bg7tbl gpsdo master reference

[Ringmodulator]

My BG7TBL, (see my pictures earler in this thread) had a gps unit with the markings removed. It was detected by ucenter as an M8n.
It could be configured with ucenter and the bootmessage was als like a genuine device and it worked just fine So I am pretty convinced, that it was the real thing.

I have replaced it with a M8T which is pin compatibe. no further mods on the pcb needed.

Be aware, that there are Neo and Lea versions, that have different dimensions.

I have also replaced the OCXO with a Morion MV89A double oven.

It is not rocket sience.

Some info here http://amateurtele.com/index.php?artikel=203&hlk=en
and here

All in all, I dont think, that these modifications are really worth the hassle and if I had nor tried, I would never have known.

Remember, not all BG7TBL gpsdos are created equal.


Chis

[AndrewBCN]

I just got a new BGTBL GNSSDO, the successor to the GPSDO, with an used (OEM-)specific OCXO.
Just would like you to know, that the parts configuration and placement does and may vary depending on the original OCXOs used. This might make it almost "impossible" to replace a defective (original) OCXO with another one without adding additional parts of unknow specs, dependig what OCXO was used originally by BG7TBL. The major problem is, that there are no schematics "in the wild" or supplied, which could be used. The latest incarnation of the GPSDO, now called GNSSDO, has a new motherboard (2021-06-10) to adapt to all kinds of (used) OCXOs which might be available right now on the used market. Unfortunately many rare OEM OCXOs are not documented, even when searching the net intensively, and may feature - eventually - a different PIN layout or other variations in specs, which make it almost impossible to find a direct replacement in those cases. Mine i.e. got an OEM specific C-MAC OCXO, which seems to have reached (almost) "End-of-life" (tuning voltage 8.212 82VDC, multiplier is X3 already) but my fully functional replacement OCXO (Oscilloquartz 8663-XS) does fit mechanically but not electrically exactly for this reason, because some parts are missing on the board (solder pads are unpopulated, no values given etc.) for substituting the OCXO. I already looked around for another replacement OCXO (same kind), but no luck...
Sending it back from Europe costs a lot too and does not make this feasiblack boxle.

If you look at page 1 of this thread, you will find an ever changing list of different board layouts, which seems to continue "endlessly" up to the current date. Many OCXOs have an identical PIN layout and usage, but some vary quite a bit not only by size.

You have listed all the disadvantages related to the fact that the BG7TBL GPSDO is essentially a "black box" product. You don't have the schematics, it uses recycled and sometimes "unobtainium" parts, and you have zero control over the firmware. Its sole advantage is being extremely inexpensive for what it does.

If you want an Open Source (hardware and software) GPSDO that you can put together yourself for less than half the cost of a BG7TBL, and where you have total control of both the hardware and software, there are at least two described right here in the EEVblog forums:
1. Lars' DIY GPSDO by Lars Wilenius.
2. The STM32 GPSDO by yours truly i.e. myself.

[burkm]

In the end an OCXO just needs a power supply and tuning voltage. It shouldn't be hard to retrace the circuit which drives the OCXO tuning voltage. Most likely a DAC and an opamp. Power supply is a matter of measuring which voltages are available where (likely 5V and 12V). And some retracing to determine where the filtering components are  and then to move these if necessary. Only things you need to find out are a DMM with continuity and a piece of paper to draw on.

I tried this already myself, but the new PCB has a somewhat different layout (more complex) in comparison to older PCBs, which makes it near "impossible" to find corresponding parts, because they didn't exist before and are of unknow specs...
Some components with the same name have different values compared to older PCB editions. This makes it hard to tell, what should be used, when switching the OCXO. I don't want to exchange components on half the PCB after all without really knowing what I'm doing...
Note: Even the multiplier (x2, x3) seems to use differently specified components and is - componentwise (resistors) - partly unpopulated, although the OP amp used (358) is the same. Not even the operating voltage of the OCXO is specified on the label, thus I have to trust BG7TBL to have selected the right one. Note: Some of the C-MAC OCXOs have operating voltages of 12V, 15V and 24V.

