F6500 GPSDO as external reference for F1500

  • 1
  • Question
  • Updated 6 years ago
  • Answered
G'day all,
Had a play with this just now.
6500 GPS locked and using 10MHz reference on the 1500.
Look at 10MHz leakage in DSB 10 Hz filter in panascope.
I had hoped that they might be synchronous.
But not so. A calibration gave a DDS offset of 3.
With the scope set to 1000000ms I can see an error of 17 cycles per 1000 seconds.
or 0.017Hz. The DDS offset is limited to whole numbers.
I suppose 1.7 cycles per 10E9 is not so bad.
Andrew
Photo of Andrew Russell

Andrew Russell

  • 270 Posts
  • 39 Reply Likes

Posted 6 years ago

  • 1
Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9153 Posts
  • 3497 Reply Likes
Andrew, this is probably related to the FLEX-1500. Neither my FLEX-5000 or the FLEX-1500 have a DDS offset of 0 when I use my Rb reference. It is always off by a slight bit. The FLEX-1500 is usually +6 and the FLEX-5000 is +2. Also remember that the FLEX-1500 has a low thermal mass and it will be sensitive to temperature fluctuations.
Photo of Graham - KE9H

Graham - KE9H, Employee

  • 28 Posts
  • 22 Reply Likes
Official Response
Andrew:

A DDS is not a continuous tuning device. It generates frequencies in discrete steps.

For the DDS as used in the 1500, when run off a 10 MHz reference, the step size is every ~93.1 milli-Hertz.(actually, exactly 400 MHz / 2^32) So, if you chose the step closest to the desired/requested frequency, you can be as much as half of that, or up to about 46 milli-Hertz in error. Sometimes you will be exactly on, but usually you will have some small error in the tens of milli-Hertz. And, it will be different on every frequency tuned. So, if you correct it with the DDS offset, then change frequencies, it will be the wrong again. I would leave the DDS offset in the 1500 on Zero when running from an external standard. That will give you the lowest average error across all frequencies.

The NCO's (Numerical Controlled Oscillators) in the Flex-6500/6700 will also have some similar granularity in the actual frequency steps generated.

--- Graham / KE9H