Topics

OCXO/Si5351A Synth kit - The wandering minstrel... #ocxo

Mike Berg
 

I enjoyed building the oven stabilized synthesizer, but in operation it has some annoying frequency excursions that I can't explain - yet.
I've seen up to 500 Hz difference in the reference frequency when the unit is self calibrating and the drift seems to vary randomly.
The oven seems stable and is set to 45C using a thermistor mounted over the adjustment hole and wrapped in insulation.

I suspect a bad crystal or one of the caps in the oscillator circuit, but they are a bit difficult to get at :)

I plan to have my brother, the machinist, mill the top off of the covers and take a look inside.
When I replaced the ocxo with the stock Si5351A synthesizer, it works so well I wondered why I bothered - however, when the seasons change, I'm sure I'll be reminded.

Has anyone one else had this problem with the OCXO?
73,
Mike N0QBH

Alan G4ZFQ
 

I suspect a bad crystal or one of the caps in the oscillator circuit, but they are a bit difficult to get at :)
Mike,

I had similar, finally replaced one of the 3 capacitors associated with the crystal. C6 comes to mind, but any could have been faulty.
Hans always says components do not fail, this one could have been slightly cracked when fitting? I did resolder all joints first.

As to dismantling after faithfully soldering all seams, I did it eventually with an iron and razor blade and some wick. Then just used a few tacks in case I needed to do it again.
Having first obeyed Hans I am not so sure of the need to solder all the seams.

73 Alan G4ZFQ

Alan de G1FXB
 

Hi Mike,
A thermistor mounted through the adjustment hole is not measuring the temperature of the inner enclosure contents that accurately. just it's outer surface temperature.
Ideally you want to get inside the inner chamber, it's the physical Xtal can temperature that gives rise to the expression of the target range being circa 45 degC.
The recommended calibration method is by observing the frequency shift rather than being a specific temperature, which is perhaps a distraction.

If you are unable to find the sweetspot for your actual Xtal, then it could be an out of tolerance component.
Xtals have being discused as one possible cause, but it's not epidemic.
In perspective they are the identical ones used in all QRP-Labs kits so well in excess of 10,000 pieces employed worldwide, there are bound to be the occasional bad one.
It's just as likely, it's something else......
Just because it's an OCXO kit, do not negate the requirement for clean & stable power supplies,
In the case of the OCXO, the heater circuit makes a heavy demand that you perhaps don't want on the main 5v one.

I would be careful on the choice of milling, the stresses of the cutter may tear the copper traces off the main PCB should it catch.
It's not a show stopper if it's isolated to just the lands securing the inner / outer enclosures,
but could be if it also damaged ones required electrically elsewhere.
Also the milling operation will result in amount of conductive copper dust that will require careful removal, trapped between and underneath components.....
Perhaps consider removing as much solder as you can using whatever is your preferred method of solder removal, if you can release just a portion it's possible to inch the assembly bit by bit being careful not to lift the traces. If you start to get impatient it's a sign for the need to walk away for a while.
Perhaps also investigate a product called Chipqwik, (it's approx halfway to the cost of a replacement kit if it's the only reason for purchase, so may not make sense.)
The principle is that it alloys with the remaining solder and "magically" much reduces it's melting point that will perhaps allow easier removal of the enclosures as a whole piece?


Alan

On 25/07/2019 13:16, Mike Berg wrote:

I enjoyed building the oven stabilized synthesizer, but in operation it has some annoying frequency excursions that I can't explain - yet.
I've seen up to 500 Hz difference in the reference frequency when the unit is self calibrating and the drift seems to vary randomly.
The oven seems stable and is set to 45C using a thermistor mounted over the adjustment hole and wrapped in insulation.

I suspect a bad crystal or one of the caps in the oscillator circuit, but they are a bit difficult to get at :)

I plan to have my brother, the machinist, mill the top off of the covers and take a look inside.
When I replaced the ocxo with the stock Si5351A synthesizer, it works so well I wondered why I bothered - however, when the seasons change, I'm sure I'll be reminded.

Has anyone one else had this problem with the OCXO?
73,
Mike N0QBH


Mike Berg
 

Thanks for the advice, Alan.
I realize the temp measurement location was probably less than ideal, however, with the insulation I used, the heat had very few options as where to go.

