Found home made rubidium reference that seems to really work well

[notfaded1]

Regarding the TAPR TICC, I see you have a 53131A. That should be fine, I just don't know if you'd want to dedicate it permanently towards monitoring a GPSDO

Maybe LH supports it already.

I concur.  It might be nice to not have to dedicate the 53131A to this for sure.  I like the idea of having something that can run 24x7 and monitor the status of the GNSSDO and it would great to in addition to having a rock solid 10MHz to also having a monitored accurate 1PPS signal.  I have a nice distribution amplifier that has it's own internal 10MHz oscillator but also can pass an inputted signal.

And hey it goes to 11!!! How can that not be good???

B

[ArthurDent]

The linear modulation technology unit isn't that bad for general use but you have to remember which BNC outputs are connected together to avoid possible crosstalk.

[notfaded1]

The linear modulation technology unit isn't that bad for general use but you have to remember which BNC outputs are connected together to avoid possible crosstalk.

Any idea why they would add the one lone 11th output?  I thought that was kinda neat but also kinda strange?  That was all the room they had in a 1U space?  Do you have one too?

B

[notfaded1]

What you could do is order a TAPR TICC (it's about $250), which LH supports, and feed it the 1PPS signal from the GPS module and the 10MHz from the LO and see what you get there. No idea if LH can work both with a GPS and the TAPR TICC but it's worth a shot.

I'm looking at a different GPSDO now.  A later model Trimble Thunderbolt.  It seems to offer up more data including the LO.  It would be interesting to compare the two GPSDO to each other.  I have enough antenna connections on my splitter for them both to run together along with my sync server.

B

[ArthurDent]

The linear modulation technology unit isn't that bad for general use but you have to remember which BNC outputs are connected together to avoid possible crosstalk.

Any idea why they would add the one lone 11th output?  I thought that was kinda neat but also kinda strange?  That was all the room they had in a 1U space?  Do you have one too?

B

The photo I posted is of the unit I own.

[notfaded1]

A little clean up on the TI measurement of the unknown Rb.  After following the directions I seem to be getting better readings than I did before. 

To do Time Interval measurements with one interface on the 53131A I followed this procedure:

Example Procedure for Using Common 1 to Make Time Interval (TI)
Measurements on a Single Signal
1 Connect a signal to channel 1 of the Counter.
2 Press Time & Period key until TI 1 TO 2 is momentarily displayed. (Note that the
Time, Ch1, and Ch2 annunciators light.)
3 Press Trigger/Sensitivity key until COMMON 1: OFF is displayed.
4 Press any one of the arrow keys until COMMON 1: ON is displayed.
When the Counter is operating in Common 1 mode, the signal applied to CHANNEL
1 is used for both start and stop event. The trigger level, slope, and sensitivity for the
start signal can be modified by using the CHANNEL 1 input conditioning keys.
Whereas, the trigger level, slope, and sensitivity of the stop signal can be modified by
using CHANNEL 2 input conditioning keys. (See the figure located at the end of the
sub-section of this section titled “Overview of Trigger/Sensitivity Menu.”)

B

[Johnny B Good]

 Hi, notfaded1

 I've been reading this topic thread to while away the time whilst I await delivery of an Efratom LPRO-101 I'd ordered from a stateside dealer just over ten days ago now in the hope that you would finally break open the wooden box to reveal what I strongly suspect will be an undervolted FE-5680A.

 For some reason I thought I'd seen a later posting dated May, hence this contribution. I guess it must have been a "Last Edit:" date/time stamp I'd seen attached to an earlier posting. Anyway, if anything, the absence of any report on what you found (if you ever did get round to cracking it open  ) after more than four months since your last posting, makes the delay worse. Lucky for me that I've been reading 'ancient history' rather than following along in real time.

 So, the unanswered question remains - did you finally discover what your "unknown Rb" frequency standard was? Before you dignify that question with a reply, let me first offer the reasons for my own prediction that it's most likely going to be an "undervolted FE-5680A" before you reply.

 I could see from your photos that it was using a 12v laptop charging brick rather than the expected 19v one typically used to meet the minimum (single rail) supply voltage requirements of an LPRO-101 (19 to 32vdc, 24v recommended). That, along with what looked like a 5v smpsu board, initially suggested an FE-5680A until I checked actual voltage requirements for the FE Rb units when I discovered that they require 15 to 18vdc for the physics package as well as a separate 5v 200mA dc supply for the supporting logic, hence my referring to it as an "undervolted FE-5680A" which would neatly explain the much protracted warm up delay and the unsteady lock lamp indication.

 I suppose it's just possible that the previous owner may have modified that 12v laptop charging brick to output 15 volts but that seems rather unlikely, judging from the use of a wooden box to house a Rb oscillator that needs to be mounted on a solid heat spreader thermally coupled to ambient air (fan assisted or not).

