[Lowfer] Experiments with an OCXO driving a DDS

[Ben Hall]

Good afternoon all,

In the earlier set of posts with subject "LOWER receiver conundrum," I 
discussed the possibility of using a DDS chip with an OCXO or GPSDO 
instead of the on-board (and drifty) oscillator.

I got this working this afternoon.  ;)

The GPS module is one of the AD9851-based units available on eBay.  I 
chose the AD9851 because it has a 6X internal multiplier, which drops 
the range of REFCLK frequency input into the range where a 5 to 10 MHz 
input will work properly.

The module:  <http://r.ebay.com/Z2vu8I>

I recall paying less than $25, so it may pay to shop around.

The controller is based on KT1F's Arduino DDS controller described here:

<http://www.theladderline.com/inexpensive-ad9850-dds-boards-ebay>

and here:

<http://www.theladderline.com/dds-60>

I had some problems getting it to work right, but it turns out my choice 
(junkbox) of mechanical encoder wasn't well matched to Ross' software. 
This was easily fixed by opening up the encoder, removing the detent 
spring, and putting it back to together.

(My encoder had the detents positioned so that both switches were open. 
  For each "clicky" I'd move the encoder, the switches changed state 
multiple times, causing the control to be very jumpy.  Removing the 
detent spring fixes this nicely.)

The OXCO I'm using at the moment is a Millren Technologies 260-series 5 
MHz unit.  Eventually, I'll get the HP Z3801 back online and will feed 
all this with its 10 MHz GPSDO output.

The OCXO feeds a sine-to-square wave converter circuit that follows Todd 
Robert's 74HC04-based design in his modifier AMRAD LF converter design. 
  This converter is built "dead-bug / manhattan style" on a small square 
of copper clad board.

The output from the sine-to-square is fed by a small piece of RG-174 to 
the pads on the AD9851 board where the 30 MHz oscillator used to be 
before I removed it.  ;)

As one would expect, the controller software reads 6 times the actual 
output frequency of the AD9851.  (Original crystal osc was 30 MHz, the 
MTI OCXO is 5 MHz)

Inspection of the output signal on the scope shows a nice sine wave.

Next up:  I need to see how drifty this setup is.  Initial measurements 
on the MTI OCXO showed it may have some drift...but the drift is likely 
less than what I'm able to measure with my counter.  Another reason to 
get the HP Z3801 back online:  the counter can accept the 10 MHz output 
as the time-base input, which would make drift measurements more accurate.

Thanks much and 73,
ben, kd5byb