[Ham-Mac] Accuracy of PSK Offset Cluster Spots

[Richard Kriss]

The following is a copy of one of my recent postings to the PSK reflector
that was pretty well received and may be of interest to the Mac ham
community as well.

73 de Dick, AA5VU
aa5vu at arrl.net  
(can't get this Mac reflector to accept a new address)

------ Forwarded Message
From: "Dick Kriss, AA5VU" <aa5vu at arrl.net>
Date: Wed, 26 Jan 2005 13:58:56 -0600
To: psk31-post <psk31 at aintel.bi.ehu.es>
Subject: Accuracy of PSK Offset Cluster Spots

How may times have you seen a DX station spotted on the cluster and found
nobody there or just watched a QSO and see it spotted on the cluster at a
different frequency offset?   This kind of stuff bugs me so I used the 2004
ARRL Frequency Measurement Test to see if my station was calibrated to
reasonable amateur standards.

I participated in the ARRL FMT using a TS-570D(G) and W7AY's Macintosh OSX
cocoaModem sound card software to determine the offset. I don't own lab test
equipment.  This is the same technique I use to determine the offset for PSK
cluster spots. Just before the FMT I double checked the Kenwood TS-570D(G)
and cocoaModem against W1AW's RTTY bulletin and WWV.  The TS-570/cocoaModem
combo seemed to be right on the money - close enough for amateur use.

When FMT test tones came up they were very obvious on the cocoaModem
waterfall spectral display on both 80 and 40 meters. I had no copy on 20
meters as the band was dead at that time.  I let cocoaModem lock on and
track the tone. It settled very fast and cocoaModem offset box read 1101.1
Hz on 80 meters and 1105.6 Hz on 40 meters.

I submitted the data and forgot about it until I received a letter from Joe
Carcia, NJ1Q, the ARRL Station Manager dated 13-Jan-04. The letter thanked
me for participating, gave exact offsets used (80 = 1105.02 Hz and 40 =
1108.26 Hz) and showed my errors as 3.72 Hz on 80 meters and 2.66 Hz on 40
meters.

I understood the FMT was not a 'contest' but a test exercise to see how well
we could determine a frequency of a tone.  Needless to say I was
disappointed with the errors in my readings because I was dead on with W1AW
and WWV before the test.

On 25-Jan I called Joe, NJ1Q, at ARRL to discuss the data. First, he did not
use the normal W1AW transmitters for the test. He use the station's
commercial Harris transmitters and had them calibrated before the test to be
sure they were on the money. He used lab test equipment to determine the
stated offsets rather than PC sound cards. His comment was the participants
using lab test equipment came pretty close (some reporting the exact tone)
and the participants using sound cards technology were not as close. Joe
felt the errors like mine were normal for sound card technology due to the
lack of a frequency standard in sound card software programs. He will be
posting the data to the ARRL web page; however, the results of the 2004 FMT
will NOT be published in QST.

After reflecting on the conversation, I was became concerned that something
may be wrong with my rig, the cocoaModem software, the Mac G4's sound card A
to D converter or the operator (me).

On 24-Jan, I started tracking the new PropNET beacon that should be on
14.071 +1500 Hz and it seems to change. On my last check it was at 1580.9
Hz.  I am now really confused because the drift could be in the N4PSN
PropNET transmitter rather than in my station.  I have questioned the value
of the PropNET beacon (another discussion topic) but it may be useful as
test so I posted the following to the cluster to see where others are
copying the beacon.

14071.0  PROPNET  26-Jan-2005 1727Z  PSK N4PSN PropNET Beacon <AA5VU>
14071.0  PROPNET  26-Jan-2005 1724Z  PSK Copied at +1578.9        <AA5VU>
14071.0  PROPNET  26-Jan-2005 1723Z  PSK +1500 Where do u copy? <AA5VU>
14071.0  PROPNET  26-Jan-2005 1939Z  PSK Beacon now at 1580.9 Hz <AA5VU>

As a test see if you can copy the N4PSN PropNET beacon and post the offset
to the cluster.  A simple [sh/dx propnet] command should show the above plus
postings from others.  The beacon text says:

  For Information, please visit http://www.propnet.org
  N4PSN>HK:[EM84mo]PHG523ØC7/^5B8A

When I mentioned the ARRL FMT errors to Chen, W7AY, [developer of the
cocoaModem software] he pointed out this is why many amateurs buy expensive
TCXOs and calibrate them against WWV. Based on his comments the errors may
be due to my rig. I contacted Kenwood about accuracy and calibration of the
TS-570 and the response was I could return the rig to Kenwood and at my
expense they would calibrate it to Kenwood standards. God only know what
that means so I passed. If it ain't broke, don't fix it.

We all tend to believe our rigs with the XX,XXX.000 digital displays and
TCXOs are stable and accurate but something tells me they may all off,
including W1AW.  I conducted some RTTY experiments with W1AW a few months
ago with several other stations using better (or at least more expensive)
rigs and we call came to the same conclusion that W1AW was transmitting
slightly off. It was reported to the station manager and W1AW was
re-calibrated just before the FMT. Note my comment earlier about checking
against W1AW before the FMT.

I was hoping the FMT test would give me some assurance that I was quoting
the correct frequency offset for cluster spots.  I now question my ability
to spot PSK stations with a reasonable degree of accuracy and do not have a
solution. I am open to suggestions.  No comments please about ditching the
Mac OSX for Windows or to ditch the Kenwood for a Yaesu or Icom.

Thanks for reading this shaggy dog story posting.

73 Dick - AA5VU

------ End of Forwarded Message