RF signal generator - Marconi 2022?

[WaveyDipole]

I am considering to replace my existing three signal generator units with one RF signal generator covering everything from LW to FM and beyond. Both AM and FM modulation are naturally required. I also need something fairly compact.

I was thinking of acquiring a Marconi 2022, but does anyone know how much difference there is between the various models, A to E? There is a 2022A on eBay right now, but I have seen a 2022D go on auction for rather less than the asking price of £485.

I was offered an Agilent E4431B for a very attractive price, but although coverage goes up to 2GHz, its starts at 250kHz so does not cover LW. The bigger problem was its physical size which is huge and I could not accommodate it, so ended up having to turn it down. Bit of a shame really because it otherwise seemed a very nice unit.

Is there anything else that is reasonably compact and ranges up to say 500MHz or 1GHz that I should consider?

[G0HZU_JMR]

The little Marconi 2022 sig gens are really popular and I think this is why they seem (to me) to be way overpriced. I think the early ones had lower (max) output power but I'm a bit rusty on this now. At work we used to have loads of them from the A version to the E in either brown or white. I think the low power versions top out at +7dBm and the later ones can deliver +13dBm.

I can't remember what the other differences are apart from the lower power level of the early ones. Some versions were for certain markets and that might explain the lettering. GPIB is a nice option to have fitted if you ever want to control it remotely.

The 2022 does have a few reliability niggles to do with the little 5 pin OM345 amplifier modules it uses in various parts of the signal path. There are quite a few of them in a 2022 and it can mean that you have to check the generator out quite carefully to make sure it works on every sub range within LF to 1GHz. The 2022 is fan cooled and I would advise you always operate it such that it can get a nice flow of air at the back, especially in summer. Otherwise I think it will be less reliable over time. This might be what kills the OM345 modules. At work, OM345 failure was rare across maybe a dozen generators and about 25 years and this may be because the generator would typically be used on an uncluttered bench with plenty of free space around it.

Also, the little front panel buttons of the 2022 can become a bit vague to operate and everything has to be done via buttons as the 2022 has no rotary controls. If you do buy a 2022 I'd recommend you keep at least one of your old generators in case you want to do any old school frequency wiggling via a rotary VFO control.

The Agilent E443x generators are very versatile, especially the vector/IQ versions. I've got a couple here and although the front panel states 250kHz to 4GHz the lower frequency limit is actually 100kHz. This may mean that some versions of firmware allow extended operation down to 100kHz but maybe it won't be as good on spurious specs down below 250kHz. Both of mine work fine at 100kHz. The phase noise across the HF bands is quite poor though.

[Andrew2]

In addition to what Jeremy says, I'd watch out for dodgy displays on these. Segments can go low contrast or missing and it seems to get more common with age. A nice little generator though, even if it doesn't have 'proper knobs'.

[G0HZU_JMR]

I did a quick google over lunch and I spotted a few minor differences. It looks like the performance of the FM modulator has been improved on the later versions in terms of the max deviation available. I've listed the deviation for the 88-108MHz band although the deviation can go much higher than this on the higher frequency ranges.

The early 2022 version covers 10 kHz to 1000 MHz and the max power level is +6dBm. The max FM deviation on the 88-108MHz band is 99.9kHz

The C version covers 10 kHz to 1000 MHz and the max power level is +13dBm. The max FM deviation on the 88-108MHz band is 99.9kHz

The D version covers 10 kHz to 1000 MHz and the max power level is +13dBm. The max FM deviation on the 88-108MHz band is 125kHz

The E version covers 10 kHz to 1010 MHz and the max power level is +10dBm. The max FM deviation on the 88-108MHz band is 125kHz

I have a 2022C version here and I could do some tests on it on FM if you are concerned about the modulator performance?

As Andy says, the LCD display can go a bit vague and shimmery. Luckily, the one I have here is in good condition with a really good LCD display. I've seen a few at work suffer with the display contrast issue although all of them remained usable.

[Radio Wrangler]

Originally Posted by WaveyDipole

I was offered an Agilent E4431B for a very attractive price, but although coverage goes up to 2GHz, its starts at 250kHz so does not cover LW. The bigger problem was its physical size which is huge and I could not accommodate it, so ended up having to turn it down. Bit of a shame really because it otherwise seemed a very nice unit.

That sound you heard to the North, was my jaw dropping. That's the fancy digital modulation version of the ESG. They are rather valuable. If it's got Dave Park's dual arbitrary waveform generator option board in it, it's unbelievably versatile. It's the I/Q modulator module that limits the low frequency coverage.

Jeremy could be getting a little excited, but he likely already has one

Edit... ah, he has two!

