Front panel layout revisions

[synx508]

At yesterday's Rugby radio rally I sold my Solartron 7150 and 7150plus. This pair looked similar but a few of the keys on the keyboard performed different functions and this was a source of mild irritation. Far worse, in my opinion, is the difference between Marconi's 2019 and 2019A. Crimes include moving the Store and Recall buttons from the black right-hand stack of buttons to the left orange buttons, adding the AF Level control (which, to be fair, is a new function) and changing the On/Off feature from being context sensitive on the 2019 to being independent of context on the 2019A. The last one would be fine, but other features are still context dependent on the 2019A, so you have this paradigm split between immediate control and context based control (all entry and up/down stepping is context based). I actually prefer the 2019A's immediate control but would've preferred it if they had changed all the controls to be immediate.

https://www.flickr.com/photos/synx508/53933384880/

[tggzzz]

A useful illustration that it isn't only operating systems that suffer cosmetic downgrades.

E.g. menus that change with use and concealing options you might need to know, flattie GUI (abandoned early 90s), dark GUI (abandoned early 80s).

Bah humbug,

[synx508]

Marconi definitely needed an extra button but not four. It turns a clean design into something a bit more confused. Four more buttons to add the ability to set the AF level target voltage seems like it should've set alarm bells ringing but if you think about aesthetics, symmetry, beauty, the grouping of logical functions, it makes some sense - or made sense to someone with influence and power over decision-making. It comes unstuck with store and recall, however you look at it, these are commands for recalling state of the whole instrument rather than toggles so perhaps they shouldn't be with the ON/OFF and INT/EXT, but that's no reason to place them with the context selection buttons. I'd have put them under the blue second function button, I think, maybe with a blank space, so they align to the bottom of the grid. They do something important that isn't related to anything parameter, they deserve their own grouping.

Edit: It's actually two new functions, the 2019A adds Phase Modulation too. That's still only two buttons needed.

I don't think HP would've done it the Marconi way, but maybe you know of an example.

[Zenith]

I can see it would irritating if you had more than one, because you get to the stage where you want to use these things by muscle memory.

I have one Marconi 2019. If I need an RF sig gen, that's the one use. The frequency reference is a bit naff, and one of my several projects, is to replace it with one based on one of the excellent Aliexpress ex cell tower units.

It's hard to say how these decisions come about. It could be that they had extra functions to add and just didn't think about compatibility with the previous model, or they did it that way because it saved a couple of quid, or the 2019A had a new project leader, who wanted to stamp his mark on it.

[mansaxel]

If any of you lot with disliked 2019s want to part with them at a price consummate with your dislike I'm probably game.

[tggzzz]

If any of you lot with disliked 2019s want to part with them at a price consummate with your dislike I'm probably game.

And that is an elegant "put up or shut up" retort! I like it

[Zenith]

If any of you lot with disliked 2019s want to part with them at a price consummate with your dislike I'm probably game.

I don't dislike my 2019, I think it's great. The frequency standard takes ages to stabilise and then to somewhere within 2 or 3 ppm, but for most things it doesn't matter. It would nice if it was quicker settling and closer to spot on, but as I said, one of the very cheap and very good Aliexpress used OCXOs could probably be adapted.

It was synx508 who was expressing dissatisfaction, but he's living life large with both a 2019 AND a 2019A. You can expect such people to be discerning.

[synx508]

I'd consider getting rid of the Marconi generators if something better came along. I should've bought a second 2019 or 2019A so I had two the same. I had the opportunity at last year's Wiltshire rally, too, but I'd just bought a HP 3312A and a RACAL counter (the 9916 that I sold this year) and didn't want to seem greedy. That 2019 was only £50 and was snapped up quickly.

Both Marconi generators are using an external reference (Droitwich), which will be a problem soon, I'll probably switch to one of my HP OCXOs after it goes off. I had intended to build a GPSDO using some nice OCXOs that I bought last year but that hasn't happened yet.

