Re: FeelElec New Arrival FY-6900 Signal Generator

[Bad_Driver]

I paid for mine about 80€, and it is worth the money. And while playing with it and with the help
of this forum I learned a lot about current arbitrary function generator technics. That is my benefit
from this little toy.

You can‘t expect HPAgiseight performance. As described you can simple fix the earth-ground wiring,
and put the fan into it‘s designated place. But it is a a very light plastic box, it slips on the desk with
every key pressed.

And it should be no problem for the vendor to improve the graphics and the fonts
and the shown digits on the display. There were enough advices in this board. But there is no improvement for a year or so.
At the end it‘s your decision, but I became during working with it the strong desire for a more professional
device, as mentioned I bought a Siglent SDG 2042 (which improved itself into a 2112 without soldering  )
and I‘m happy with it (but that costs about 6 times more).

At the end my conclusion is this: it does the job in many cases but it is no pleasure to work with it.
It is a good beginners device but you get appetite for more. I tried to stay fair in relation to the price.

Any comments?

[Labrat101]

Hi Bad-Driver
As you said is correct If one requires AWG that they spend more time fixing than using
its  ... OK .
  From my experience it was a good time waster . 
   Pros .. meet interesting friends on this forum .
   Con's  Yes  Too many to list.  at least 1200 blogs on this forum alone .
 But its the same as buying a clock That does not tell the correct time .
 It works, it chimes on the hour . But not on any particular Hour .   

Also it comes complete with all Fake chips ie the cyclone .. The Heart 

 Live long and Prosper spend the extra  bucks

[Johnny B Good]

 Incredibly, my modified FY6600-60M proved to be a considerable improvement in many ways over an SDG1032X that I'm now returning for a full refund on account of a rather strange and seemingly rare instability afflicting its internal clock reference (I suspect it's the same 25MHz +/-25ppm rated smd xo as used in the earlier SDG1025 model - the SDG2000X series use a 10MHz metal lidded +/-10ppm smd part for its internal clock reference).

 After discovering the work around solution of feeding its external 10MHz clock reference socket from a low phase noise GPSDO reference to eliminate this peculiar LF jitteryness in the internal clock (which, BTW, can be exported via this socket and show the exact same jitteryness), I discovered that the frequency control UI would only accept up to 8 digits worth of resolution when selecting each digit for adjustment via the rotary encoder.

 However, it is possible to enter a total of 11 digit's worth when using the keypad before pressing the multiplier key (MHz in my case) whereupon the placement of the MHz multiplier would stamp over the last three digits, causing the displayed frequency to be rounded to the nearest 8 digits even though it had actually accepted the full eleven digits that had been entered. Furthermore, the lack of the 'sticky digit cursor position of the FY6600's frequency selection UI adds even more pain to what is, quite frankly, an execrable UI experience (and there's even worse which I won't bore you with).

 Now you might be thinking, "But, Johnny, that's only the UI which can be fixed with a firmware upgrade - it's not fully set in stone, surely the rest of the specifications are considerably better than what Feeltech's "toy" has to offer?" Well... (a small) yes and (a large) no.

 The extra 200 quid didn't buy me any enhancements to the Feeltech's feature set (and this is for a modified FY6600 rather than the better FY6900!) that I couldn't live without. To achieve that state of AWG nirvana, you need to spend yet another 200 quid more on an SDG2042X before you can safely consign your FY6xxx to the spares cupboard.

 For instance, most, if not all of the DDS ARB functions are limited to a maximum frequency of just 6MHz versus the 18MHz they can typically be pushed to with little ill effect in the case of the Feeltech "toys" (the trueArb option for instance, limits the Sinc pulse function to a maximum limit of, not the 1.8MHz I'd initially assumed was being shown in that hard to read when selected frequency display but, incredibly, just a mere 1.8KILOHERTZ!! 

 These SDG1000X and 2000X series of AWGs do provide a nicely built fully shielded enclosure to prevent ingress of any EMI within the lab/workshop environment. However, that does introduce the issue of mains earth ground loops (solvable with a pair of heavy duty low voltage silicon diodes wired in anti-prallel with a 1K resistor across them placed in series with the protective earth connection to open circuit this king size magnetic loop antenna under normal operating conditions - the diodes short out any otherwise dangerous voltage in the event of a live contact with full mains voltage from any source, thus blowing the safety fuse involved or trip out any ELCB/GFCI circuit breakers at the CU). The low jitter full frequency square wave feature of the SDG1000X series, nice as it is, simply doesn't justify all the other performance shortfalls as a result of downgrading from an FY6xxx series AWG to this Siglent model range.