David

[WaveyDipole]

Thank you for the heads-up on the display issue. I will check any prospective purchase carefully.

Originally Posted by Radio Wrangler

That sound you heard to the North, was my jaw dropping. That's the fancy digital modulation version of the ESG. They are rather valuable. If it's got Dave Park's dual arbitrary waveform generator option board in it, it's unbelievably versatile. It's the I/Q modulator module that limits the low frequency coverage.

Jeremy could be getting a little excited, but he likely already has one

Edit... ah, he has two!

David

I was offered it at well below what they go on eBay but there were a number of factors: it would have been a bit of a stretch financially; I am not keen on doing off eBay deals; it was overkill for what I really need. Overall it seemed rather too good to be true, but I was sorely tempted to take it.

The trouble is that the thing is huge and has a significantly bigger footprint than even my PM3094 CRT oscilloscope which I plan to replace with a more modern smaller unit. I just don't have shelves that go that deep or that kind of desk space. I expect it is designed to be rack mounted - a lot of them seem to be. Otherwise I would have taken it and perhaps kept the RF1U for LW alignment. As it is, it was offered and did sell on eBay quite quickly. Not sure whether it had the Dave Park's arbitrary waveform generator option with it. Didn't notice anything like that in the specs.

There were useful outputs on the back including IQ I think, but I had virtually no chance of getting cables in the back without risk of them getting squashed/damaged. It just was not a practical solution for my environment.

BTW, the 2022A currently on offer on eBay has the connectors on the back and although I'm not sure how much trouble would it be to move them to the front, this is one reason why I am hesitant to put in an offer and looking at other possible options.

[G0HZU_JMR]

Stewart of Reading have a 2022E in their special offer list at £332+VAT.

I think it has been there for a long time or maybe they have a few of them. I wouldn't pay that much for a 2022 sig gen but it does seem to be slightly lower than the current going rate.

In the early 1990s there used to be a trader at various amateur radio rallies who always seemed to have a Marconi 2022 on offer on his stall at £450. I suspect it was the same one year after year. Here we are 25 years later and the prices for a 2022 (in good condition) seem to be about the same. On US ebay the prices seem to be about double this. Crazy.

[PJL]

The push buttons are still available, quite a few of mine are poorly. I bought these and replacement electrolytics but have never got around to fitting them as it still 'works' and best to let sleeping dogs lie.

[G0HZU_JMR]

Probably the best advice I can offer is to be patient as it looks like you already have a few older/larger sig gens that you can get by with for now.

None of my sig gen purchases were ever planned, I just kept an eye out for bargains. Test equipment rental companies sometimes have clearout sales and traders sometimes need the cashflow. Otherwise it's a case of waiting to find something on offer from someone looking to offload a surplus generator. If it has a minor fault then the price tends to drop a lot.

I paid £25 for my 2022C for example. It did require a minor repair as one of the OM345 amps had failed. I fudged the repair with an external bias resistor across the OM345 and the cost of the repair was about 1p for the resistor. I keep meaning to swap out the OM345 for a healthy one but never seem to get round to it. The generator works fine with the temporary bias resistor fitted.

[Wendymott]

I had two 2022's for a while.... both decided to go faulty OM345's at one time or another, MAR6 mmics modules cured the issues. However one of them.. I think it was the 2022 C had a Matrix button issue and needed "double press" to persuade it to operate. Cleaning was not successful. Its a great general purpose generator though. I sold my two for an Anritsu, which required a restucturing of the shelf it sits on... and the Anritsu fan is quite noisy,, If you can get a 2022 for £25 as Jeremy did..snap their hands off.

[G0HZU_JMR]

I hope this isn't too nerdy but see below for an old phase noise plot of one of the company 2022 sig gens at an output frequency of 14MHz. This plot was taken back in 2012 and I suspect that this old 2022 generator was already suffering some degradation in spurious terms close to carrier. However, it still looks acceptable to me.

To generate frequencies across an output frequency range of 10kHz to about 62MHz the Marconi 2022 uses a downconverter using a 160MHz LO. To generate this 14MHz signal I assume it runs VCO #1 at 348MHz and this gets divided by 2 to get 174MHz. This then gets mixed down to 14MHz using a fixed internal 160MHz LO.

The phase noise performance of the plot below is quite good considering the VCO #1 tunes 250MHz to 353MHz. If I throw some numbers into a VCO design spreadsheet it shows that the VCO designer did very well to get that phase noise at offsets above 2kHz. It's close to theory for a 10mW oscillator for example.