 If I ever felt the need to modify a 12v laptop charging brick for use as an integrated 15vdc PSU, I'd be rather inclined to remove its guts and scrap the plastic housing to improve its reliability by allowing it to run some 15 to 20 deg C cooler.

 Luckily for myself, I can just use a bog standard 19v laptop charging brick as an external psu without any need to increase its output voltage ( a ripple free 19vdc supply is just about the optimum for an LPRO-101 to minimise excess heat from its internal 17v analogue voltage regulator). I've no shortage of 65W and 90W rated 19v charging bricks to choose from

 One final thought has just occurred to me regarding an outside chance of it being an LPRO-101 and that is: the 12v charging brick's output could have been placed in series aiding with the output from the 5vdc smpsu board to create a 17 volt supply, resulting in a similar undervolt condition unless the 17v analogue regulator had been strapped out of circuit inside the LPRO-101 module.

 However, otoh, a 17 volt supply is nicely within the 15 to 18 volt range for the FE-5680A' physics package and the 5v smpsu, if it's being used to lift the zero volt rail of the 12v supply, can still provide the required 5v to the support logic rail.

 It can be fun to speculate but could you please report whether or not you finally took the plunge and cracked open the box and, if so, what did you find inside?   

JBG (still awaiting delivery of an LPRO-101 that's been stuck in a global air freight terminal in Erlanger, KY since last Thursday afternnoon )

[EDIT 2020-08-25] It arrived yesterday afternoon, just before 12am. I've now got it trimmed to within 60ppt (2 cycles drift over a 9 hour overnight  run) compared to my MK II GPSDO. I'm looking to improve on that calibration attempt once I've got it housed in a thermally regulated case.

[thinkfat]

A little clean up on the TI measurement of the unknown Rb.  After following the directions I seem to be getting better readings than I did before. 

To do Time Interval measurements with one interface on the 53131A I followed this procedure:

Example Procedure for Using Common 1 to Make Time Interval (TI)
Measurements on a Single Signal
1 Connect a signal to channel 1 of the Counter.
2 Press Time & Period key until TI 1 TO 2 is momentarily displayed. (Note that the
Time, Ch1, and Ch2 annunciators light.)
3 Press Trigger/Sensitivity key until COMMON 1: OFF is displayed.
4 Press any one of the arrow keys until COMMON 1: ON is displayed.
When the Counter is operating in Common 1 mode, the signal applied to CHANNEL
1 is used for both start and stop event. The trigger level, slope, and sensitivity for the
start signal can be modified by using the CHANNEL 1 input conditioning keys.
Whereas, the trigger level, slope, and sensitivity of the stop signal can be modified by
using CHANNEL 2 input conditioning keys. (See the figure located at the end of the
sub-section of this section titled “Overview of Trigger/Sensitivity Menu.”)

B

I still don't think this is the Rb you're seeing... I'd try something different:
* Measure "Frequency 1"
* In "Gate&ExtArm", set Gate Time to 2 seconds, this yields an additional digit.

Capture the absolute frequency data with TimeLab, remember to set the interval to 2 seconds when importing.
Remember to set the scaling correctly in TimeLab. Sometimes the 53131A prints MHz, sometimes just Hz.

[tkamiya]

Do you think HP53131A has enough resolution for this purpose?  As I understand it, it's LSD resolution is 500ps.  Even if you go up to HP53132A, it's 150ps.  On top of it, reference oscillator's stability comes into play. 

500ps is 500 x 10^-12, which is 0.5 x 10^9, which isn't much.  This is regardless of how many digits you have displayed.  Typical Rb is orders of magnitude better than this.

I'm struggling with this kind of limitations myself. 

[thinkfat]

500ps is the single shot resolution, if I'm not mistaken. Measuring frequency, with a 2 second gate time he should at least be able to see something of the bg7tbl he's using as external reference.

 I'm using this setup for measuring my DIY GPSDO and while the small tau < 10s is down in the noise, I can clearly see it for larger tau. I cannot see any of my Rb though. But I can measure another commercial GPSDO just fine.

[notfaded1]

Thanks @thinkfat for this comment.  I've recently figured out that you have to set it to 2 seconds to get meaningful data from 53132A.  If you leave it at the shortest time period it's just too short.  I have a couple Cs now and a bunch of Rb so I've figured a lot of this out with trial and error.  I have a Pendulum CNT-90 as well now.  I will say if you set it to 2 seconds the 53132A gets much much better data you can use in timelab.  No one ever really told me this I had to figure it out on my own... for too long I left it at sub second and couldn't understand why it was so hard to reach -13's.  I'm using an Agilent 5071A now as my reference for the counters in some measurements which has helped.  New to me soon will be a 5065 Rb I've always wanted but I was never was around when they were really available.  I'm hoping after letting it run for a while (weeks most likely) I'll be able to test short and medium term stability with much better precision which for a reference the Cesium isn't as ideal I don't think.

