REALLY old HP5065A on eBay right now!

[5065AGuru]

This 5065A is the oldest one I have ever seen (SN prefix 00808)(1968)! Possibly one of the fist units sold! It has the oldest style optical unit (Blue colored) and most likely the -6010 suffix A7 which HP quickly replaced with the -6080 suffix A7. Also has the old style 00105-6013 Quartz oscillator vice the 10811 used in the late versions. If the lamp oven has not shorted and fried stuff it most likely is repairable. Watching to see how high it goes!
Already past what I'd be willing to pay! If someone here gets it I'd be happy to give advise.

https://www.ebay.com/itm/HP-5065A-Rubidium-Vapor-Frequency-Standard/274060976102?hash=item3fcf4e9fe6:g:bjoAAOSw-a5drMy4

Cheers,

Corby

[james_s]

Wonder why people want it so bad? Do people collect these things?

I've never understood why anyone bids so early either, I mean it's got 10 bids already with more than 4 days left. There's never any point in bidding before the last day except maybe an opening bid to nix the Buy It Now option. Bidding against others early on accomplishes nothing but driving the price up as people get antsy and try to outbid each other.

[BravoV]

Wonder why people want it so bad? Do people collect these things?

Just look at the OP's nick, isn't that clear enough ? 

[edpalmer42]

Wonder why people want it so bad? Do people collect these things?

Almost all the Rb standards you see on ebay are designed to be used as a module in another piece of equipment.  As such, they are designed to a price/performance point and nothing beyond.  HP designed the 5065A as a stand-alone piece of lab equipment.  Like all HP equipment from that era, they optimized everything they could.  The result is that the 5065A is arguably the best commercial Rb standard that will ever exist.  The market for such a unit is small which is part of the reason that they're no longer produced.  As a result, they are highly sought after and on the very rare occasions when one shows up, they tend to sell for surprisingly high prices, regardless of condition, and people like Corby get really excited!   

As a second point, the analog clock included in this unit was manufactured by a very famous company (Patek Philippe) that makes very high end clocks.  That name alone gives the unit significant value.

[dietert1]

Several years ago we bought two FE5680 and put them into VXI-VME adapters with little 4x fanout boards. Total cost per unit about € 200. Frequency comparisons yielded stability of better than 10 ** -11 after 1 hour. I doubt the HP unit offered "untested" on ebay performs any better. I mean those smaller Rb modules are several technology generations ahead.

The HP unit may be a collectors item, but i doubt it performs any better.

Regards, Dieter

[thinkfat]

Several years ago we bought two FE5680 and put them into VXI-VME adapters with little 4x fanout boards. Total cost per unit about € 200. Frequency comparisons yielded stability of better than 10 ** -11 after 1 hour. I doubt the HP unit offered "untested" on ebay performs any better. I mean those smaller Rb modules are several technology generations ahead.

The HP unit may be a collectors item, but i doubt it performs any better.

Regards, Dieter

http://www.ke5fx.com/rb.htm

The FE5680A with its DDS synthesizer was designed for a totally different purpose than the HP5065A. For example, the wide trimming range of the FE5680A has a catastrophic impact on the phase noise. The stability of the physics package is also worse. I don't think they're in the same ballpark.

On the topic of "several technology generations ahead", yes, true, but the design goals were completely different. The driver for the telecom grade standards development was more "how cheap can we go and get away with", so they've been cutting corners wherever they could. For example, the Datum LPRO design doesn't feature an ovenized XO and so it requires a larger loop bandwidth and results in more phase noise on the output signal. Then there's temperature drift of the C-field, stability of the temperature control, susceptibility to external magnetic fields, ...

[dietert1]

When i have two Rubidium oscillators and they agree within 10 ** -11 plus some fixed offset, this is absolutely sufficient for all practical purposes.

That old "untested" ebay HP device may not work at all, no matter what it was meant to be. The buyer will eventually need a new physics package, a complete refurbishment and a calibration. That device wasn't meant to run on for years without regular maintainance.

Regards, Dieter

[thinkfat]

When i have two Rubidium oscillators and they agree within 10 ** -11 plus some fixed offset, this is absolutely sufficient for all practical purposes.

For _your_ purposes. And probably mine, too, but that's not the question.

That old "untested" ebay HP device may not work at all, no matter what it was meant to be. The buyer will eventually need a new physics package, a complete refurbishment and a calibration. That device wasn't meant to run on for years without regular maintainance.

"eventually" -> "probably", "maybe". I often see this mistake, the German "eventuell" does not translate into "eventually" at all.

That being clarified, yes. That's the risk you take with these old devices and its unclear origin, but that's not stopping people who need such a device for _their_ purposes, which are not limited by yours or my imagination.

[dietert1]

Sorry, when i write eventually, i mean eventually (German: letztlich). That device is probably sold as "untested" because the physics package arrived at the end of its life and it is unobtainium. My five cents after several hundred ebay deals. In fact, we also bought a total of three used FE5680 oscillators because one of the first two did not lock after some minutes.