However, the main reason I'm posting up the plot below is that you may find that a typical 2022 sig gen won't be this clean today. There are quite a few aluminium electrolytic caps in the VCO/synthesiser area and these seem to have a marginal voltage rating in the circuit. Because of this and because this generator runs quite warm internally I think these caps may begin to degrade in terms of ESR over time.

To compare results I had a look at my 2022C and there are a lot more spurious terms on a 14MHz signal compared to the plot below. Mine is much worse for spurious terms on the VCO #2 range so I suspect I really should be changing some caps in mine. For example, my 2022C looks much worse on 17MHz than it does on 14MHz and I think it swaps across to the VCO #2 somewhere inbetween these two frequencies.

It could be that most users will be unaware of any ageing effects on the spectral purity but a lot depends on what you expect from a generator like this.

Also, if you try and use it to tune in small steps of (say) 100Hz then you can expect to see a lot of close to carrier sidebands appearing within about 1kHz of the centre frequency. This is a limitation of the synthesiser it uses. These spurious terms would probably only be an issue if you wanted (expected?) to use the Marconi 2022 as a clean NBFM signal source and you wanted to test the spectral purity of a NBFM demodulator. I think the 2022 would be the weak link here especially if it was being tuned in 100Hz step sizes.

If you keep it tuning in larger step sizes of 1kHz it is much less of an issue. This is partly why I don't recommend the 2022 if you like 'wiggling' the frequency either side of the desired output. For one thing it quickly becomes tedious to use the buttons like this and the other reason is that the spurious performance degrades when you start tuning in fine step sizes of 10Hz or 100Hz.

[G0HZU_JMR]

See below for a plot of the predicted phase noise of the VCO #1 in the 2022 after a divide by two to 174MHz. This was created by a simple excel spreadsheet and I added the thick blue lines to show the impact of the frequency divider and the PLL cleanup within the loop bandwidth of the main PLL.

Across the 10kHz to 62MHz range the phase noise of a healthy 2022 generator should be fairly similar to the output of the spreadsheet below or to the phase noise plot of a real Marconi 2022 also shown below. The PLL cleans up the phase noise below about 150Hz and the phase noise floor will be limited to about -150dBc/Hz by the frequency divider and also the rest of the signal path in the 2022 sig gen.

The spreadsheet assumes a 10mW oscillator, with loaded Q of 20 and an 8dB degradation due to the noise factor of the transistor any any compression effects. I guessed a flicker corner of 5kHz.

This phase noise performance is better than the ESGD E443x generators on the HF bands but much, much worse than signal generators designed to have low phase noise on the HF bands. It could be that your existing old sig gens have better close to carrier phase noise than the 2022 for example.

[WaveyDipole]

£332 is £398.4 with the VAT added on and then there is carriage on top. Still, it would work out almost hundred pound cheaper than the 2022A on eBay. How on Earth did Jeremy get one for £25! I would definitely snap someone's hands off for one at that price! The last working 2022D I saw on eBay was sold via auction for £440! I have found that these days even "spares and repairs" stuff often goes for close to the price of a working unit.

I did ask for alternatives and Jeremy's suggestion to be patient get by with what I have for now might the best option for now. I have a search set up on eBay so anything that crops up should be automatically flagged in my mailbox.

[G0HZU_JMR]

I have given it some thought but it's difficult to suggest alternatives because of the requirement for small size. The Marconi 2023 or 2024 is a really nice signal generator but it will typically cost about £800 and it has about the same footprint as the E4431 generator. It is only half the height and is a fair bit lighter though.

The classic Marconi 2018/2019 is the same size and weight as the E4431 but often sells for much less than a 2022. The other popular classic is the HP8640B but these are huge and extremely heavy. Even I've avoided buying an HP8640B because of the size and weight.

The 2022 is fairly unique because of its small size. There are modern compact alternatives from the far east such as Rigol but these will be very expensive to buy new.

One other concern with the 2022 is the impact on performance when the OM345 amplifier modules fail and get replaced with low cost MMIC amplifiers. A typical MMIC won't have the same large signal performance and it won't have the same I/O impedance as the OM345 and the reverse isolation will probably be worse as well.

I keep meaning to replace the MMICs in the 2022C I have here. I think the previous owner fitted them. The problem is that the OM345 has been obsolete for many years so they are quite rare these days. So the only alternatives are to buy a dodgy MMIC based module on ebay or to try and make a decent copy of the original OM345 or to try and repair the existing OM345.

I suspect that part of the reason my 2022C isn't working quite as well as the unmolested ones at work is because of the cheap MMIC modules that have been fitted to it in place of the OM345.

[G0HZU_JMR]

How on Earth did Jeremy get one for £25!