Now if we could just get a timepod or phasestation that's less than 5k we'd by stylin'      Wouldn't hurt if BVA's grew on trees either!

[notfaded1]

Also now that I have other better references I'll maybe tear into the wood box this weeknd and find out what's inside since it's not really working anyhow.

Bill

[5065AGuru]

Bill, I did not read through this whole post but wondered if you had considered using a DMTD system along with your 53132A.
I used a DMTD system for a couple years with either an SR620 or an HP 53131A and had good results. The DMTD system I use now has a built in counter that eliminates a minor dead time problem that occurs with the SR620 or the 53131/132A. It works so well I sold my SR620!
I managed to work around that problem until the new counter went operational.
A DMTD would give you a AD floor at 1 Sec. of less that 2 to 4 X10-13th.
You would have to build it yourself but there are schematics and info on the web to help.

https://nvlpubs.nist.gov/nistpubs/Legacy/IR/nbsir75-827.pdf

http://www.stable32.com/A%20Small%20DMTD%20System.pdf

Also I've thought of replicating the input stages of the EFRATOM TS-105 Atomic test set. (another DMTD unit) This would allow one a much easier build for those wanting to try a DMTD system!

Cheers,

Corby

[notfaded1]

I really want a good DTMD system Corby.  I was watching this and dreaming:  https://www.ebay.com/itm/Symmetricom-TSC-5120A-Single-Box-Phase-Noise-Analyzer-Allan-Deviation-Test-Set-/164356367460

DANG even used look it sold for 15.5k!

I have an SR620 myself and both 53131 and 53132 both with 3G third interface and the 010 LO built in.  Recently I bought a nice Pendulum CNT-90 too which I like.  I bought one of the Chinese DTMD from TURN Dynamic Systems but it doesn't seem to get great numbers at all.  Lately I've been just using my 53132 with the 5071A feeding it as an external reference and it's working ok but you don't get any real phase data like you do with a DTMD.  I've been reading a bunch of papers about different devices including "Comparison of high-precision frequency-stability measurement systems"  I wish I could post the full document but it's IEEE that licensed to me at work and it stamps the PDF with my company name.  Here's the Abstract:

The performance of two commercial high-precision frequency-stability measurement systems, timing solutions 5110A time-interval analyzer and Quartz lock A7 frequency and phase comparator, has been compared with two laboratory dual-mixer time-difference (DMTD) systems developed at the institute of radio engineering and electronics (IREE), Prague, and istituto elettrotecnico nazionale "Galileo Ferraris" (IEN), Turin, respectively. Two BVA oscilloquartz 5 MHz oscillators with Allan deviation (ADEV) around 1/spl times/10/sup -13/ at averaging intervals 1 s to 30 s were used as test-signal sources. In terms of the best-measurement-capability (BMC), the compared systems have shown the background noise at 5 MHz of ADEV=4.2/spl times/10/sup -14/ at 1 s in IREE-DMTD, 4.4/spl times/10/sup -14/ in IEN-DMTD, 7.4/spl times/10/sup -14/ in A7, and 8.8/spl times/10/sup -14/ in 5110 A, respectively. The 5110 A analyzer behaved the best concerning the dependence on departures from the BMC conditions.

Great paper!  Here's the noise floor samples from the devices:

I'm learning fast but I'm a Computer Scientist and not really an EE so it takes me a little longer to figure some things out but I learn fast.  I want to do exactly what you're talking about Corby.  Even with many of these DTMD they still use an SR620 as the TIC for the experiment.  See second picture.

Bill

[FriedLogic]

There was this 'DIY' version from a decade ago:
http://www.wriley.com/A%20Small%20DMTD%20System.pdf
I did hear something about there being an update for it, but I don't know the details.

A variation on the theme:
http://www.ke5fx.com/tpll.htm

Quite a few people have made their own, but other than the Small DMTD above I've not come across any that have been documented to allow others to easily play along, which is a pity. Many people have little enough time for their hobby without having to design and build the test gear from scratch, so I'm sure it would be appreciated. Maybe even with money, although that can bring more problems than it's worth.

[5065AGuru]

FriedLogic,

There have been a couple nice systems offered in the past on TimeNuts at VERY inexpensive costs.
Unfortunately not enough takers, and as is sadly the case a lot of the time, people nit picked it and it went away.
It and a later version are my main and backup DMTD.
Like I mentioned before I will look over my notes on using the EFRATOM TS-105 mixer scheme and make a board to test. If it has promising results I will share the performance/parts list/schematic/and ExpressPCB file.
The SR620 and 53131/2A are major overkill with a DMTD
With a 1Hz offset oscillator you would count the time interval between the DUT and REF 1PPS outputs.
If operating at 10Mhz and the counter has 9 digits the resolution of the LSB is 1X10-15th.
So using 12 digits from a fancy counter will give you and LSB of 1X10-18th! Even with 9 digits the last couple digits are "noise"!