I had dialed in (or thought I did) the OCXO on a test bed using a PIC to control the Si5351 while monitoring heater amps and zero beating against WWV 10 MHz signal.
I'm sure I would have seen the excursions there if I'd let it burn in longer - before everything was soldered :)

Anyway. the autopsy is mostly to look around.  I'll probably order another and give it a go so that my U3S behaves when the weather turns colder.

No complaints really.  At Han's prices we can afford to experiment.
Mike

Alan de G1FXB
 

Hi Mike,
Sorry!

I question choice of a "beating against" method just so it matches something else, if this is your sole setup procedure.
The aim is to find the temperature the raw 27MHz xtal shows the flattest response to change.

(In my mind you are tuning the xtal so that the post 5351 the 10MHz output compares closest to WWV.
effectively substituting the traditional physical trimmer capacitor for trimming with temperature change?
(I know the build manual has a chart showing accuracy also compared to WWV.
it's just that, once the xtal sweetspot has being found, a plot of typical performance Vs. ambient temp change.
It's not that an alternative that it's possible to calibrate against WWV)

I know not what the PIC code contains in your application.
Apologies, if I have interpreted your methods very, very wrong....)

With a U3S, depending on your use perhaps over 95% will find the OCXO an unnecessary complication.
users report the vanilla 5351 kit OK up to 6Mtrs, some have at least performed tests on 2M even if not long term use on WSPR.
(Others please report practical findings.)
If you are using WSPR chances are high you also use a GPS if for no other reason of taking control of the timing aspect for this mode.
GPS stabilization to frequency also works well with WSPR, and will tame most all changes in ambient shack temp given you employ just normal good practices.
The limiting factors are just as likely the stability of the RX station.....

There is a reason to employ the OCXO kit for instance higher VHF use, and if transmitting long duration modes such as QRSS etc.
And a multitude of users own specific applications.



Alan

On 25/07/2019 15:14, Mike Berg wrote:

Thanks for the advice, Alan.
I realize the temp measurement location was probably less than ideal, however, with the insulation I used, the heat had very few options as where to go.

I had dialed in?(or thought I did) the OCXO on a test bed using a PIC to control the Si5351 while monitoring heater amps and zero beating against WWV 10 MHz signal.
I'm sure I would have seen the excursions there if I'd let it burn in longer - before everything was soldered :)

Anyway. the autopsy is mostly to look around.? I'll probably order another and give it a go so that my U3S behaves when the weather turns colder.

No complaints really.? At Han's prices we can afford to experiment.
Mike


Chris Wilson
 

Hello Alan,

Thursday, July 25, 2019

I have often thought it would be a nice option for future kits to
have an input for a 10MHz GPSDO as these are now pretty commonplace
and cheap.







pretty

Best regards,
Chris 2E0ILY mailto:chris@...


AdGvGI> Hi Mike,
AdGvGI> Sorry!

AdGvGI> I question choice of a "beating against" method just so it
AdGvGI> matches something else, if this is your sole setup procedure.
AdGvGI> The aim is to find the temperature the raw 27MHz xtal shows
AdGvGI> the flattest response to change.

AdGvGI> (In my mind you are tuning the xtal so that the post 5351 the
AdGvGI> 10MHz output compares closest to WWV.
AdGvGI> effectively substituting the traditional physical trimmer
AdGvGI> capacitor for trimming with temperature change?
AdGvGI> (I know the build manual has a chart showing accuracy also compared to WWV.
AdGvGI> it's just that, once the xtal sweetspot has being found, a
AdGvGI> plot of typical performance Vs. ambient temp change.
AdGvGI> It's not that an alternative that it's possible to calibrate against WWV)

AdGvGI> I know not what the PIC code contains in your application.
AdGvGI> Apologies, if I have interpreted your methods very, verywrong....)

AdGvGI> With a U3S, depending on your use perhaps over
AdGvGI> 95% will find the OCXO an unnecessary complication.
AdGvGI> users report the vanilla 5351 kit OK up to 6Mtrs, some have
AdGvGI> at least performed tests on 2M even if not long term use on WSPR.
AdGvGI> (Others please report practical findings.)
AdGvGI> If you are using WSPR chances are high you also use a GPS if
AdGvGI> for no other reason of taking control of the timing aspect for this mode.
AdGvGI> GPS stabilization to frequency also works well with WSPR,
AdGvGI> and will tame most all changes in ambient shack temp given
AdGvGI> you employ just normal good practices.
AdGvGI> The limiting factors are just as likely the stability of the RX station.....