The 2019A was £100 from one of the scrappies at a McMichael rally (I think in 2019, coincidentally) but had a very poorly oscillator block, seemingly affected by contaminants ruining the dielectric properties of its PCB, a problem because it uses stripline inductors. I sourced a replacement osc block for a reasonable price from the Israel TE breakers on ebay and this worked properly, then spent a while repopulating the EAROM calibration data using my HP 8901A and HP 5335A. It didn't take as long as I was expecting.
In around 2021 the 2019A developed an intermittent power problem which turned out to be a dry joint on the RF amplifier board, apparently a common fault on these. The BFO also stopped working at one point, making the LF range vanish. I can't remember the fix but it was probably another dry joint.

I think the 2019 was £80 (from the late Lonsdale twins of Banbury, at Stockwood Park in 2017), that only had a display problem, missing segments caused by a CMOS level shifter IC losing a few units. Replacement would require removing the LCDs which supposedly requires a special tool. I didn't fancy that so I cut the leads on the old IC, extracted the leads individually from the component side of the board then soldered a new part in from the component side, hoping that Marconi's dodgy through hole conductors stayed in position (they did).

[Zenith]

My 2019 came from Stewart of Reading I think five years back. My HP8640A had the scale finally break and had the usual problems with the plastic gears cracking which meant it needed delicate handling. Being stuck I bought a 2019 from Stewart. He said he'd sorted out a good 'un and burned it in.

When it arrived it didn't work, so I phoned to let him know, then opened it and made and remade contacts and it came to life. So it had suffered from being bumped in transit. The next problem was that the frequency was off and the reference wouldn't adjust through 10MHz, it was quite a lot below at best, which is sort of what you expect with aging OCXOs. It wasn't a calibration problem because it was incapable of calibration. I had another brief chat with S of R and he asked me how I knew there was a problem. I said it was well below two Racal counters which agreed with each other and he agreed it was a problem. Neither of us wanted for it to be sent back. I was starting to think a solution might be for him to send me a reference from one of the units on his junk pile. I said I'd keep him posted. The same reference was used in various pieces of Marconi kit and someone had traced and published the circuit. I opened it up and tweaked an internal trimcap. It could now be adjusted to 10MHz. It took overnight to stabilise properly and always at a different frequency. I'd rather it was spot on, but to within a few ppm is good enough for most things. I phoned S of R and told him it was sorted out. As I say, I'm getting round to making up a new reference with one of the Aliexpress recovered OCXOs.

Eventually I found a replacement scale for the HP8640A on fleabay so that's back in commission, but I like the 2019 better.

I saw what looked like a Marconi 2018 or 2019 at Newbury last year. It was on a Get Rid Of It stall, where I picked up a Tek 2213 and a Tek 2465B at very reasonable prices. I was sort of tempted to go back and investigate the Marconi 201X, but I thought that I'd come away with enough to keep me occupied for a time. I'm sure the same people were there this year, but they didn't have the same range of goodies on offer.

I don't really have any objections to the 2019 apart from that it's a 19" rack unit and so is a bit big. There are smaller RF sig gens of comparable performance, but they are quite dear. The just about OK frequency reference is not out of the ordinary for a 40 year old OCXO. I could use an external reference.

[synx508]

There was a Marconi reference in a cabinet of parts that I bought at Milton Keynes last year, it wasn't working properly so I replaced the trimmer. The rubber tube that Marconi stuck over the little Philips style trimmer had degraded and destroyed the very thing it was intended to protect. The 22pF trimmer measured as 42pF (probably also conducting DC, I didn't check that). Because there was too much capacitance it could not be tuned to 10MHz when warmed up. I put a PTFE version of the same trimmer in (I believe they're 18pF rather than 22pF but that's fine) and it now works fine and can be a spare if I need it.

The reference oscillator in the 2019A was not working at all because that same rubber tube had eaten the variable capacitor and the leads of the crystal, which was unfortunately past saving. I had a very nasty cheap 10MHz crystal in my box of crystals at the time, so that went in, with a new trimmer. I'm not sure if the oven circuit even works, but it's good enough to keep the instrument working if the external reference goes away.

Here's the Milton Keynes oscillator from a thread on what was then Twitter (before picture)…

[Zenith]

Yuck! Looks like a very bad case of a NiCd battery leaking. I think a lot of things about plastics weren't understood, particularly how they would age and how they interacted. Mine was slightly crusty inside, but nothing at all like that. Of course if you replace the original crystal with one from the collection, it will get things working, but it almost certainly won't be the proper cut for an OCXO.