 For anyone looking to upgrade out the "cheap toy AWG" category, the best bang for your buck option appears to be Siglent's SDG2042X right now. Rigol stopped being any competition against Siglent some years back and it's the likes of Feeltech Siglent should be keeping their eye upon as of at least the past two years.

 Closer inspection of Feeltech's competitor products in the same 100 to 150 dollar price bracket just looks like a switch would be pretty much an exercise in swapping one frying pan of pain for another to my mind. If you've outgrown the limitations of an FY6xxx then it's time to stop wasting money on more of the same and contemplate, at a minimum, an SDG2000X series or one of the A brand offerings and spend some real money. 

 Harking back to LF jitter issue with the SDG1032X I received just over a week ago, I do believe I'd witnessed a similar phenomena with my FY6600 before I'd upgraded its crappy ten cents 50MHz XO chip almost two years ago to an off-board 50MHz 0.1ppm TCXO. Back then, the only other signal sources I could play with had been a collection of salvaged DIP14 XOs with which I'd played a game of "Chase Will 'O' the Whisp" trying to match frequencies against the FY6600.

 Not only was I fighting the execrable stability of a ten cent XO being roasted alive at 55 to 60 deg C from the 65 to 70 deg heat off a trio of LDO regulators just 1 cm away as well as the slower thermal drift in my DIP14 oscillators, I do believe I was also having to contend with a similar LF jitter effect in the ten cent smd XOs Feeltech's bean counters had insisted upon using. ISTR that this game of "Chase Will 'O' the Whisp" seemed a whole lot easier after upgrading to the 0.1ppm 50MHz TCXO board.

 I've long since replaced that upgrade with a 10MHz OCXO and a 3n502 clock multiplier chip. In the light of recent events regarding an all too familiar looking clock reference instability in the SDG1032X, I decided to repeat this game of"Chase Will 'O' the Whisp" with 5 and 16 MHz DIP14 oscillators to refresh my memory. I had little difficulty in matching frequencies and, after covering the solderless breadboard I'd plugged the oscillators into with a cotton rag to shield them from random cooling air drafts, I could only discern a small amount of frequency drift, comparable to the 50MHz 0.1ppm oscillator board I've just shot a ten second movie clip of by way of comparison to the jittery sine wave output from the SDG1032X I'd recorded last week.

 The 50MHz is the green CH4 trace compared to the GPSDO stabilised 10MHz sine (yellow CH1) and Sinc pulse (blue CH3) outputs from the FY6600 with the RFS connected to CH2 (purple). I'm triggering from CH1.

 I've attached both movie clips and I'd be interested if any of these look familiar to anyone who has ever tried playing this particular game of "Chase Will 'O' the Whisp" with any of the FY6900's predecessors (I doubt the FY6900's 10MHz XO would be similarly afflicted but you never know). BTW, if you haven't already guessed, I'm displaying the SDG1032X's output on CH2 in this video.
 
 I've had to append the .zip extension to the attached files in order to bypass EEVBlog's waste of time filename restrictions. For any newbies, after downloading them, either remove the .zip extension or else use the "Open as" option and choose VLC media player (or whatever your poison of choice happens to be).

John

PS I almost forgot to mention that I'm expecting delivery later this morning of an SDG2042X from my favourite T&M kit supplier.

[EDIT] It did arrive this morning (11:45 UTC). Pretty snappy delivery considering they'd only confirmed that they'd received the payment less than 24 hours ago.

[Labrat101]

Hi Johnny
I tried to open your ZIPs both seem to be corrupted .  Could you re do them .
That was a very long story .
I have said all along that the FY6600 cyclone chip was original .
The Fake chips were in the FY6800 .. &up
But I have found a new use for my FY6800 I have Locked the frequency @ 40 Hz Cmos Square and added a power transistor
to the output .  And I am using it as a Gamma Strobe for my wife who suffers from Dementia .
I left the OXCO in place and calibrated with my new Leo Bodnar GPSDO .
 And it has been running for about a 2 week  . It has help a little but as she is in the really last stages .
 Today I off set the frequency by just 0.05 hz .. WOW wont do that again . The brain sinc has to be the
same locked frequency zero drift . So it's back to 40.Hz 10^-9.  . This should have been done 8yrs ago.