[AndrewBCN]

Many people here in this thread have been obsessed with swapping out the u-blox GPS module in their GPSDO for a much more expensive and supposedly more precise one, and supposedly that would make their BG7TBL GPSDO perform "better" or "more accurate".

Of course most people don't even have access to the equipment required to back these claims with solid data.

But one person (John Ackermann N8UR) went ahead and did a rather good job of collecting the data on a variety of u-blox receiver modules, and the verdict is quite clear imo. They all perform more or less the same.

https://hamsci.org/sites/default/files/publications/2020_TAPR_DCC/N8UR_GPS_Evaluation_August2020.pdf

A very good read, and perhaps that can settle the question.

[0xFFF0]

Many people here in this thread have been obsessed with swapping out the u-blox GPS module in their GPSDO for a much more expensive and supposedly more precise one, and supposedly that would make their BG7TBL GPSDO perform "better" or "more accurate".

Of course most people don't even have access to the equipment required to back these claims with solid data.

But one person (John Ackermann N8UR) went ahead and did a rather good job of collecting the data on a variety of u-blox receiver modules, and the verdict is quite clear imo. They all perform more or less the same.

https://hamsci.org/sites/default/files/publications/2020_TAPR_DCC/N8UR_GPS_Evaluation_August2020.pdf

A very good read, and perhaps that can settle the question.

Thats not 100% true. The F9 series is half an order of magnitude better.

[AndrewBCN]

Many people here in this thread have been obsessed with swapping out the u-blox GPS module in their GPSDO for a much more expensive and supposedly more precise one, and supposedly that would make their BG7TBL GPSDO perform "better" or "more accurate".

Of course most people don't even have access to the equipment required to back these claims with solid data.

But one person (John Ackermann N8UR) went ahead and did a rather good job of collecting the data on a variety of u-blox receiver modules, and the verdict is quite clear imo. They all perform more or less the same.

https://hamsci.org/sites/default/files/publications/2020_TAPR_DCC/N8UR_GPS_Evaluation_August2020.pdf

A very good read, and perhaps that can settle the question.

Thats not 100% true. The F9 series is half an order of magnitude better.

"better"? That is a poor way to describe the data in the article. 

Also, the "better" you refer to would make zero difference in a BG7TBL GPSDO. But I would be happy to be proved wrong, if you or anybody else have the data to show for it. 

[burkm]

The final question - for me - is: do any improvements of the GPS chip or OCXO provide either significant higher accruracy and/or stability of the generated 10 MHz signal, because that is the main purpose of my unit ?

[MegaVolt]

The final question - for me - is: do any improvements of the GPS chip or OCXO provide either significant higher accruracy and/or stability of the generated 10 MHz signal, because that is the main purpose of my unit ?

The output phase noise is determined mainly by the OCXO parameters. So choosing a better OCXO can give a noticeable improvement if a better one can be found.

[burkm]

That seems to be my "problem" right now, because the BG7TBL installed C-MAC OEM-OCXO 48S1164 (DC1212=> December 2012) drifts slowly downward (constant voltage) or needs a slowly increasing voltage to keep its frequency constant. This - probably - seems to be the original reason, why the C-MAC (now Rakon) OCXO has been replaced so "early". My Oscilloquartz DOCXOs (8663-XS) perform much better with respect to frequency stability compared to the Oscilloquartz OCXOs. After searching the internet for some time now, I still didn't find any specific information about this C-MAC OCXO. The unusual size (50.84 mm both dimensions) and the pin-out seems to be a specialty of the C-MAC OCXO, because I found similar units only in a rather "old" C-MAC catalogue (2000) of this manufacturer.
Unfortunately BG7TBL had its multiplier already set to x3 probably because of this drift. During measuring the C-MAC OCXO it never actually reached the 10 MHz...

[AndrewBCN]

The final question - for me - is: do any improvements of the GPS chip or OCXO provide either significant higher accruracy and/or stability of the generated 10 MHz signal, because that is the main purpose of my unit ?

GPS chip: no.

OCXO: exactly as MegaVolt wrote.

The output phase noise is determined mainly by the OCXO parameters. So choosing a better OCXO can give a noticeable improvement if a better one can be found.

