[hpsdr] Resolved RE: Further Information: Help Needed: PowerSDR TX FrequencyInaccuracy

Tokio Endo rfcgf600 at ybb.ne.jp
Mon Jan 14 06:44:02 PST 2013

            Hi Joe, Dave,

  Thank you for your comments.
  Dave's reply reminded me of some discussion on the list. I searched
e-mails by a keyword  "frequency", then I found mails whose title is
"Frequency Calibration in PowerSDR" dated in 2012, two years back.

  I've never done the frequency calibration with HPSDR neither with Hermes
so far.

  Joe's case is quite good -0.000 00000 030 at 3530KHz, by that my OCXO is
very poor, but the transmit frequency difference between HPSDR and Hermes
exits in may case.
  So the suspicious part in case of my Hermes might be OCXO used, so I used
two different OCXOs for Hermes there was no change between them. However I
didn't verify the clock just in the Hermes board.
  There was a loose contact in the connector of 10MHz. The output level of
OCXO is 1.7dBm terminated to 50ohm, but the level at the board e.g., at C215
was around -30dBm.
  That was the cause. Now the transmit frequency by Hermes is quite correct
in terms of my OCXO accuracy.

  Thank you again.

  Tokio Endo

    -----Original Message-----
    From: Joe Martin K5SO [mailto:k5so at valornet.com]
    Sent: Sunday, January 13, 2013 3:23 AM
    To: Tokio Endo
    Cc: hpsdr at lists.openhpsdr.org
    Subject: Re: [hpsdr] Further Information: Help Needed: PowerSDR TX

    Hi Tokio,

    The absolute frequency accuracy issue has been discussed here
previously.  As Dave mentioned, the finite size CORDIC steps necessarily
result in less-than-perfect frequency output for some frequency selections
and this is highly dependent upon the exact frequency specified.  Some
frequencies are essentially perfect whereas others suffer somewhat, just as
you indicate.  However, I have never found any frequency more than 1Hz off
here, with most frequency selections yielding errors of less than a few tens
of milli-Hertz absolute error.

    The specific frequency of 3.530 000 you mention yields only a -0.000 000
030 MHz error (-30 milli-Hertz error) here on my system when running with a
Rb-vapor frequency standard into the 10 MHz ext 10MHz clock port on Hermes
(and properly selected via J20 on Hermes, of course!), as measured using an
HP-5345A counter locked to a second Rb-vapor frequency standard.

    If, on the other hand, I do not use the ext 10 MHz input ref clock but
use the on board 10 MHz osc on Hermes I observe a +2.903 Hz error at 3.530
000 Tx; not surprisingly.

    I don't observe any unreasonalble/unexpected behavior at all with
frequency accuracy here.

    73,  Joe K5SO

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20130114/eda76fa1/attachment-0003.htm>

More information about the Hpsdr mailing list