Thanks for the detailed reply Bill. You've spent as much time with this
circuit as I have! Only, my focus has been on changes to the BFO switching
circuit, rather than on BFO stability. I have presented the following
information in bits and pieces which always prompts additional questions.
Others have encouraged me to post my circuit changes for future reference
but I haven't had the time to start and end topic in one concise message.
Here goes...
Ten-Tec has two designs for the 9 MHz BFO Circuit Board: the older
generation uses the USB crystal for CW transmit, while the LSB crystal is
used for CW receive. The newer-generation board uses the LSB crystal for
both functions. Here's my theory for what it's worth:
I suspect the circuit change was made as a quiet attempt to improve the Omni
Six's CW note. Rather than switching between two crystals during the T/R
function, one crystal could be used, provided that logic changes were made
to components on the OMNI's Control Board. In Ten-Tec's attempt, the total
parts count, in fact, actually decreased. The new control board only works
with the new 9 MHz BFO board. Similarly, the old circuit boards must also
match together. Independently, I came to the exact same conclusion as
Ten-Tec and documented the changes. On paper, the new circuit appeared to
be a vast improvement. In reality, the new design is more prone to slight
chirp than the older design. Here's why: Ten-Tec did not change the value
of the LSB crystal. It remains 9,000.000 kHz. The Omni Six uses a fixed CW
transmit offset of 9,000.400 kHz. In CW transmit, the BFO crystal is
shifted *positively* to 9,000.400 kHz. These VXO circuits tend to produce
more audible chirp effects, albeit slight, when suddenly shifted positively
rather than negatively. When transmitting in CW, the poor LSB BFO crystal
is being rapidly frequency-shift-keyed and may or may not settle in time for
the QSK & T/R circuitry to quickly shift from receive to transmit.
To create a slight negative shift, I custom-ordered a 9,000.500 kHz crystal
and further changed the logic such that the crystal frequency is shifted *on
receive* rather than transmit. The initial "FSK" shock to the crystal is
only evident when switching from *transmit to receive.*
The older circuit board even with its inherent switching problems is still
superior in my opinion because although the USB crystal is shifted 2.7 kHz
on CW transmit, it represents a *decrease* in frequency as far as the VXO is
concerned.
That's the first problem. The second problem exacerbates the first. Here's
why:
The second problem is associated with the OMNI Six's 2.4 kHz @ 9 MHz filter.
As I've stated before and at the risk of sounding boring, the lower knee on
*some* Omni Six filters meets square-on with the Omni's CW fixed transmit
offset frequency of 9,000.400 kHz. A CW note that exhibits slight chirp
from the root-cause of problem #1 as stated above, becomes exacerbated from
the strangle-hold that the 2.4 kHz filter places on the BFO at 9,000.400
kHz. In theory, the 2.4 kHz filter is centered at 9,001.500 kHz. The
lower filter skirt, by design, cuts off at approximately 9,000.300 kHz. In
reality, the true lower filter skirt cutoff is not precise and its
variability can be attributed to imprecision in the filter's QC production
accuracy. The margin of filter error, even by design is only 100 Hz.
Imprecision during filter production can, and does, place the filter's knee
dangerously close to the CW transmit BFO frequency of 9,000.400 kHz.
THIS IS WHY SOME OMNI SIXES SOUND FINE ON CW WHILE OTHERS HAVE A SLIGHT
CHIRP AND DIRTY NOTE.
The Omni V does not manifest these problems even though the CW circuitry is
nearly 100% identical to that of the older-generation Omni Six. Here's why:
Although the Omni V and Six share the exact same 2.4kHz filter, the Omni V
uses a fixed CW BFO frequency of 9,000.600 Hz. This is 200 Hz higher than
the Omni Six's BFO frequency. When you're ever-so-close to that lower
filter cutoff frequency, 200 Hz makes all the difference in the world. I
confirmed this by slowly adjusting the BFO frequency while monitoring the CW
waveform on an Oscilloscope. On my Omni Six, a BFO frequency of less than
9,000.500 kHz resulted in a dirty CW note where the waveform begins to show
signs of visible distortion and a slight truncation of the waveform's
trailing edge. It's almost all in the filter accuracy, the remaining
difference rests with the BFO shift scheme. A BFO that chirps slightly,
will become much more intensified when it tries to pass through the
bottleneck of a sloppy filter skirt.
The absolute worst configuration is a newer Omni Six Plus, with an
inaccurate 2.4 kHz @ 9 MHz I.F. filter. (emphasis added to "inaccurate").
Listen to W1AW's 20-meter signal and you'll see and hear why. Perhaps we
can all take up a collection and present INRAD's 2.8 kHz filter to the W1AW
operating staff as a gift. I'll let Clark take up the cause from here!
-Paul, W9AC
-----Original Message-----
From: Bill Cook <wdcook@ibm.net>
To: Paul Christensen <paulc@mediaone.net>
Cc: tentec@contesting.com <tentec@contesting.com>
To: <tentec@contesting.com>
Date: Tuesday, December 29, 1998 5:46 PM
Subject: Re: [TenTec] Re: 2.4khz 9mhz pri INRAD Filters
>
>
>Paul Christensen wrote:
>
>> Bill,
>>
>> Can you elaborate on your BFO mods and what values/part types you used to
>> improve temperature stability? At one time I had considered replacing
the
>> majority of BFO trimmer caps with subminiature ceramic NP0 types.
>>
>> -Paul, W9AC
>>
>
> Paul,
> I replaced the BFO Osc, Buffer, and Agc xstrs with the metal can (4
pin)
>2N918 type. These are supposedly very slow heating. Collins Radio (who I
work
>for) always seemed to use these types in xtal oscillators in the HF and
slightly
>above area (I am referring to older designs). They have even published svc
>bulletins on various equipments changing 2N5179 to 2N918 because of noise.
The
>Osc base coupling cap was changed from a .01 discer to a .022 monolythic
cer.
>This was a SWAG because .022 is closer to 9mc for self resonance. The
>variables, except for FSK , were changed to ceramic piston type and the
fixed
>ones were changed to ceramic COG temp spec. I have silver micas in for
the two
>470pf, although I do not understand these high vaules in this colpitts ckt.
>Realistically they should be about half that value for 9mc. I am sure TT
has a
>reason. I do plan to play around with this some more as the SM caps are
unkowns
>as far as temp variations are concerned. I would like to change the
switching
>diodes, 1N4148 type, to PIN diodes but I am not sure about putting 20-30ma
>through them and screwing up the xtals.
>
>This is all fun for me, no matter how much the TT Omni cost.... hi hi
>
>73 Bill Cook N4WC
>
>
--
FAQ on WWW: http://www.contesting.com/tentecfaq.htm
Submissions: tentec@contesting.com
Administrative requests: tentec-REQUEST@contesting.com
Problems: owner-tentec@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|