Instead of "noticeable", I would say "measurable", but that of course entirely depends on what measurement apparatus you can get a hold of. And again: most people don't have access to instruments that can measure an OCXO's phase noise.

Just to give you an idea, here is the data for a Vectron OX-204 OCXO: "The OX–204 achieves a close in phase noise performance of −140dBc/Hz at the 10Hz offset with a phase noise floor of <-175dBc/Hz."
https://www.vectron.com/products/ocxo/ox-204.htm

To measure -140dBc/Hz phase noise you need rather expensive instruments. And if you have such expensive instruments at hand, I would guess you are probably not using a $150 BG7TBL GPSDO.

Btw even used inexpensive 10MHz 5V OCXOs such as the ones that can be found online are specified for −120dBc/Hz @ 10Hz phase noise and that too requires expensive instruments (that most of us cannot afford) to measure.

Finally, there is the matter of cost/benefit. A new OCXO with superior phase noise specifications will cost you more than the BG7TBL GPSDO itself - if you can buy it at all in single quantities. And it will have no measurable benefit for you, since you won't have the means to measure and compare the phase noise before/after.

That seems to be my "problem" right now, because the BG7TBL installed C-MAC OEM-OCXO 48S1164 (DC1212=> December 2012) drifts slowly downward (constant voltage) or needs a slowly increasing voltage to keep its frequency constant.
...

That is not phase noise, that is frequency drift. Either you are not waiting long enough for your GPSDO's operating temperature to stabilize, or you have a defective OCXO, or something else (e.g. temperature compensation, power supply, etc) is not working correctly, in which case you are better off just buying another "black box" GPSDO and hoping that it comes with a better OCXO and works within specs. Or you can build your own Open Source GPSDO for < 50€ and experiment with / improve on it as you see fit.

[rf-loop]

And find that every single xtal in XO or OCXO or DOCXO have its ovn individual aging curve. And it is not drift in oven temp but just xtal itself is ageing.

And then. If some OCXO have aged example year(s) running all times and then someone cut power and keep it some time without power... after then turn on... it have retract and is dropped more or less far back to history on its aging curve... but now... it recover much more fast back to its original nominal aging curve... how long this fast recovery takes... just who knows but here one week is short time. Xtals have also random hops. Other things may be also that owen teperature drift slowly due to components aging/drifting.
It is whole world where is not any single constant, everything is jumping randomly, drifting and walking more or less. (been there done that)

If in GPSDO OCXO control voltage do not slowly rise or fall over weeks and months... I can tell it have problem. Even if DOCXO is "wolds best" it have aging curve and retract if stored without power.

It need remember that after new power on it looks like aging more fast... because it typically drives back to the original aging curve much faster than the original aging is. As told,  time when it reach original aging curve is mostly unknown but after start long time used and long time cold stored OCXO this aging after new power up may be quite fast in first hours and days. (until it reach nearly this original aging)
There is one proposal. Do not turn OCXO off, keep it always constantly running. Days, weeks, months, years.  After power  off, wait enough time. There may exist more random jumps and "fast"  return to its individual already reached aging curve after retract. ("fast" mean much much faster than original aging)

Many peoples have alarmed when second hand OCXO drift so fast after they start use it. If there is not  other reasons still this aging retract alone can be surprisingly fast (related to normal long time aging after example 1 year run) in first hours and days. (naturally it can only look after Oven have reached full temperature equilibrium.
But then there is also some OCXO products what are not perfectly hermetically sealed, example models where is mechanical adjustment (even if they also have electrical tuning).

These theory and practice can find whi is interested about xtals and ocxo secrets... it also have mysteries what science today do not know exactly, many mysterious things in xtals. (and every xtal is finally just as individual as every human) 

[MegaVolt]

If we are talking about GPSDO, OCXO aging has no effect on the output parameters.

[AndrewBCN]

If we are talking about GPSDO, OCXO aging has no effect on the output parameters.

Exactly. Phase noise and other output parameters are not affected by OCXO aging.

[AndrewBCN]

...
It need remember that after new power on it looks like aging more fast... because it typically drives back to the original aging curve much faster than the original aging is. As told,  time when it reach original aging curve is mostly unknown but after start long time used and long time cold stored OCXO this aging after new power up may be quite fast in first hours and days. (until it reach nearly this original aging)
...