AdGvGI> There is a reason to employ the OCXO kit for instance higher
AdGvGI> VHF use, and if transmitting long duration modes such as QRSS etc.
AdGvGI> And a multitude of users own specific applications.



AdGvGI> Alan


AdGvGI> On 25/07/2019 15:14, Mike Berg wrote:
AdGvGI>
AdGvGI>
AdGvGI> Thanks for the advice, Alan.
AdGvGI> I realize the temp measurement location was probably less
AdGvGI> than ideal, however, with the insulation I used, the
AdGvGI> heat had very few options as where to go.
AdGvGI> I had dialed in?(or thought I did) the OCXO on a test bed
AdGvGI> using a PIC to control the Si5351 while monitoring
AdGvGI> heater amps and zero beating against WWV 10 MHz signal.
AdGvGI> I'm sure I would have seen the excursions there if I'd let it
AdGvGI> burn in longer - before everything was soldered :)
AdGvGI> Anyway. the autopsy is mostly to look around.? I'll probably
AdGvGI> order another and give it a go so that my U3S behaves when the weather turns colder.
AdGvGI> No complaints really.? At Han's prices we can afford to experiment.
AdGvGI> Mike

AdGvGI>



--
Best regards, Chris Wilson (2E0ILY)

Mike Berg
 

My previous method using the PIC involved setting the output to 10.000MHz connected to a relatively accurate frequency counter for monitoring purposes.
Then I ramped the oven temp up to about 40C for a place to start. The outter shield wasn't soldered yet and I had the thermistor taped to the outside of the oven above the crystal.
This was wrapped in foam for insulation in order to stabilize my readings.
I had a current meter in series to monitor the oven's state.
I tracked the freq shift down til it leveled out and my temp was showing abit under 45C.
I waited for everything to stabilize which took 20 min or so.
My PIC code has a calibrate function so I entered in the offset to make my counter show 10.000MHz.
Then I played with that while using a receiver to zero beat WWV, calibrating both the PIC and the counter in the process (for the moment, it's a $100 counter :)
I didn't notice the wayward drifting til I started using it in the U3S and racking up some hours.
I hope this clarifies things.
73
Mike 

Alan de G1FXB
 

Hi Mike,

If your wayward wandering minstrel is more U3S specific, than in your custom PIC project:-
http://qrp-labs.com/images/appnotes/AN001_A4.pdf
Has suggestions on what's known in that application


Alan

On 25/07/2019 20:50, Mike Berg wrote:
My previous method using the PIC involved setting the output to 10.000MHz connected to a relatively accurate frequency counter for monitoring purposes.
Then I ramped the oven temp up to about 40C for a place to start. The outter shield wasn't soldered yet and I had the thermistor taped to the outside of the oven above the crystal.
This was wrapped in foam for insulation in order to stabilize my readings.
I had a current meter in series to monitor the oven's state.
I tracked the freq shift down til it leveled out and my temp was showing abit under 45C.
I waited for everything to stabilize which took 20 min or so.
My PIC code has a calibrate function so I entered in the offset to make my counter show 10.000MHz.
Then I played with that while using a receiver to zero beat WWV, calibrating both the PIC and the counter in the process (for the moment, it's a $100 counter :)
I didn't notice the wayward drifting til I started using it in the U3S and racking up some hours.
I hope this clarifies things.
73
Mike?

Mike Berg
 

"With a U3S, depending on your use perhaps over 95% will find the OCXO an unnecessary complication."

Agreed 100%. The stock synthesizer works great with my U3S.

73
Mike

Hans Summers
 

Even I agree 100%. And I designed the thing :-/

73 Hans G0UPL

On Fri, Jul 26, 2019 at 8:39 PM Mike Berg <mikeberg@...> wrote:
"With a U3S, depending on your use perhaps over 95% will find the OCXO an unnecessary complication."

Agreed 100%. The stock synthesizer works great with my U3S.

73
Mike