[Elecraft] Re: N6KR's 4 MHz oscillator cal method for the K2
Steven Gibbs
[email protected]
Thu Aug 21 15:37:00 2003
Wayne Burdick wrote:
> Not if you use my new method, and that's the entire point.
> No iteration needed.
But Wayne, step 2 of your method says
> Now alternately move the K2's internal counter probe between
> TP1 (VCO) and TP2 (BFO), adjusting C22 in small increments
> until the kHz and Hz digits at the two test points match as
> closely as possible."
That *IS* an iterative method, even though CAL-PLL & CAL-FIL
are not included in the loop. It suffers from CAL-FCTR's
10Hz display ambiguity which cannot afford the precision
needed to ensure MCU oscillator setting within the one part
in a million which is needed.
> Your objection regarding accuracy of tuning of the reference
> signal can be addressed in other ways, as noted by other
> posters.
You will need to define these in a manner understandable by
average K2 builders who have been misled by so many scatter-
brained schemes posted here.
> But from the standpoint of adjusting C22, this is the only
> non-iterative adjustment method.
*Not so*. My comment gave two *real* non-iterative methods,
repeated below. Please take time to verify that they are
straightforward and that they really do *consistently* work
to the needed accuracy.
> +/- 30 Hz at 10 meters is easily achieveable.
Only if MCU reference is within one part in a million of 4MHz.
For the reasons I outlined in my previous comment, average K2
builders using your method cannot be assured they would achieve
this.
This comment is intended to be helpful to the Elecraft community,
and must in no way be taken as any intention to 'flame'.
Regards & 73, Steve GU3MBS.
======
Elecraft K2 Control Board 4MHz Oscillator Adjustment
k2-4mhz-set.txt 07 August 2003
�2003 Steven Gibbs GU3MBS mailto:[email protected]
Here are two methods of correctly setting the K2 4MHz clock
frequency with greater confidence than the methods suggested
in the K2 Owner's Manual, Chapter 6, 'Alignment and Test,
Part II' and Chapter 9, 'Frequency Calibration Techniques'.
Direct comparison of the MCU clock and a standard frequency
is made, so these methods avoid pitfalls which commonly
trouble K2 builders:
* Neither method is iterative - you don't have to
refine your adjustment by chasing around a loop;
* Neither method calls for specific frequency read-
outs on the K2 display - the 10Hz dial uncertainty
is not a factor.
Method A requires that assembly of the K2 is complete. It
requires reception of a standard frequency transmission, and
needs a PC running the Spectrogram program.
Latest Spectrogram version is available at
http://www.visualizationsoftware.com/gram.html
Freeware Spectrogram version 5.1.7 is available at
http://www.guernsey.net/~sgibbs/gram50.zip
Method B can be performed at the Chapter 6, 'Alignment and
Test, Part II' stage, or when assembly of the K2 is complete.
It needs an oscilloscope and a 1MHz frequency standard. Lab-
standard signal generators & counters commonly have a 1MHz
square-wave reference output. I used an off-the-air standard
derived from the high-accuracy carrier frequency of the BBC's
198kHz LW transmission.
________
METHOD A
A1. Set the K2 to USB. Identify and tune (approximately
zero-beat the carrier) a standard frequency (WWV) transmission
on 10MHz, 15MHZ or 20MHz. (To improve reception on 15MHz you
could temporarily peak C21 & C23; to improve reception on
20MHz you could temporarily peak L10 & L11). Tune the K2 dial
about 1kHz lower. On Spectrogram you should see the WWV
carrier as a tone of about 1000Hz. Disregard any modulation
tones that WWV may also be transmitting.
A2. Disconnect the antenna; you should see a weak signal
somewhere on Spectrogram - this is the 10th (or 15th, or 20th)
harmonic of a 1MHz signal derived by the MCU from its 4MHz
clock (you may recall that a similarly-derived 7MHz signal is
used in initial 40M alignment). You can verify that you are
looking at the right signal by touching the Control Board MCU
crystal X2 or its associated C22 trimmer and noting that the
frequency on the Spectrogram display changes slightly. If you
don't see the MCU signal you can tune the K2 dial up or down a
couple of hundred Hz until you locate it.