I bought it from a forum member a while back. I had to drive a fair distance to collect it though. I think he got fed up with the OM345 module failure issue and he had replaced several of them already.

[G0HZU_JMR]

If you don't mind having a non-synthesised generator then the little Marconi TF2016 covers 10kHz to 120MHz and it has a decent ALC and attenuator system. It would have to be used with a frequency counter though.

It is possible to buy a synchroniser to make it act like a synthesised sig gen but this module is the same size as the 2016 and it is quite clunky to use. I think I'd rather use the TF2016 on its own with a frequency counter as I would expect it to be quite stable once warmed up.

The little TF2016 should sell for under £100 although test gear prices seem to have gone up a lot recently.

[Radio Wrangler]

Originally Posted by G0HZU_JMR

The other popular classic is the HP8640B but these are huge and extremely heavy. Even I've avoided buying an HP8640B because of the size and weight.

I'd steer you well away from an HP 8640. They'r 19 inch rack width and about as tall and deep as the other full width ones. They are full of cast aluminium housings, but the total weight will make you swear lead is involved. They have great modulation versatility and legendary low phase noise.but have a few problems:

1) The legendary status reflects in the price despite the other problems.

2) There are brittle plastics in gears and switches. Maintenance is now a nightmare. See other threads on this forum.

3) There are custom RF microcircuits (sapphire substrates!) inside as well as custom ECL hich speed divider chips.

4) The output microcircuit is easily blown up by some fool transmitting into the sig gen when he was supposed to be testing the receiver!

They become a major epic in themselves to get one going. But the low noise is special.

HP brought out the 8645 to try to reproduce the low noise behaviour with a synthesiser. You won't believe how much heavier they are. 'Two person lift' labels are common additions.

They then brought out the HP 8648 family as a cheap sig gen with no pretence at high performance. Small, neat, light, basic, cheap (for HP). Their big selling point was having TWO knobs on the front panel, one for frequency, one for amplitude!

David

[G0HZU_JMR]

The weekend is here now so I've got some free time to look at my 2022C. I think I've had it for about three years but it only gets used when I need a basic sig gen on another work desk. For tasks like this it is ideal because it is so small and light. However, maybe it's time to try and restore some of the original performance as it does appear to have a lot of spurious responses on the main carrier on some frequency ranges.

I also had a look at the FM and AM modulation performance and below 62.5MHz the AM modulation is really grim when viewed on a spectrum analyser. I think this is the first time I've tried using AM with this generator. Something is definitely wrong and it could be caused by one of the MMIC repairs. A typical 12dB gain MMIC will have less reverse isolation than the OM345 so this could be part of the reason the AM appears to have a lot of FM on the signal.

If I find anything interesting I'll start a thread about it. If it is caused by a dodgy OM345 >> MMIC repair them maybe lots of 'repaired' 2022s will have similar issues with AM below 62.5MHz.

[G0HZU_JMR]

I spent some time this evening playing inside the 2022C and found out the main reason the spurious were so poor on VCO #2.

It wasn't what I was expecting and it appears that the (triple) VCO lid needs to have a very good electrical fit to the main board or the VCO can pick up interference from other parts of the sig gen. Just by playing with the screw torque I could send it crazy on the VCO #2 range.

The main cause appears to be a large solder splash on the gold screen tracking on the main PCB next to VCO #2. In the old image below you can see that someone in the past looks to have replaced the VCO #2 transistor and they managed to get solder all over the gold layer on the PCB (circled in yellow). This solder splash was actually quite proud compared to the gold layer when viewed under a microscope and I think this was partly why VCO #2 was so sensitive to pickup. It prevented the metal lid from achieving a flush seal to the main PCB.

It took over an hour to carefully remove the soft layer of solder from the PCB and I did this under a lab stereo microscope. See the 'after' image below where there is just the gold layer left intact. All the solder is gone!

I also cleaned the metalwork face and the gold layer on the PCB until it was like a mirror. There did seem to be a thin layer of oxide on both faces. The result now is that the spurious terms have all but vanished and the phase noise is now nearly as good as the old works 2022E. See the phase noise plot below. It manages just under -120dBc/Hz at a 10kHz offset on a 14MHz carrier. This phase noise performanceisn't that great for a lab sig gen on the HF bands but I think this is what I'd expect to see from a 2022 because of the way the main signal path is designed for signals across the 10kHz to 62.5MHz range.

The performance on AM is a lot better now as well. There is still some residual FM on a 30% AM signal at 17MHz but it is about 200Hz of FM rather than 2kHz.

[Radio Wrangler]

A lumpy blob there would lift the whole cover and create a slot. Someone was slapdash.

David