[time-nuts] Salvaged OCXO by ar electronique

[Andreas Seltenreich]

Bob kb8tq writes:

> If you have not already spotted a Vref pin on the unit, your magic 
> 8V pin may be the Vref for the EFC. It also could be “factory use only /
> do not connect “.

Thanks - I think I did spot such a pin: there's 5.3V suspicuously close
to the tuning input, and wiring a pot to it with the wiper to tune
appears to do the right thing.  I guess the 8V will have to remain a
mystery then…

Here is my current best guess in case someone else searches the archives
in the same situation:

             +-------------------+
 10 MHz sine | o               o | 12V --,
  1.3Vpp     |                   |       | connected to
             |  micrOSTAR-FB 509 |       | supply on PCB
             |     (top view)    |       | 250mA…100mA
         GND | o               o | 12V --^
             |                   |
             |                   |
     5V3 out | o                 |
        TUNE | o               o | Mystery 8V out, measures open on PCB
             +-------------------+

> There are OCXO’s on eBay in the < $30 range that should do about as
> well as what the spec sheet shows for the part you have. The unanswerable
> question is of course “which ones?”. As noted in many posts sorting
> through this and that to find a good part is never easy.

I skimmed the archives in that regard before, but thought I'd get this
DVB-modulator instead - the same price but lots of extra RF critters
(and a too-large-to-be-useful machined aluminum case).

Thanks also for the off-list mail from Tim about what to expect from the
tuning pin.  The entire unit actually had a 1PPS-input, but it went into
the FPGA on the digital board.  The connections were

+----------+            +----------------+
| RF board | --10MHz--> | digital board  | <-- SMA "1-pps"
| w/ OCXO  | <---IF---- | w/ 300MHz DAC  | <-- rs485
+----------+            +----------------+
 rs485--^

So maybe the oven was not disciplined and the digital side just shifted
the IF around to compensate.  Both had a PIC controller with a rs485
connection to the outside world though, so something external could have
talked to both and adjust a DAC to the oven, although I didn't spot a
discrete one on the PCB yet.

I don't have a frequency counter in the shack, so I improvised with my
HackRF (0.5ppm-TCXO).  To take a closer look at its performance and
tuning it, I'm squaring the OCXO with a 74AC04 and look at a
high-resolution-FFT of the 200th overtone.  The FFT-bins are 0.1Hz wide.
Here's a screenshot of the warmup.  The screen is 190Hz wide (apparently
gqrx can't draw sensible axis labels for such a high resolution FFT)

    http://aurora.npff.co/~andreas/ofen2.jpg (110kB)

I guess this is what the texbooks say a SC-cut warmup curve should look
like. The residual drift is probably the TCXO in the Hackrf: Touching
the HackRF causes dents in the curve while touching the oven doesn't
cause a visible change.

regards,
Andreas