Marconi 2019A Intermittent RF Output Fault

[Ti Pwun]

I bought this sig gen from another member here and overall it's been an excellent, very reliable unit. Recently, though, it developed an intermittent fault.

Because I didn't need to do anything to it when I got it, I have only just looked into doing any work on it... After removing the top and bottom covers, it's a tad daunting, I have to say. Doable for me, no doubt, but I want to be cautious.

The problem is that there is no RF output (intermittently, but more and more frequently). In fact, while writing this post it has suddenly started outputting the 600KHz I programmed over an hour ago. I hear it in the radio speaker.

I have been at this point before but when changing to 1.5MHz for the next alignment procedure, there was no output and yesterday I was unable to set it to 1.5MHz at all. All looks good on the displays but no output/

Everything else seems to work properly so I'm thinking perhaps a dry joint somewhere, or maybe a failing transistor.

I'm planning to study the manual fully today and then do some testing but before I do, I'm wondering if there is a common issue someone is aware of that may point me in the right direction.

Unlikely, I know. because there could be any number of problems, but it's worth asking. So, if anyone has encountered intermittent output problems concerning the Marconi 2019A, your suggestions or thoughts would be appreciated before I start digging into this thing.

TIA,

Jon.

[trh01uk]

Jon,

I have one of these too. Unreliability is built in probably because of the interboard connectors. In mine it shows up as the display doing odd things and buttons not working, particularly around the modulation frequency selection. Consider that it might be that the front panel board is sending false "output on/off" signals to the RF system.

I've not cured mine totally, but stripping down the front panel and cleaning everything has helped a lot.

You might also want to ask about this on the Marconi Test Instruments email list on Yahoo

Richard

[trh01uk]

Jon,

you may be able to pin-point the problem by checking to see just how much RF is coming out when the output fails. Is it truly "OFF"? In other words is the level nothing at all, as it would be if the output is switched off? Or is it some low level?

If its the latter I would consider the output attenuator and its control system.

Richard

[G0HZU_JMR]

I've owned a couple of these generators for about 20 years and they were both quite unreliable for the first 5-10 years I owned them.

After numerous stripdowns to fix intermittent problems with phase noise and signal level problems plus the odd component failure mine have both been pretty good for the last 10 years.

The only issue I have with one of them is that the output attenuator can misbehave if the sig gen isn't used much and this means it doesn't give the correct output level unless the attenuator is cycled a few times. So try setting the generator to 10dB step sizes using 'Delta' 'RF Level' '10' 'dB' and then use the up/down keys to cycle it through its complete range a few times. The mechanical attenuator should give a clonk noise every 10dB change in level and I'd recommend you run it up and down from about -120dBm to +10dBm several times in 10dB steps.

These generators aren't that reliable in my experience and we had several of them at my place of work in the 1990s and they were all scrapped eventually.

However, I do like the 2019 because it is so easy to use and the phase noise performance is quite good across the shortwave bands. Also they don't use cooling fans so they are silent in operation. So it's definitely worth trying to repair it.

Common faults/failures include the internal 10MHz OCXO, the VCOs can fail, various caps can go high ESR and the signal interconnection system isn't very reliable. There is also an OM345 device used in these generators and these little amp modules are known to be unreliable. Especially in the little Marconi 2022 where there are quite a few of these OM345 devices.

[Ti Pwun]

Great info guys - many thanks.

This one has been spot on since I got it and I love using it. I've been very impressed with it. But it isn't used extensively, I'll admit that, so I will certainly give it plenty of work to do over the coming days.

I did go around checking cables to make sure everything is seated properly and after testing it seems to be working properly (typical - I hate intermittent problems) so I'll get on with the work I'm in the middle of while it's behaving and then I'll cycle through lots of settings.

The comment about lack of use definitely struck a chord, though, because other than some radio alignment work, it doesn't see much action and I don't leave it on for longer than an hour or so warm up time plus however long it takes to do the job so I ought to leave it on for much longer for a while as I 'give it a workout'.

I will also clean the front panel thoroughly as well - every press registers and I get what I'm expecting on screen but it certainly can't harm to do that.

Thanks again - I'm pretty ill at the moment but I'll post an update in a few days one way or the other. I was surprised to hear that there are reliability issues with this unit but on looking inside, I can see what that would be. Maybe I've just been lucky that it's been super reliable until recently. Hopefully a workout will be all it needs...