A3. If WWV is coming through strongly there may be enough
breakthrough to show WWV's signal also. Use an insulated
tuning tool (sorry, but the Elecraft tool is rather poor for
this job) to adjust Control Board C22 so that the two signals
coincide - final proximity may show a beat which can be
reduced to zero. If WWV isn't very strong you won't see its
breakthrough signal so you'll have to alternately connect and
disconnect the antenna until you have the two signals showing
the same frequency on the Spectrogram display.
NOTE: the exact setting of the K2 dial and the
exact frequency on the Spectrogram display doesn't
matter. The sidetone frequency which you have set
doesn't matter either. These are the important
features of this method - the MCU oscillator is
being compared DIRECTLY with the standard
frequency transmission. If you have a KAT2 or a
KAT100 you can switch to an unconnected ANT2
instead of disconnecting your antenna.
A4. Perform CAL-PLL.
A5. Perform CAL-FIL.
CAUTION: remember that the K2 manual says
'Whenever you switch modes or filters, the K2 will
first record your new settings, IF THEY HAVE BEEN
CHANGED' [my caps]. This means that to properly
calibrate the BFO to the new MCU crystal frequency,
the BFO control parameter must be changed, and then
changed back to its original setting.
A6. SPOT the WWV carrier in CW-nor and again in CW-rev; in
each case the K2 dial should show 10000.00kHz (or 15000.00kHz
or 20000.00kHz) probably plus or minus about 0.03kHz (30Hz).
NOTE: significantly greater error means that
either your MCU 4MHz isn't properly set to the
WWV (re-check), or that you forgot CAL-PLL, or
that CAL-FIL wasn't properly performed (re-do).
If CW-nor & CW-rev dial accuracy is OK, then
LSB & USB should look after themselves,
provided CAL-FIL has been correctly done.
________
METHOD B
If you have access to a 1MHz frequency standard, here's a way
of setting the K2's 4MHz Control Board oscillator with greater
certainty than trying to beat with WWV.
You will need a 4MHz bandwidth oscilloscope having a high-
sensitivity y-amp (5mV/div worked for me), with an independent
trigger input derived from your frequency standard.
B1. Temporarily reduce the y-amp sensitivity, connect the
probe to the trigger source, and adjust the 'scope's timebase
and trigger level to stably display one or two cycles of the
1MHz standard frequency.
B2. Connect the probe's ground connection to the K2 Control
Board ground point. Move the 'scope probe to the vicinity of
X2-C21-C22 on the Control Board, but do not touch anything,
just rely on loose capacitive coupling. Switch the 'scope to
maximum y-sensitivity, and look for a trace showing pick-up
from the 4MHz clock oscillator.
B3. Use an insulated trimming tool (sorry, but the Elecraft
tool is rather poor here) to carefully adjust C22 so that the
scope trace is resolved into a sine wave moving slowly or
rapidly either to the left or to the right. Note that four
cycles occupy the same time as one cycle of the 1MHz reference
oscillator. Careful tuning will result in a near-stationary
display when the trimming tool is removed, and the 'scope
probe is gradually drawn away. This indicates very close
matching of the K2's 4MHz oscillator to four times the
standard frequency. (A left or right shift of 4 cycles in one
second means the frequencies are matched to one part in a
million. Temperature effects will degrade frequency accuracy,
so it's probably not worthwhile trying for better than this).
If you are still building your K2, continue with 'Alignment
and Test, Part II, PLL Reference Oscillator Range Test'.
Otherwise, continue with step B4.
B4. Perform CAL-PLL.
B5. Perform CAL-FIL.
CAUTION: remember that the K2 manual says
'Whenever you switch modes or filters, the K2 will
first record your new settings, IF THEY HAVE BEEN
CHANGED' [my caps]. This means that to properly
calibrate the BFO to the new MCU crystal frequency,
the BFO control parameter must be changed, and then
changed back to its original setting.
B6. Loosely couple the frequency standard output to the K2's
antenna. You will hear 1MHz harmonics at 2, 4, 7, 10, 14, 18,
21, 25 & 28MHz. SPOT these in CW-nor and again in CW-rev; in
each case the K2 dial should show probably plus or minus
about 0.03kHz (30Hz).
NOTE: significantly greater error means that
either your MCU 4MHz isn't properly set to the
frequency standard (re-check), or that you
forgot CAL-PLL, or that CAL-FIL wasn't properly
performed (re-do). If CW-nor & CW-rev display
accuracy is OK, then LSB & USB should look after
themselves, provided CAL-FIL has been correctly
done.