In OCXO jargon this is called "retrace". And yes, OCXO aging is a complex phenomenon affected by various variables. I wouldn't worry too much about OCXO aging, though, because it is exactly for that reason that GPSDOs operate with a closed loop (either PLL or FLL or a hybrid PLL/FLL) to control the OCXO frequency.

[burkm]

I did not talk about "phase noise", I did not even mention it (?)...

I talked abou frequency stability or drift, which I measured with my Rigol DM3688 DMM. Within 2 days (48h) the tuning voltage drifted (for almost constant frequency measured by my Fluke PM6685) from 5.24322 up to 5.24756 VDC. The runtime of the OCXO (GNSSDO) and Rigol DMM is currently more than 22 days without interruption. This would justify the x2 settings of the multiplier but it was already set for x3 by BG7TBL, which makes me wonder.
Accelerated retrace of a used OCXO should usually be finished after that period of time. I got an older version (BG7TBL GPSDO 2015-6-19), which retraced a lot faster after all, although the OCXO was correspondingly older.

May be there is someone here, who got a "standard" sized OCXO in his rather current BG7TBL GPSDO or GNSSDO, who has a higher resolution photo of the board, which I could use as a template for changing mine...

[AndrewBCN]

...
Within 2 days (48h) the tuning voltage drifted (for almost constant frequency measured by my Fluke PM6685) from 5.24322 up to 5.24756 VDC.
...

That is way too little data to reach any conclusion. 

[rf-loop]

...
It need remember that after new power on it looks like aging more fast... because it typically drives back to the original aging curve much faster than the original aging is. As told,  time when it reach original aging curve is mostly unknown but after start long time used and long time cold stored OCXO this aging after new power up may be quite fast in first hours and days. (until it reach nearly this original aging)
...

In OCXO jargon this is called "retrace". And yes, OCXO aging is a complex phenomenon affected by various variables. I wouldn't worry too much about OCXO aging, though, because it is exactly for that reason that GPSDOs operate with a closed loop (either PLL or FLL or a hybrid PLL/FLL) to control the OCXO frequency.

Yes as I told "Even if DOCXO is "wolds best" it have aging curve and retract if stored without power."
(I have used tens of years term retract independent of "Merriam-Webster")

And as you told it is norma, no need worry so much. But when someone wonder how it drift perhaps faster than he think is ok, reason may be this retract related phenomenon.

In cheap GPSDOs as example these many BG7TBL produced  are, there is not so much what average user can do  for optimize Discipline / LOOP parameters. As can do example in quite cheap old TrimbleThunderbolt and some others.

[AndrewBCN]

Yes as I told "Even if DOCXO is "wolds best" it have aging curve and retract if stored without power."
(I have used tens of years term retract independent of "Merriam-Webster")

And as you told it is normal, no need worry so much. But when someone wonder how it drift perhaps faster than he think is ok, reason may be this retract related phenomenon.

I entirely agree with you.

In cheap GPSDOs as example these many BG7TBL produced  are, there is not so much what average user can do  for optimize Discipline / LOOP parameters. As can do example in quite cheap old TrimbleThunderbolt and some others.

Yes, in these "black box" GPSDOs the average user has zero control over the firmware, the PLL parameters, etc. These GPSDOs are inexpensive, but not designed to be hacked or serviced by the average user.

[rf-loop]

...
Within 2 days (48h) the tuning voltage drifted (for almost constant frequency measured by my Fluke PM6685) from 5.24322 up to 5.24756 VDC.
...

And problem is?
It is its work, keep OCXO output "roughly" *) around 10MHz, it need discipline and 2 days... perhaps just after turn on, if so, it is totally nonsense and totally lack of enough data for anything.
Keep least one month constantly on or more... then you can see just bit some kind of curve how it goes and also perhaps some less change later.
Just look/watch it example using Lady Heather  (yes full use of it may need some experience, its UI is bit garbled), you get all trends etc how it goes.

*) yes I mean really just quite roughly.
But I am retired time nutter so it is better than I stop before I even start.

