Hi John
great job! I think I have ordered the same enclosure you used for the GPSDO from China for my project!
And what a beautiful scope! Are you happy with your upgrade?
As mentioned before I found a tear down of an CTS OCXO
http://syncchannel.blogspot.com/2016/03/schematic-of-cts-1960017-10mhz-ocxo.html
The schematics is interesting. As expected there is a varactor diode for f-trimming. But Vref comes "only" from
an resistor and a Zener diode.
Unfortunately I found out that my awaited Trimbles seem to have no Vref out. I have to check it when they wil have found there way to me.
Question regarding the Rubidium standard:
As I learned the rubidium frequency is 6.834.682.610,904324 Hz (sorry folks, outside the imperial world we use "," instead of "." and vice versa)
So there must be a FLL involved to get 10 MHz. Have you looked on phase noise? I this a problem?
If it is the same enclosure, it will measure 100 by 100 by 50 mm where the split halves interlock (need to be slid apart meaning the "Lid" can't just be lifted up after undoing the four topmost screws). That's not a problem for the MK II version since, unlike the MK I where I'd soldered the front panel LEDs' flyleads directly to the PCB, I've used a plug ended flylead to connect them to the main board. I only need to undo (just) four screws to release the rear panel, allowing me to slide the PCB out and unplug the front panel LEDs.
I chose those particular enclosures out of a range of similarly sized ones mainly on account the end plates were a nice clean rectangular shape rather than be cursed with a half a millimeter or so trimmed off each of the short (or even all four) sides to within a few millimetres of each corner. It seems to me very likely that was also an influencing factor in your own case.
The 'scope is nice enough but it is really an 'overkill' solution for the testing and monitoring I'm currently doing. If I had enough room on my bench to keep the 1202 alongside, I'd be using that instead since it consumes some 30 watts less power (22W versus the 52 or 54 watts of the 2104X+ (now hacked to the 2504X+ specification).
It's a little disappointing that instead of just a 50% increase, I'm seeing a 150% increase over the 1202's energy consumption with the added insult to this 'injury' of energy consumption in the form of a threefold increased boot up time (46 seconds versus the 16 seconds for the 1202).
The extended boot up time might be an inescapable consequence of a larger and more precisely initialised calibrated feature set that no amount of cpu 'grunt' could speed up but the energy consumption increase came as a bit of a shock.
I could see the energy requirements of 2GSa/s DACs being double that of the 1GSa/s DACs used in the 1202, assuming the same generation of silicon technology being used but I'd have thought the energy budget for the supporting processes would have remained pretty much the same, hence my expectation of a 50% increase in overall energy consumption as opposed to the surprisingly high 150% increase I actually observe.
At the end of the day, the increased energy cost for the improved performance and features is a relatively small price to pay in the larger scheme of things so there's little to be gained by worrying unduly about it. I only mention it to warn others of these hidden costs (protracted boot times and a 2.5 increased energy consumption) and to make it clear that Siglent should do better unless they want to advertise their next but one generation 'scope as being higher output space heaters than those old dual beam Tektronix 'scopes of the fifties and sixties if this increase of energy consumption keeps following this trend.
Other than that 'minor' criticism, it's a very nice DSO indeed.
I had a look at that blog article on the CTS OCXO. I wasn't at all impressed with the design of the thing. A resistor fed zenner for a 4.1v reference??? Really!?!?
and, to make things worse, a 150 ohm resistor in series with its output -150mV drop for a 1mA load (and here was I thinking the 1mV drop with a 1mA load on the CQE OCXO's 5.127v Vref pin was a bit much but, "what the Hey, it's a fixed load in addition to the other internal fixed loads that will only introduce a tiny fixed offset that can readily be perfectly compensated for in my source of fixed dc offsets for my EFC monitoring circuit.").