 Sorry to hear that your Siglent AWG was not up to standard .
I have been very Happy with my New UTG 962 .
 I have not been able to reproduce any of the jitterery stuff that the FY6*** has ..
 The display as I call it is What u see is what you get on the scope exactly .
The oc is very good the spec says its 50ppm or less . my one is 30.4 ppm. Maybe later I might be temped to add an external
Reference  . mean time I just add the error in on the frequency setting and leave it .
I was very temped to go for the UTI ,s big version . which does have the ref in/out and it is far superior .
 But for my needs the UTG962 does every thing I want when I want  just peachy .
 The display input allows to add example 10.000,000,000,001 without any problem . but I have to add 30 ppm to the
value to make 100% perfect to my Leo 
But on every day stuff the general hobbyist would never notice this small error .
For the faint hearted never check the OC in your PC . its Horrible.

Johnny I would like to see your video's if you could up them again 

 

[DC1MC]

Labrat101, regarding your 40Hz Gama Strobe, I have some very bright leftover LED bars used for LCD backlight and a couple of 10MHz TCXOs, what is the waveform for this generator, sine, rectangular or some special shape, I have an elderly relative that could benefit form it.
Also, is really the 40Hz frequency that critical, that +/- 0.5 produces adverse effects ? How do thse effects manifests and actually is there any visible, measurable improvement when is the right frequency ?

Many thanks,
DC1MC

[Labrat101]

Labrat101, regarding your 40Hz Gama Strobe, I have some very bright leftover LED bars used for LCD backlight and a couple of 10MHz TCXOs, what is the waveform for this generator, sine, rectangular or some special shape, I have an elderly relative that could benefit form it.
Also, is really the 40Hz frequency that critical, that +/- 0.5 produces adverse effects ? How do thse effects manifests and actually is there any visible, measurable improvement when is the right frequency ?

Many thanks,
DC1MC

BTW This should really be made into a new post ..

Don't use the leds that are too bright .. that is not the object .
 Be careful on the chose of leds they are not all equal and can cause eye damage.
as some have high uv output .
There are many documents on the subject from Professors & uni's
The wave form is cmos square and the pulse must be 50% on 50% off .
Don't use the leds from a TV they have a lenses 
i found the best were from the torch with about 40 flat plan leds are very good
 they should be as close as possible to day light.
 the frequency is critical if the person suffers from fits or other problems
 Do not try it.
You need to research it .. its not a try & see . Yes it does help
 If you read the other post I did mention the wave form also.
Yes the frequency can not be change after you start the treatment

[Johnny B Good]

@Labrat101

 I've had a similar complaint about these "zip" files being corrupted before. I did explain why I'd had to append a false extension onto the ends of these mkv H.264 movie clips and the need to rename them to delete the final ",zip" endings. or else use the "open as" option in the right click drop down menu to choose a suitable media player rather than let it default to whatever is the default application for dealing with actual .zip files which will get all confused and report that this .zip file must be corrupted because it fails to recognise the expected metadata.

 Just by way of a test, I downloaded the second file and tried to open it with the default application which in this Linux Mint system is "Ark", and it refused with a "corrupted archive" error message. If you're using windows, then I'd guess this is what you're seeing with whatever you use for opening such compressed archives (PKunzip? or similar).

 Microsoft has always been scammer/malware/virus friendly with its habit of hiding the vital clues revealed by the filename extensions by default (Aping Apple's "user friendly" reliance on metadata rather than filename extensions to decide how files are handled by default). I suspect the .mkv part of the .mkv.zip ending is being neatly suppressed, leaving you to suppose that removing the .zip extension will prevent the default mediaplayer application from opening these files.

 All of that microsoft nonsense is why I offered two ways to avoid the very trap you appear to have fallen into. Since it's now been 6 years or more since I last used the only good version of windows ever released by ms (win2K Sp5, aka NT5.0), I'm now a little unfamiliar with the list of filetypes windows doesn't hide from view (especially as I'd long since disabled all this nonsense with file names in their previous best version of windows, win95OSR2).

 The only reason, afaics, for this issue of "Korrupted Downloads"(tm) is due to the EEVBlog's filter on allowed file extensions necessitating the use of fake extensions and the nonsense that is windows habit of hiding the critical file name extension types like .com .exe .dll .bat .lnk .pif and so on (example: receiving an email attachment with the name of "naked lady.jpg.exe" which windows disguises by hiding the .exe extension yet, curiously, it still treats as an executable program file regardless of the consequences, leaving the hapless victim wondering where Hell his naked lady picture has disappeared to).

