly the same -- 2.7GHz signal. I also tried running it with the shield on the output board (mixer chip not bypassed), and it shifted to 1.1GHz (but still no sign of the signal it should be putting out).

I think I might have had a faulty connection in the measurements labelled 2 in the last post. That would suggest that there might be a dead component somewhere later in the RF signal path. If the desired signal isn't getting through, and there's a lot of 2.7GHz around, then I suppose the input to the next stage might just be coupled 2.7GHz, leading to that appearing at the final RF output.

I'll do some more probing around this weekend. I'm starting to get a bit fed up with this. The number of parts to check is quite small though -- since the problem is there for the whole instrument output range, it must be a part of the RF chain that is shared. That leaves just a few amplifiers and some PIN diodes to check. If I don't make some progress soon I will consider selling the unit for someone else to have a look at.

[vaualbus]

Interesting thought, next I could bypass the switched filters bank? Like remove C140 and inject a signal there?
But I guess having excluded the IQ is in some regard a good idea.
But yeah, this is getting a very weird failure. I would guess that you have already checked all the power supplies (I mean also the secondary one in page 5/5 to the output scheamatics)

[vaualbus]

Guess by the lack of response that you give up at the end on reparing this? 

[microbug]

Not quite. I am low on motivation but I’ll try a bit more, it’s not satisfying to leave it unfinished.

I left my job yesterday (starting a new one in a month’s time) and so don’t have access to their spectrum analyser any more. I have ordered a TinySA Ultra but it’s not here yet, not much I can do until then.

[microbug]

Well that weekend came and went. I now have a TinySA Ultra, which I'm very pleased with.

I decided to try using a freezer spray on the vector output board while in operation. I tried freezing the mixer chip and no change (still a small 2.7GHz peak). I tried freezing U80 (the RF amp after it) and the peak disappeared. Interesting. So I tried measuring at the mixer output and found the 4GHz signal I expected. The 2.7GHz signal was also there, but I think it was just coupled from somewhere else.

I removed the baluns between the mixer and U80 and tested them. They are fine, at least at DC. No damage visible under magnification.

I then tried measuring the input to U80 (disconnected from U80 -- straight into the spectrum analyser). What do you know, the signal I wanted! Also, U80's output is 6 Ohms to ground. That seems pretty suspicious, there aren't any external DC bias components on that pin.

So I suspect U80 (which is the same part that caused the problem in the Signal Path video, although not sure if it is the same device inside the Agilent package). I think I can try bypassing it with some coax, it will be a bit of a nasty job but should prove the principle.

Fingers crossed for this. There have been a lot of red herrings on this repair, some definitely because of dodgy coax connections. Others because of that 2.7GHz signal coupling everywhere. It's all a learning experience...

Edit: it seems unlikely that U80 is the same MSA-0686 as on the board revision shown in the TSP video. The IC on my board has +9V and -5.3V power inputs, the MSA-0686 has a single-ended power input.

[vaualbus]

Cool finally an update!
Wow good find yeah this repair was "cursed" we always believed about the IQ modulator but at end it seems this due to the next component in the chain
So yeah 6ohm is way too low, I guess. Those are custom Agilent parts, I think they use the same parts, as I said to you before, in different instruments of the same period.
This must be a bad design from Agilent as a lot of people have trouble with this 1GG43-4205 https://groups.io/g/HP-Agilent-Keysight-equipment/topic/esg3000a_e4421a_signal/86763436?p=
I am glad it is not (yet) failed in my unit I would say
I think later revision boards use more off the shelf components instead of this hybrid IC.

By the way U140/U180 are the same IC, so you could try remove it tested it and see if it has the same 6ohm.
Maybe swap them and see if the error is now more downstream.

Also are you enjoying the TinySA?
I also looking into buy a proper SA eventually.

[microbug]

Fingers crossed it is that. I'm now moving house and then going on holiday so that will probably be another long delay

Yes I like the TinySA. It's a bit slow to sweep (around 5 seconds for 50MHz-4.1GHz) but it's quite usable. The touchscreen is surprisingly fast and easy to use compared to a 'real' SA. Built-in storage (microSD slot) is nice, and it comes with a 32GB card.