I'm very surprised that those trimbles lack a Vref pin. I'd have thought that any self respecting OCXO manufacturer would not be so stupid as to give up the opportunity to make their internal temperature stabilised oscillator and thermal controller Vcc rail available for external use via the Vref pin as a "value add" product feature to help boost their sales figures, especially when the cost of the Vref pin itself outweighs the cost of the internal wire link to an existing temperature stabilised Vcc rail. When I see such monumental stupidity as an unused Vref pin (or even none at all), my immediate thought is, "Yeah, they must have shit for brains!".
When a Vref pin voltage is available, that usually (the CTI OCXO being an obvious exception) gives you immediate access to a reference voltage that's only second to a very expensive lab standard voltage reference. The absence of such a pin where the datasheet specifies for example, a maximum 1mA current loading, would be best interpreted as a manufacturing statement which says, "Our internal voltage reference is just too shite for use elsewhere so we're not going to let you see just how shite it really is!". IOW, steer clear of any OCXOs that don't provide a Vref pin voltage with a specified safe maximum current loading.
The caesium atomic clock redefined the second as being precisely 9,192,631,770 periods of the hyperfine transition frequency of a caesium atom in the ground state at absolute zero
to the nearest whole period but all atomic standards based on other alkali metals such as the rubidium atom don't neatly fit an exact number of oscillation periods into an atomic second. The actual frequency being 6.834 682 610 904 29 GHz or 6,834,682,610.90429 Hz to five decimal places of accuracy. I've just realised (dang the European habit of using the comma and fullstop arse about face as numeric separators!) that you've used a six decimal accuracy. You must have tracked down a more recent figure than mine.
If the rubidium atomic clock had come first, the definition of the second would no doubt have been defined as "6,834,682,611 periods of the hyperfine transition frequency of a Rubidium atom in the ground state at absolute zero
to the nearest whole period" instead, leaving the caesium definition with five trailing decimal places as a secondary standard.
Since the second is now defined, arbitrarily as being precisely 9,192,631,770 periods of the hyperfine transition frequency of a caesium atom in the ground state, all other atomic based oscillator references are, by definition, secondary standards. In the case of the compact rubidium oscillators or frequency standards (RFS), this allowed the freedom to use a FLL and a rubidium gas that didn't require a cryostat to hold it at absolute zero.
It's not just the fact that an RFS can be fine tuned and therefore needs to be calibrated against a primary standard or frequency reference, such as a GPS timing source, that's directly traceable to the primary standard, that makes it a secondary standard so much as the fact that being a secondary standard it is freed from the need of a cryostat, allowing the atomic physics package and its supporting electronics all to be housed in a physical package smaller than a half height optical disc drive which can be used to provide a frequency reference with at least an order of magnitude better accuracy than the best double ovened XOs can achieve with the added bonus of not suffering the random frequency shifts that can afflict even the best double ovened XOs which makes it a very attractive alternative frequency standard where weight and bulk are major considerations (they work even better in the vacuum of space which completely eliminates the issue of barometric sensitivity that plagues ground based RFSs)
The Datom and Efratom LPRO-101 bends the 6.834GHz excitation frequency ever so slightly to allow a 20.0000000MHz VCXO to generate a 60MHz lamp excitation RF field and a precise 6.834GHz. The 20MHz VCXO is also divided down to 10MHz and filtered to remove the odd harmonics to produce the 10MHz sine wave output.
The FE-5680 uses DDS technology to generate its 10MHz from a less inconvenient choice of exciter reference frequency which unfortunately introduces a high level of spurs and jitter noise, requiring the use of a separate 10MHz VCXO to clean the output up for use at frequencies of 1GHz and beyond. The 10MHz sine output from an LPRO-101 can be used directly to drive transverters going all the way up to 24GHz. Therefore, since I don't have any means to measure jitter and phase noise, I take it on faith that jitter and phase noise won't be an issue for any of my envisioned future projects.
https://febo.com/pages/oscillators/rubes/JBG