[Ti Pwun]

OK, I found some dry joints on the power supply board so I decided to replace the capacitors now, rather than just resolder them - I didn't like the look of a couple of them and even though those ones tested OK, others I removed didn't. I also cleaned the front panel and went through the various attenuation settings several times.

Anyway, I've been working with it all day and it is working perfectly again. I'm sure resoldering the dry joints alone would have been enough to sort it out because prodding on that board caused the display to act up, but I had the caps and the board was off so I replaced them anyway.

Some of the buttons are more responsive as well so cleaning everything on the front panel was worthwhile.

Thanks again for the input - I love this unit so I am glad it was a something and nothing repair in the end. Although after studying the manual it isn't as daunting a prospect as I had first thought if I do ever need to delve deeper into it.

[trh01uk]

Before you go, this is all useful stuff for those with these generators. Sounds like you had a power supply fault due to dry joints and/or faulty caps. Did you actually see any symptoms in terms of varying supply rail voltages?


Richard

[Ti Pwun]

Unfortunately not, but there must have been something like that going on. After pressing on the board the unit reset itself and the displays came back dim (at the default settings).

At that point I lifted the board to inspect it and I could see straight away that some of the connections were dodgy (visible separation between component lead and pad). Likely a heat issue, I would say.

After repairs, though, it was rock solid and pressing at various places on the PCB didn't have any effect so I ran it all day with the lid off and it's been fine since. I therefore didn't get as far as establishing voltages. With the lid on it's still stable so I'm calling it done...

...until something else fails, I guess. I may consider installing a couple of small quiet fans behind the vent slots at some point; I don't need any heating on in the room when this thing is running, that's for sure, so it wouldn't hurt to vent a little bit of that heat!

[trh01uk]

Fine, yes, you describe the symptoms of voltage loss with the displays coming back on dim.

I am not sure that having no fans is a "good thing". In a multimodule unit like the 2019 heat can build up locally, and have nowhere to go. In theory many items are fine with high temperatures, but I suspect that continually cycling between room temp and some higher temperature means repeated heating-cooling cycles which can damage parts. And of course, the way that you and I use our 2019s - with occasional use only - means that they thoroughly cool down between times. In many professional labs, gear like this is left on 24/7 to avoid this problem, but that's a bit costly for those of us paying our own electricity bills!


Richard

[Ti Pwun]

Yep, the Mrs would have a fit if I left it on 24/7 and I'd be uncomfortably hot in this room - when I left it on all day I had to remove the filter from my solder fume extractor and spin it around to get cool air blowing at me.

As you say, though, cycling between cool and hot cannot be good for a unit like this. If I need to remove the lid again I'll think more about cooling if I can do it without hacking holes in the case - even with the slots I can see that it's mostly blocked by the innards so even a thin fan isn't going to help much. Adequate venting clearly wasn't considered when it was designed.

[Ti Pwun]

Well, the fault is back. I was doing some FM work at 90-100MHz yesterday (after assuming the 600KHz/1500KHz problem was solved/working) and it worked flawlessly but I went back to 600KHz on AM today and after about a minute, the signal disappeared. Looks like I have some troubleshooting to do.

I'll start with voltages in this state and then when I have an output (assuming I can get one) and see where that leads.

[Herald1360]

Ahhh..... it's a proper intermittent then, complete with red herrings, wild geese and probably mare's nests

It'll likely need some unobtanium to fix it as well.

[Ti Pwun]

Ha! It's giving me a headache, that's for sure.

I left it on 600KHz while I study the manual some more but I connected the output to my Racal Dana 1998 - it was fluctuating between about 95KHz and 107KHz (all different seemingly random frequencies between that range) and it was doing that for a few minutes and now it isn't displaying any output at all.

I imagine I'm going to have fun with this...

[Ti Pwun]

Just an observation as I check things out: I can go down to 2.1MHz (was set to FM) and an accurate frequency is displayed on the RD 1998. But if I drop that to 2MHz on the 2019A, the frequency doesn't change on the RD1998.

It seems that any frequency below 2.1MHz is where the intermittent fault lies...

[Radio Wrangler]

Sounds like the synth going out of lock at one end of the VCO range? does it have missing patches at higher frequencies?

First check the power supplies. Then check that the frequency reference oscillator is going. Then check that the main VCO can be tuned across its full specified range.