 If you follow my instruction on how to open these attached files (either remove the fake .zip extension from the filename (and don't worry about not seeing the mkv extension) before double clicking on them to let your default media file handler open them or else right click and select an appropriate mediaplayer - VLC is an excellent choice in this case), you will be able to watch these movie clips.

 BTW, if anyone else has any better suggestion on how best to get past EEVBlog's attachment filename filtering, feel free to chime in.

John

[Labrat101]

   @ Johnny B Good

Ok . I removed the Zip part and they worked fine.
 If you had put TXT at the end it would have made it a little more obvious 
I watched them OK compared to my FY68 with the fake chip I get a double sine one is stationery and the other slides over the top  the square on my one is horrible over 5Mhz.
That square wave you have the sloping top & bottom is from the relay contacts put a very small cap 8pf or 10pf before the relay input will solve that .
 But I have locked it now for one purpose ONLY and @ 40hz cmos it is working fine . with only a small tweet.

 BTW Thanks . I also moved away from windows about 15 years ago .  Now on Mint 20 
I have a laptop that I bought about 5 years ago that came with the shity 10 .
 which I only use for Multisim & Texas Tina-Ti.
and my wife's computer which is windows 7 as it has her game that she has played for nearly 8yrs 
 and with Dementia can't be changed . (still on level 1).

 Now with the Pulse strobe & VLF music  there is a slight improvement 
 I have to keep an eye on the VLF pulse as it can make things move like Dr. Who's sonic screw driver.
 
  RNS

After an hour in the same room . So Relaxing  make U feel like your out of body

[Johnny B Good]

   @ Johnny B Good

Ok . I removed the Zip part and they worked fine.
 If you had put TXT at the end it would have made it a little more obvious 
I watched them OK compared to my FY68 with the fake chip I get a double sine one is stationery and the other slides over the top  the square on my one is horrible over 5Mhz.
That square wave you have the sloping is from the relay contacts put a very small cap 8pf or 10pf before
the relay input will solve that .
 But I have locked it now for one purpose ONLY and @ 40hz cmos it is working fine . with only a small tweet.

 BTW Thanks . I also moved away from windows about 15 years ago .  Now on Mint 20 
I have a laptop that I bought about 5 years ago that came with the shity 10 .
 which I only use for Multisim & Texas Tina-Ti.
and my wife's computer which is windows 7 as it has her game that she has played for nearly 8yrs 
 and with Dementia can't be changed . (still on level 1).

 Now with the Pulse strobe & VLF music  there is a slight improvement 
 I have to keep an eye on the VLF pulse as it can make things move like Dr. Who's sonic screw driver.
 
  RNS

After an hour in the same room . So Relaxing  make U feel like your out of body

 I'm glad you solved the problem and that suggestion to use a fake .txt extension instead has been made before so, in view of the fact that it's a fake extension anyway, I'll take you up on that advice. I guess I was being just a little too conservative in sticking with the .zip idea out of the seeming logic that .zip files can contain any type of file or files - unfortunately, that's only valid when you actually take the trouble to encode them as real .zip container files.

 I've just been looking through my earlier attempts at attaching short video clips and I've just realised why I'd chosen to use .zip as a fake extension. My very first video clips had been actual zip files, requiring the recipient to unpack them into their original .mkv format. I then received a suggestion that I could simply append a fake extension to get past EEVBlog's file type filter, hence my sticking with the .zip as a fake extension, little realising that this would be more trouble than using something like .TXT From now on, it's .TXT all the way

 I don't envy your situation but no one can know what's ultimately in store for them and their loved ones. You just have to manage whatever comes along as best as you can since you have little choice in such matters. Whilst that AWG leaves a lot to be desired as a piece of test gear, it's good to see that you've managed to repurpose it rather than send it straight to landfill or clutter up cupboard space just gathering dust. BTW, it looks like you've found yourself a drugs free alternative to the narcotic experience.

John

[Labrat101]

@ Johnny

All is good . I have been following an idea from a Nobel prize winner Pro.
 and I am working on the VLF - EMIC which was designed about 2 decade ago.
   Using 7.38Hz which was proven to work . The usage placement differs .
 I did think of opening a New post .. But I think that this field is outside the
 scope of this forum . & not for Novice .
 