I'm sure they used the same frequency reference in several things including the 2955. This was the circuit diagram I found. I recall it was on the UK Vintage Radio Repair and Restoration Forum. Someone had traced the circuit and was good enough to publish it. R5 looks much too high.

It might be useful if you have to repair one or build a substitute.

PM3 OCXO.jpg

I converted if to jpg because this forum doesn't like pdfs. You should be able to find the original pdf with Google.

OCXOs bought at rallies are quite likely to have been removed from something because they were not up to snuff. The 04A option OCXOs I've come across in Racal frequency counters, normally take ages to stabilise and don't quite meet the original spec, but none have been as bad as the Marconi OCXO.

[synx508]

Thanks for the circuit diagram, I think I have a slightly different version from a different source.

The white stuff is thermal grease that someone clearly applied enthusiastically, I've left this and not cleaned it up at all - the oven part seems fine. The problem was really just the rubber tube around the capacitor - perhaps it's there to stop the grease getting inside?

The rally OCXOs came from a source that, by my detective work alone, would probably not have kept hold of out-of-spec parts without making it clear that it was faulty (the broken one was from the same source and had an accurate fault synopsis written on the side in permanent marker!). Whoever it was, they had access to some *very* nice kit. Other stuff included lots of random Watkins-Johnson parts from very high end receivers, mostly ROMs with incrementing firmware release numbers and various OCXOs including some older ones made by ITT that seem to behave correctly and one that's of a type used in a Symmetricom GPSDO. My guess is they were harvested from perfectly good gear being decommissioned.

I'm probably going to end up buying a Leo Bodnar GPSDO in the next 12 months, though.

[Zenith]

That explains it. You really shouldn't need to dollop thermal grease around like that.

I bought a TM4313 from Aliexpress for £70 or so about three years back. The more common and slightly dearer one is the BG7TBL-D.

At the time there wasn't that much about them. It didn't come with documentation.

Since I've found these

A teardown and discussion:

https://tomverbeure.github.io/2023/07/0 ... rdown.html

A datasheet:

https://tomverbeure.github.io/assets/tm ... asheet.pdf

A comparison between six low end GPSDOs including the BG7TBL-D, the TN4313 and the Leo Bodnar number:

https://reeve.com/Documents/Articles%20 ... DOComp.pdf

For his application he chose the TM4313. He points out that any of them would cost less than a new OCXO.

From time to time I've contemplated buying a Leo Bodnar GPSDO as well, but for practical purposes I doubt there's any difference, and the money hasn't been burning a hole in my pocket.

[bd139]

So you got my stolen Marconi OCXO


https://www.eevblog.com/forum/testgear/ ... nd-refurb/

[mansaxel]

I have a Meinberg GPSDO, a M300 rack mount one. It was bought, then retired by its first user, as NTP server, in which capability I also inherited it, but it does have 10MHz and 1PPS outputs. It gives my computers time, and it gives my instruments 10MHz when they need it. I've a very slow project to make it drive slave clock displays (the alternating 24V pulse variant) too.

[synx508]

I've now got a Leo Bodnar LBE-1420 and I'm pleased with it, very small, does what it was supposed to do, easy to get working. The output level's perhaps a bit excessive so I've reduced it with a 50Ω 6dB attenuator. It's also a square wave output which I've filtered into a reasonable sine, but I was doing this with the previous RCS/Quartzlock box, too. I'm now listening to the whine of the 53310A and three 5335As, one of which is around 38mHz high but has been falling very slowly all afternoon.
The output jitter is negligible, seemingly beyond the abilities of the 53310A (I have the high stability timebase option and it has been running for >48 hours).

Here's a little video of the Droitwich derived RCS box wobbling relative to the LBE-1420.

[synx508]

And oh dear. I shall be contacting the company tomorrow as I suspect there's a problem, possibly just the active antenna though. It's flipping between being able to see 35+ satellites and use >12 of them and being able to see 20 to 23 satellites and being able to use 3 or 4. I tried substituting the active antenna for a wholly inappropriate VHF whip and it was seeing about 22 satellites which suggests the problem is the active antenna and not the receiver. Still, it's annoying. It kept time in that state to around 1mHz of wobble but it should be better than that. *sigh*

[synx508]

I panicked, perhaps, I think it's still settling down.