The manual should have fault-finding trees which will guide you

David

[G0HZU_JMR]

Originally Posted by Ti Pwun

Just an observation as I check things out: I can go down to 2.1MHz (was set to FM) and an accurate frequency is displayed on the RD 1998. But if I drop that to 2MHz on the 2019A, the frequency doesn't change on the RD1998.

It seems that any frequency below 2.1MHz is where the intermittent fault lies...

The design integrity of the 2019 is compromised for frequencies below about 2MHz because the signal generator resorts to using a mixer/downconverter to synthesise signals below 2MHz.

The frequency plan for this generator is loosely based on the classic HP8640 sig gen where the generator can generate a signal across an octave range of about 260-520MHz using a bank of four VCOs (rather than a single cavity osc in the HP8640)

But the basic idea is that if you can generate a signal across an octave range up at UHF then you can generate every frequency 'below' this range using frequency dividers and a bank of switched filters to remove the harmonics from the dividers.

On the HP8640 this division is used to take the signal right down to about 500kHz and the benefit of the division is you get a 6dB improvement in close in phase noise for every divide by two.

So by the time you get down to maybe 2MHz the phase noise is very low.

But the Marconi 2019 doesn't use 'just' division for frequencies below 2MHz.

In the interests of cost (and also the desire to operate right down to about 80kHz) the 2019 generates 2MHz by dividing one of the UHF VCOs down to 12MHz and then it gets mixed with the 10MHz OCXO to get 2MHz. If it tried to do this using dividers then you would need a whole load more dividers and filters and the cost/complexity of the generator would go up a lot.

The penalty for this is the phase noise will not be as good as it could be for signals below 2MHz and also the spectral purity of the generator is slightly compromised on the 80kHz to 2MHz range.

So maybe it's worth checking to see if the generator 'works' when you dial in front panel frequencies of 10.08MHz to 12MHz because this is the range of frequencies it has to synthesise to generate 80kHz to 2MHz on this lower range. In other words, you may find it is also dead across 10.08MHz to 12MHz if there is a PLL based issue.

It could be that one of the four UHF VCOs has failed (intermittently) or maybe there's something wrong in the mixer/downconverter circuit that is used below 2MHz or maybe there's an intermittent connection in the associated switching network.

[Ti Pwun]

Thanks guys - I have the AB3 board on the workbench now (oscillators 1-4) - it only had one screw in the centre so with the other eight screws missing, I think the board needs to be tested given that the problem could be there.

It's all one groundplane so missing screws are not an issue, obviously, but it does mean the board has been worked on, as can be seen by the soldering.

I'll go back to checking voltages after I've tested and refitted this board aqnd made sure there isn't a bulging cap lurking somewhere.

[G0HZU_JMR]

I think you can work out the frequency plan for, say, a front panel frequency of1MHz as below.

The VCO divider chain needs to produce 11MHz at its output in order to mix with 10MHz to get 1MHz. This will presumably mean that VCO no2 (TR2 on AB3) will be selected and it will be locked at 352MHz because 11 x 32 = 352MHz and this lies inside the 309-368MHz range of VCO no2.

So VCO no2 will be running at 352MHz. This will get divided down by 32 to get 11MHz. This 11MHz gets mixed with the 10MHz OCXO to give 1MHz at the sig gen output if the generator is healthy.

I think VCO no2 will also be at 352MHz when you dial in a front panel frequency of 11MHz.

[Ti Pwun]

Thanks, I'm slowly getting my head around the way things are and should be working - I'm delving into it again tonight so I'll see how I get on. I need to rig up a cable so I can power it up with the lid open but I dug out a couple of connectors so I can do that easily enough.

[Ti Pwun]

Update on this...

All the lower frequencies are present again. Very tricky repair but I found the problem; on OSC 2 the sliding adjuster (piece of wire soldered across two tracks) had a failing connection. I initially had the 8v on both sides of the wire but because there was only a central screw holding the AB3 board down, touching the track was enough to make the connection.

Luckily I noticed that sometimes the 600MHz I was trying to get was present when I probed the board. I located the problem area by flexing the board ever so slightly on both sides of the tracks for Osc 2 and bingo. It was a nice shiny solder connection but it was flakey.

After resoldering that connection, all is well. Really glad because I use this unit regularly and I've had a tough time without it.

Thanks again for all the comments - this helped me to focus on the right area. Shame I hadn't discovered this intermittent before I replaced all the SMD caps and resistors but it's working now so it was a small price to pay.

Needless to say, I also located some screws to hold down the AB3 board properly.