Regards, Dieter

PS: Any Rubidium oscillator is at least a factor 1000 more stable than the best crystal oscillators you can get, and anybody interested in precision frequency should get one, but a modern unit please. Spend your time getting it locked to GPS instead of trying to fix that old HP unit.

[edpalmer42]

One advantage of these older units is that they can be repaired.  The Rb cell itself has an indefinite lifetime.  It has no defined wearout processes that will limit its lifetime.  Corby has the experience to say more on that matter.  So if the cell is intact, everything else can be repaired, depending on how much time and effort you're willing to put into it.  The electronics are all through-hole boards.  There are no processors, therefore no software, and almost no digital circuitry.  The schematics and circuit descriptions are widely available.  I can think of a few hard-to-find parts, but since full documentation is available, parts or full modules could be substituted or replaced with new designs.

So why would you want to bring one of these units back from the dead?

I did a 5-day data run where I measured my 5065A against my HP Z3801A GPSDO.  Averaged over the 5 day period, the frequency drift was ~1e-13 per day.  But that number is so small that you really need to measure it over a longer period to get an accurate value.  I couldn't really measure the Allen Deviation because the inherent noise of the GPSDO masked it.  Other tests suggest that the noise level of my 5065A is at an Allen Deviation level in the low e-13 range.  These numbers are significantly better than the specs, but are typical for the 5065A.

Why would you want that kind of performance?  Well, that's where the 'Nut' part of TimeNuts comes in. 

By the way, I see that the bid is at $910 with 8 bidders.  I'm expecting a feeding frenzy in the last few minutes.   

[thinkfat]

The Rb cell itself has an indefinite lifetime.  It has no defined wearout processes that will limit its lifetime. 

Interestingly, the newer Rb standards seem to suffer from Rb migration in the lamp, which will limit their lifetime. I looked at the operating manual for the HP5065A, it's built from the same principles as, say, an Efratom FRK. Optical pumping, Rb lamp with a 100MHz RF exciter oscillator, passive absorption cell, SRD multiplier.

I'm really curious about those newfangled SOIC-size Rb standards I keep hearing about.

[maat]

So if the cell is intact, everything else can be repaired, depending on how much time and effort you're willing to put into it.

Definitely true and even if the cell is broken, rubidium is a lab commodity. You can get a cell that fits into the oven in any flavour you like.

[5065AGuru]

OK, Here is some info for this discussion.
The 5065A stability can't be beat except by an Active Maser or for short Taus a selected Quartz (BVA8607 opt . A Cesium at long Taus will eventually beat it also.
The plot shows some standards and the 5680 for comparison(black plot).
Most 5065A will easily match the green plot. Mine (yellow plot) is an exceptional unit!
There are no "replacement physics packages" however they can be repaired as long as none of the 3 glassy bits are damaged. (lamp/resonance cell/filter cell)
Although in the very rare instance of a bad lamp you can replace it with a lamp from a dud telecomm unit. (with some work)
Rubidium may be a commodity but trying to manufacture new cell is not happening! (trust me!)
The cell will not fail unless mechanically damaged and opened to air.
A Quartz oscillator (8607A option is at least 100 times more stable at low Taus than a 5680.
Even the Quartz in my Passive maser which is not a particularly good performer, (purple line) is 10 X better at low Taus than the 5680.
Stability and accuracy and drift rate are not the same thing!
Long baseline array astronomy uses Masers as they are more stable than Cesium for the time periods of interest, also the 5065A was used early on at several astronomy sites for the same reason.
For an accurate frequency source a GPSDO is the way to go, but to measure stability you need a reference better than the device you are testing!
I do agree that the "first article" 5065A on eBay seems way overpriced but it should be repairable and the analog clock is neat.
However a late SN unit (prefix 2816A) in good working condition is worth $3500-$4000!
Here is another bit of info to chew on: There is a company out there that uses 5065A units in a proprietary application as no other modern device is good enough, last I heard they had over 35 5065A in use around the world!
The 5065A was built between 1968 and 1988 and I find it remarkable that 31-51 year old instruments easily outperform the modern units, and as someone who hates it when a piece of my test equipment dies due to software or computer issues, I love the older non processor HP stuff with voluminous manuals/parts lists/theory of operation.
Well enough  for now!

Cheers,

Corby

[CJay]

http://www.ke5fx.com/rb.htm

The FE5680A with its DDS synthesizer was designed for a totally different purpose than the HP5065A. For example, the wide trimming range of the FE5680A has a catastrophic impact on the phase noise. The stability of the physics package is also worse. I don't think they're in the same ballpark.

You need to know a little more about the FE5680 module and its variants before you make blanket statements like that but in general, you're correct, the 5065 is in a different league, however, it's also from a different era and as such good examples are rare, most have been worn out, abused and many are in need of significant attention before they'll work, never mind meet specs.

[thinkfat]

http://www.ke5fx.com/rb.htm

The FE5680A with its DDS synthesizer was designed for a totally different purpose than the HP5065A. For example, the wide trimming range of the FE5680A has a catastrophic impact on the phase noise. The stability of the physics package is also worse. I don't think they're in the same ballpark.