 RNS  ¹⁰¹

[gameru]

Good day
I want to buy a Function Arbitrary Waveform Signal Generator and i can't decide between FY 6900 and Fy 6200
Can someone tell me the diference between them?
Thank you

[davebb]

Hi when you set the internal osc frq offset for 10mhz in the secret system menu, how do I save it as the cal setting that I set is lost when I switch off then on
Thanks For any Help
Dave 2E0DMB

[Labrat101]

Hi when you set the internal osc frq offset for 10mhz in the secret system menu, how do I save it as the cal setting that I set is lost when I switch off then on
Thanks For any Help
Dave 2E0DMB

Long press on the OK button ..

[davebb]

Thanks for the fast reply,
Thanks Dave

[davebb]

Hi I have also done a mod to connect my gpsdo that is using a converted multi outlet 50ohm old video amp,
I cut the track going from the output of the tcxo and put a switch so I can select the gpsdo or the on board tcxo, the Problem I am having  is that when I am connected to my gpsdo and switch on off any of my mains test eqt the 6900 stops outputting a signal , I have tried adding caps and rf chokes / inductors, grounded  copper tape over the tcxo and the cpu, this did not help, changing input level to the 6900,
can anyone give me any other things to try ?
Thanks Dave 2E0DMB

[pizzigri]

Hi when you set the internal osc frq offset for 10mhz in the secret system menu, how do I save it as the cal setting that I set is lost when I switch off then on
Thanks For any Help
Dave 2E0DMB

Long press on the OK button ..

I am not familiar with this and after a little while trying to sort it out I could not find the "hidden service menu"... as I have a large offset from the FY6900 due to my upgrade of the stock crap XO with a 0.9 PPM TCXO, I'd like to zero the offset so that when I dial 10MHZ I do get it instead of 10.000235...
now I calibrated my counters to the GPSDO and it's quite obvious, it does not drift like it did before but  the offset is... off putting.
Labrat, could you elaborate more on it?
Cheers!
Franco

[davebb]

Hi you have to press the on button and the ch1 button at the same time when switching the power switch at the back , then once started press the system button then the more button
once you have adjusted it you use the ok button to save
yes the STD tcxo does drift a bit,
Let me know how yours goes, the STD one is of by 263hz at 10mhz,
Dave 2E0DMB

[pizzigri]

that is GREAT!
I now have my crappy FY6900 pulling out frequencies with an offset of 0.1 ppm, and with the upgraded TCXO i placed inside it doesn't even drift too much.... at least today, I'll leave it on for a couple days connected to my conter (with GPSDO external ref attached) and let's see how it fares.
But it's amazing how better that is, it always bugged me how the thing was always off by 230Hz!!!!

THANK YOU guys!

[davebb]

that is GREAT!
I now have my crappy FY6900 pulling out frequencies with an offset of 0.1 ppm, and with the upgraded TCXO i placed inside it doesn't even drift too much.... at least today, I'll leave it on for a couple days connected to my conter (with GPSDO external ref attached) and let's see how it fares.
But it's amazing how better that is, it always bugged me how the thing was always off by 230Hz!!!!

THANK YOU guys!

Hi was the upgraded TCXO a swap in to the same place with no mods ?
Thanks Dave

[pizzigri]

Well I did a mot of mods actually, with the help of EEV members...

one was the drop-in replacement of the original crystal with a 0.9 PPM TCXO, this one:
VC-TCXO IQD LFTVXO009900BULK
suggested by MasterX81, it is somewhat expensive at about 14 Euro shipped, purchased thru ebay.
It can be replaced without any other mod, and it works, as I have basically no drift. However, as you imagine, I had a very noticeable offset...
also, if you read previous posts, it may "break" the Magic thingy that FY introduced to reduce jitter.

[pizzigri]

So I actually am in search of a decent generator to replace the FY.
The  Siglent SDG 2042 is a prime candidate, however I do not have the money to purchase it new, and used are nowhere to be found, at least unless I find a good deal...

[davebb]

Well I did a mot of mods actually, with the help of EEV members...

one was the drop-in replacement of the original crystal with a 0.9 PPM TCXO, this one:
VC-TCXO IQD LFTVXO009900BULK
suggested by MasterX81, it is somewhat expensive at about 14 Euro shipped, purchased thru ebay.
It can be replaced without any other mod, and it works, as I have basically no drift. However, as you imagine, I had a very noticeable offset...
also, if you read previous posts, it may "break" the Magic thingy that FY introduced to reduce jitter.

ok thanks for the info,
Dave

[davebb]

Well I did a mot of mods actually, with the help of EEV members...

one was the drop-in replacement of the original crystal with a 0.9 PPM TCXO, this one:
VC-TCXO IQD LFTVXO009900BULK
suggested by MasterX81, it is somewhat expensive at about 14 Euro shipped, purchased thru ebay.
It can be replaced without any other mod, and it works, as I have basically no drift. However, as you imagine, I had a very noticeable offset...
also, if you read previous posts, it may "break" the Magic thingy that FY introduced to reduce jitter.