[burkm]

The tuning voltage of the OCXO after 3 days of measurements (this morning) has now reached 5.24963 VDC as shown by my labs DMM. The unit itself is up and running continously for almost 23 days already as stated before...

I checked the frequency output of the "new" BG7TBL GNSSDO frequently during that period (22-days +) and both displays (Fluke PM6685 + external star 4+ ATDC ref. freq. and the built-in LCD-display) - always - show frequencies below 10 MHz, which made me wonder, because the shown frequencies by my other GPSDOs usually "oscillated" around 10 MHz during that "startup" period (more or less), when arrived, after a few days run-time. Even my older BG7TBL GPSDO from 2015 provided a much "better" stability and accuracy measured by the instruments I got. I contacted the seller already in China by e-mail but got no satifying answer from him, because he refered me to BG7TBL.

I relate those experienced differences to the C-MAC OCXO installed in the "new" GNSSDO. This was the main reason for wanting to exchange the installed OCXO against one of those Oscilloquartz 8662-XS I got, which show a date code of "08/47" and which seemed to be much more stable after only several hours of runtime than the one installed, judging by the oscilloscope i hooked up to it. But due to the new and only partly populated PCB I have not been able to do this for various reasons.

How long would you expect me to wait for the OCXO to stabilize during retracing ? 30 days, 40 days or even more

I wonder how those units and their OCXOs have been tested by BG7TBL, if reaching stability of the OCXO will take that long ...
What made me wonder too, that the tuning voltage multiplier was set for "x3" right from the beginning although current measurements (see above) show only a need for "x2" including some proper headroom as a reserve.

[MIS42N]

I have been following various GPSDO threads for quite a while now, and I'd like to ask, what do people use their GPSDO for? It seems to me there are two groups, those that play with them, trying to make them 'better'. And those that use the GPSDO as their primary reference. Is the bg7tbl useful for either of these purposes? From this thread, it seems the bg7tbl is a changing object, so people buying it as a primary reference have no idea of what they are buying and cannot rely on tests by others because they are all different. And those that play have much better equipment and it seems pointless to try modifying a cheap box when they already have a better standard. Since I am close to finishing my GPSDO design, it would be interesting to know who, if anyone, would use it,  and for what.

[usagi]

cheap primary reference that is good enough.

it doesn't satisfy the level of time-nuts and volt-nuts metrology obsessions, but it's far beyond the requirements of almost any casual hobbyist. after all, most hobbyists don't even have a frequency reference at all.

if someone comes up with a better, cheaper gpsdo solution than bg7tbl, they should easily be able to take over the market for such devices.

[MIS42N]

cheap primary reference that is good enough.

it doesn't satisfy the level of time-nuts and volt-nuts metrology obsessions, but it's far beyond the requirements of almost any casual hobbyist. after all, most hobbyists don't even have a frequency reference at all.

if someone comes up with a better, cheaper gpsdo solution than bg7tbl, they should easily be able to take over the market for such devices.

Well I think that is what I am trying to achieve. I don't know about better, as I pointed out there are many bg7tbl versions. However, I am working on cheaper. My design criteria is to achieve a 10MHz output +-0.01Hz and be able to verify it. I am working on the idea that some people would do some construction, so the design is BYO power, BYO GPS module, BYO case. The actual GPSDO board I think can be populated for under AU$20. So far, it is all working and I am writing it up here . The first attempt at a PCB board has been produced and I am waiting for it to arrive.

[usagi]

i think most people buying bg7tbl aren't looking for BYO DIY. they want something inexpensive that you plug in and it works. at least that's why i bought mine.

[burkm]

My point of view is, that anything, i.e. a GPSDO or GNSSDO, must - at least - fulfill the manufacturer or seller provided "official" specs, because that is the main reason I am buying something. Sometimes even a 2nd time lateron, if something has changed in a - for me - relevant matter. Others seem to share this view, because they are "testing" what they got. If a seller gives a clear answer to a provided question, this is at least what I want to be delivered, when I receive the item bought.
I don't care about "improving" something, which follows the data sheet already, because this depends on so many - mostly unknow - data, that I wouldn't / couldn't care about it.