Cheers,

Corby

[notfaded1]

Good Morning Corby-

I'm totally game if you put together a formula for a working DTMD.  I can assemble just about anything if we have a good PCB and known parts.  I've been studying all the diagrams and papers but it's just a bridge too far for me to just whip one together at this point (and build the pcb layout).  At its foundation it's really a few different parts that work together.  But then what parts really will work together.  What mixers, what splitter, how to offset the reference, zero crossing detector and it is RF after all too which is a kind of black magic.

Best Regards,

Bill

[5065AGuru]

Bill,

I've just about finished the board layout. Hope to have it double checked this weekend!
I'll order the Standard mini board which will give me 3 boards.
Once they arrive  I'll assemble one and make sure it does what we want.
Will let you know!

Cheers,

Corby

[thinkfat]

Bill,

I've just about finished the board layout. Hope to have it double checked this weekend!
I'll order the Standard mini board which will give me 3 boards.
Once they arrive  I'll assemble one and make sure it does what we want.
Will let you know!

Cheers,

Corby

I'm in the same board. I'll happily assemble something, provided I have a PCB and BOM. Gerber files and Schematic would be even more awesome!

[notfaded1]

Bill,

I've just about finished the board layout. Hope to have it double checked this weekend!
I'll order the Standard mini board which will give me 3 boards.
Once they arrive  I'll assemble one and make sure it does what we want.
Will let you know!

Cheers,

Corby

I'm totally psyched Corby!  I really appreciate your experience with this.  I've made some strides in my measurement this year for sure.  A working DMTD that I could interface to timelab or load data into Stable32 would be ideal.

Bill

[notfaded1]

I'll be firing up the 5065 this weekend as well... lot's of fun!  Can't wait to let it burn in a little while.    I like the timenuts email forum but there's something really nice about having a true HTML5 web board like we have here on eevblog.  I have some other Rb that would be perfect to test against the 5065 once it settles in... couple AccuBeat AR40A, Datum 8040, Efratom FRK-S, Efratom M-100, a few Efratom LPRO-101, LEA-6T disciplined SRS PRS-10, GNSS disciplined Latlon-Rb-102, and some other GPS OCXO including FEI PicoSync, Thunderbolt (gold type), Thunderbolt E, BG7TBL GNSS OCXO, Datum Exactime 6010, HP Z3805A, Symmetricom Mark V (this one is neat it has Trimble Force 22 GPS in it and Rb LO).  I've been saving all my data from TimeLab as I progress... I've learned so much by trial and error and experimentation.  I think I'd like to do an Rb study and a separate GPS disciplined Rb/OCXO one.  I love seeing the graphs overlaid on top of each other comparing the results.

Bill

[FriedLogic]

There have been a couple nice systems offered in the past on TimeNuts at VERY inexpensive costs.
Unfortunately not enough takers, and as is sadly the case a lot of the time, people nit picked it and it went away.
It and a later version are my main and backup DMTD.

I cringe when I see someone asking for features that others would like to have in some project. The resulting daft wish list can delay or derail the whole thing.
What I more had in mind was just making the design files and maybe some basic notes available on a 'you might find this useful/interesting' basis, like a lot of the stuff that finds its way onto GitHub.

Like I mentioned before I will look over my notes on using the EFRATOM TS-105 mixer scheme and make a board to test. If it has promising results I will share the performance/parts list/schematic/and ExpressPCB file.

Thanks, I'd be very interested in seeing that.

The SR620 and 53131/2A are major overkill with a DMTD
With a 1Hz offset oscillator you would count the time interval between the DUT and REF 1PPS outputs.
If operating at 10Mhz and the counter has 9 digits the resolution of the LSB is 1X10-15th.
So using 12 digits from a fancy counter will give you and LSB of 1X10-18th! Even with 9 digits the last couple digits are "noise"!

With the increasing price of electricity, another issue these days is energy consumption. A relatively simple TDC based timer might work here.

[notfaded1]

With the increasing price of electricity, another issue these days is energy consumption. A relatively simple TDC based timer might work here.

The electricity is the least of my issues... I live in the southwest desert of Arizona to begin with and the AC runs basically 8 months out of the year.  Very little humidity though!!!

I also have a TURN Dynamic Systems DMTD but didn't have great results at first with it... I should probably try it some more now that I have better references.  Can't wait to see what Corby comes up with.

Bill

[Johnny B Good]

@notfaded1

 Just a quick question. Did you ever discover what was actually inside that wooden box? 

[EDIT 2022-01-08]

 Well, did you?