Hi pizzigri
How long does it take your 6900 to give a stable / no drift frq output with the new TCXO fitted ?
Thanks Dave

[pizzigri]

About half an hour or so. I added a fan to the housing and also used some Raspberry Pi heat sinks on the badly placed LDOs that are near the TCXO location.

[Johnny B Good]

Hi I have also done a mod to connect my gpsdo that is using a converted multi outlet 50ohm old video amp,
I cut the track going from the output of the tcxo and put a switch so I can select the gpsdo or the on board tcxo, the Problem I am having  is that when I am connected to my gpsdo and switch on off any of my mains test eqt the 6900 stops outputting a signal , I have tried adding caps and rf chokes / inductors, grounded  copper tape over the tcxo and the cpu, this did not help, changing input level to the 6900,
can anyone give me any other things to try ?
Thanks Dave 2E0DMB

 That sounds just like the problem I used to have with my FY6600 over 18 months back whenever I connected it to any hard grounded DUT or my DSO. I cured it by upgrading the cheap 'n' nasty C8 figure of eight mains socket to a C6 cloverleaf (trefoil) socket as used with most laptop charging bricks - you can get the more flexible 6A mains leads for these. and wiring a 10KR between the PE tag and the BNC ground rail to kill off the 120Vac leakage from the Y cap which had been creating an ESD transient upon making contact to the DUT's grounded connection almost every time with sufficient energy to corrupt the FPGA's current processing task.

 Most times you could disable then enable the channel output or switch between menu items to get it to resume but sometimes you'd have switch off and back on again, thankful for the FY6600's mercifully brief 6 seconds boot up time.

 If you haven't already done so, try applying the 4k7R grounding mod (after making good Feeltech's gross vandalism on the ground return wire in the ribbon cable joining the psu to the main board). You'd think leaving Feeltech's grounding vandalism intact would prevent this but connecting to kit powered from a cheap wallwart such as a solderless breadboard could induce a similar ESD event as I found out when connecting to my not yet fully grounded GPSDO fed from an as not yet modified wallwart blessed with a real but unused brass earth pin. I could have simply run a ground wire to directly earth the GPSDO and any associated distribution cabling but you really want to avoid creating low impedance earth loops into any coax based distribution cabling scheme.

 In your case, it would seem what you've described is a crude but simple changeover switching arrangement between clock sources (the internal 10MHz TCXO and an external GPSDO reference) wired directly to this switch providing a direct injection path for any common mode interference to appear on the FPGA's clock input line via the GPSDO cable connection.

 If all your kit, including the GPSDO, has a hard connection to the mains protective earth, you've created an almost ideal demonstration of the problem, so familiar to Hi-Fi enthusiasts, of ground loops (or as they naively call them, "Hum Loops"). In our case there are far greater issues than mere mains hum and its harmonics to worry about as I believe you're only just now discovering.

 When I added an external 10MHz clock reference input socket to my FY6600, may main concern had been the several milliseconds break in clock supply to the FPGA during the switching action of a simple mechanical change over switch, hence my designing a complicated injection locking circuit that would automatically and gently lock the OCXO into phase with the external reference by simply plugging the external reference into the socket to avoid any such risk of disruption. Likewise, unplugging the external reference would gently release the OCXO back to its own devices, also without any disruption.

 I'd overlooked the bleedin' obvious of phase loop locking it via the EFC connection otherwise I would have implemented the changeover between internal and external clock sources using that method which would also have provided a similarly glitch free changeover characteristic but with a wider lock in range than the +/-10ppb I'd been aiming for. Either way, injection or phase locking to an external reference, provides considerable immunity to any interference from mains borne switching transients that might be induced into the lab's clock reference cabling.

 If you have used a simple mechanical changeover switch with no transient limiting protection between the raw external clock signal and the FPGA's delicate clock input, you run the real risk of permanent damage to the FPGA. This is one modification you have to be really careful of in its implementation even if it's just the use of a disposable 74HC14 with two or more of its inverters cascaded to act as a buffer between the GPSDO input socket and the changeover switch (not forgetting to use a resistor attenuator network between the 5 volt 74HC14's output and the 3.3v clock input pin of the FPGA).

HTH and HAND

John