You need to know a little more about the FE5680 module and its variants before you make blanket statements like that but in general, you're correct, the 5065 is in a different league, however, it's also from a different era and as such good examples are rare, most have been worn out, abused and many are in need of significant attention before they'll work, never mind meet specs.

As far as I know, there's a good number of different FE5680A modules. Some feature fixed 10MHz output, some only 1pps and there's at least two different DDS chips, some are programmable over RS232, some not. But the principal operation mode is always the same. The actual rubidium locked oscillator outputs a signal at around 50 MHz, which is fed to a DDS chip (through a /2 divider I think), which then outputs the actual reference frequency. So, I assume that all variants have somewhat identical performance.

As for this HP instrument, I wouldn't touch it. I have no use (not even space) for it and no time to give it the proper attention. But others obviously have, or they wouldn't sell. And apparently there are enough success stories of such purchases around to attract buyers.

[CJay]

http://www.ke5fx.com/rb.htm

The FE5680A with its DDS synthesizer was designed for a totally different purpose than the HP5065A. For example, the wide trimming range of the FE5680A has a catastrophic impact on the phase noise. The stability of the physics package is also worse. I don't think they're in the same ballpark.

You need to know a little more about the FE5680 module and its variants before you make blanket statements like that but in general, you're correct, the 5065 is in a different league, however, it's also from a different era and as such good examples are rare, most have been worn out, abused and many are in need of significant attention before they'll work, never mind meet specs.

As far as I know, there's a good number of different FE5680A modules. Some feature fixed 10MHz output, some only 1pps and there's at least two different DDS chips, some are programmable over RS232, some not. But the principal operation mode is always the same. The actual rubidium locked oscillator outputs a signal at around 50 MHz, which is fed to a DDS chip (through a /2 divider I think), which then outputs the actual reference frequency. So, I assume that all variants have somewhat identical performance.

As for this HP instrument, I wouldn't touch it. I have no use (not even space) for it and no time to give it the proper attention. But others obviously have, or they wouldn't sell. And apparently there are enough success stories of such purchases around to attract buyers.

And that's why you need to know a little more. The 'fixed' output ones don't all have that second DDS chip fed by the locked Rb oscillator, the 1PPS devices with a 15 pin connector (FE-5680B) usually don't either.

[edpalmer42]

Okay, I didn't expect that!  $2938 ??    And for a really old unit!  Someone really, really, really wanted a 5065A.

I can't decide if the buyer was an amateur who overpaid or a professional who really needed a 5065A. 

Ed

[james_s]

Yikes. Well it could just be a really dedicated collector. I mean I know guys who paid more than that for an old pinball machine, several of the classic arcade games I own are now worth more than that. I've seen people pay orders of magnitude more for a painting that I wouldn't even want if it was offered to me for free. I wouldn't particularly want an old boat anchor  like the 5065A, but I'd much rather have that than a sculpture or something that doesn't even do anything.

[dietert1]

Yesterday german TV had a show about gold. They explained how useless gold is (except for some applications in electronics) and that it really does nothing. It's expensive because it's rare and difficult/impossible to make.
This was the start of the thread - indication of a rare offer. Little to do with the performance of one Rb standard in comparison to another. That's pure fantasy.

Regards, Dieter

[james_s]

I would argue that gold is extremely useful stuff, it's just far too expensive to be used in many of the applications for which it is well suited. It is an excellent reflector of heat and makes very good thermal insulation. It is nearly inert and makes an excellent coating that is highly resistant to oxidation. Virtually any metal hardware could benefit from gold plating if the stuff was cheap and plentiful.

[edpalmer42]

Which brings us right back to the 5065A and most (all?) HP equipment from that era.  The circuit boards didn't have tin plated traces, they were gold plated!   

Ed

[thinkfat]

Okay, I didn't expect that!  $2938 ??    And for a really old unit!  Someone really, really, really wanted a 5065A.

I can't decide if the buyer was an amateur who overpaid or a professional who really needed a 5065A. 

Ed

Well, there was one bidder with a history of over 29000 transactions on Ebay who bid $2888. That definitely wasn't a noob. The winning bid was $50 higher, from someone who really, really wanted it. He had set a bid limit of $2500 and then had to raise it manually to take this one home. We won't know how much he was willing to spend, the $50 are the automatic ebay increment...

[dietert1]

Nobody needs to know anything about oxidation to be the owner of gold and nobody needs to know anything about atomic clocks to be the owner of that 5056A. Sometimes those ebay deals fail later on as soon as the parents realize what their youngsters have done.

Regards, Dieter

[notfaded1]

I can see how a collector would want to clean up and get one of these working again... the Patek Phillipe watch alone is rare.  The HP5065A is a rare bird and one that many would like to have I think.  I'm guessing that the main payload 3 tubes were all intact.  It's pretty clean really if you look at it.

Bill

[Drewbie]

Here is another bit of info to chew on: There is a company out there that uses 5065A units in a proprietary application as no other modern device is good enough, last I heard they had over 35 5065A in use around the world!

That sounds intriguing!

Can you expand on that please?