++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
From: Larry East <w1hue at ida.net>
Subject: [Elecraft] L33, BFO Drift, and SSB Carrier Suppression (Long)

The following are observations, not complaints. If you feel an
uncontrollable urge to tell me that I'm full of crap, please do it off-line.
 
You may recall my post shortly after Field Day bemoaning my bad experience
with a T37-61 toroid used to replace L33 as per an application note on the
K2 web site. The purpose of the replacement was to increase SSB carrier
suppression. My K2's carrier feedthru was a little out of spec on 40m:
- -36dB rather than the specified -40dB or greater (relative to 10W). On all
other bands, the carrier was more than 40dB down. Replacing L33 with a 39uH
inductor wound on a T37-61 toroid provided an additional 6dB of carrier
suppression. Carrier suppression problem solved … but a new problem now
appeared: severe BFO drift! This became very evident during Field Day when
the cold night air caused the BFO settings to drift almost 400 Hz! Not good
- -- back to the original L33.

I measured the warm-up drift using the original L33 ferrite-shielded choke
and found it to be 22 +/- 2 Hz (down in frequency). I had to use an
external frequency counter to see the drift because the K2's internal
frequency counter drifts a similar amount in the opposite direction,
leading one to believe that the BFO drift is less than 10Hz. The warm-up
drift using the T37-61 inductor was 70-80 Hz.

After my post, Eric sent me a T44-8 toroid that is now being used for L33
in production K2's. I wound a new L33 using the #8 material core following
the instructions provided. I also changed C174 from the original 330pF to
82pF as per a note from Wayne. The thing worked, and did indeed result in
somewhat more SSB carrier suppression (but I failed to make a note of the
exact feedthru level…) However, the BFO range was now 7.6 kHz -- about 2X
more than really required -- and the BFO warm-up drift was a little more
than twice that of the original cylindrical choke (and in the opposite
direction). The course BFO frequency steps resulting from the wide BFO
range made it impossible to match the center frequencies of the CW filters.
I also noted that flexing the bottom cover caused a noticeable change in
BFO frequency -- not so with the original choke.

I decided once again to go back to the original L33 (I also had to go back
to the original 330pF for C174 in order to get sufficient BFO range). If
there is a trade-off to be made between SSB carrier suppression and BFO
stability, I'll go for BFO stability. Why? Because BFO drift will change
the transmit/receive offset as well as effect filter alignment. 10-30 Hz
drift is certainly acceptable, however. The VFO frequency will also change
slightly with temperature (my VFO warm-up drift on 40M is about 60 Hz), but
I'm not so concerned about that (as long as it is <100 Hz or so) because
VFO drift only affects frequency calibration. (One should make sure that
the PLL calibration is performed with the rig at normal operating
temperature.)

Just to see what effect a large temperature change would have, I put my K2
in the refrigerator for an hour or so to cool it down to about 45 deg-F.
Measuring the BFO frequency immediately after removing it from the 'fridge
showed that it was about 120Hz higher than the "nominal room temperature"
value-- guess that's acceptable. I suspect that the change would be much
greater (and probably unacceptable) using the T44-8 inductor, but I'm not
going to going to put it back in and try it. Maybe someone else will make
some precision temperature drift measurements (using an external counter)
on a rig using the T44-8 version of L33.

Well, 36 dB of carrier suppression really ain't so bad … but just maybe
something else can be "tweaked" to increase it. I reduced C174 slightly
from 330pF to 220 pF; that change resulted in some additional (but
insignificant) carrier suppression and a slight BFO range reduction -- 4.9
kHz with 220 pF vs. 5.0 kHz with 330pF. Before trying any smaller values
for C174, I decided to check the SSB carrier null pot-setting. Low and
behold, I got some additional suppression, and the nulls now occur at
essentially the same setting for LSB and USB. That certainly wasn't the
case before -- the null for USB was previously lower than the LSB null and
did not occur at the same pot setting.

So what's changed? The BFO frequency used for the "OP1" SSB filter in LSB
mode, for one thing. The SSB manual sez to use a BFO frequency of 4913.5
kHz for LSB and 4916.3 kHz for USB. However, in my rig at least, those
choices resulted in a significantly lower low-frequency filter cutoff in
LSB mode than in USB. In order to make the filters more symmetrical between
LSB and USB, I increased the LSB BFO frequency to 4913.64 kHz (my USB BFO
frequency is 4916.31 kHz). Oh yes -- I also installed the "SSB filter mod"
to improve SSB filter flatness written up in a recent application note.
Anyway, my 40m SSB carrier suppression in now 40.5 dB (for both LSB and
USB). Guess I'd better quit while I'm ahead of the game!

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Mark Erbaugh wrote:
> I don't think that there is any temperature compensation in the K2. Is 

Mark and others,

>From my experience with my own K2 I would say it's the BFO that's
drifting. The modified L33 on a T44-8 core may be better than the
original solenoid L33, but I can't really tell how much, because I
didn't make that many notes. Anyhow in my case doing L33 on a T44-8 core
didn't make the pass-band visibly more stable with changes in
temperature than what I'd seen before on the Spectrogram display.

A few days ago I started experimenting with white #7 material - it has
the lowest TC of them all at 30 ppm - and eventually I installed a T50-7
with 95 turns (0,2 mm) #32 wire on it as the modified L33 on the bottom
side of the PCB.

Spraying this inductor with cold spray doesn't make the BFO frequency
move unlike in the case of the T44-8 core, so this L33 certainly looks a
lot better, as expected. I haven't had enough 'miles' on my K2 with the
newest L33 installed to be able to say anything more definite, but IMO
it looks pretty good.

Please note I have only one #8 type core and only one K2, so my test
material is limited!

>From cold spray tests the two varactor diodes seem to be the single main
source of temperature dependence left in the BFO.

To some the prospect of winding 95 T of pretty thin wire on a half inch
toroid is probably abhorrent, but I can assure you it isn't all that
bad. The thinner the wire the more QRP?

To make the clearance between L33 and bottom cover as large as possible
I would like to try a T37-7 core, but the thinnest wire I have isn't
quite thin enough, hi.

Putting L33 on the top side of the PCB turned out less succesfull,
because in that case the BFO signal seems to get into the S-meter/AGC
circuit.
++++++++++++++++++++++++++++++++++++++++++++++++++++++
Date: Fri, 11 May 2001 18:03:31 -0700
From: Louis Hlousek <lhlousek at nvbell.net>
Subject: [Elecraft] K2 Bottom cover freq sensitivity

Elecraftisti,

One thing I've been aware of with my K2 is that if I pick it up or otherwise
press on the bottom cover while listening to a CW sig the frequency shifts
10 or 20 Hz due to the pressure on the bottom cover.  It had no bearing on
my use of the K2 but it bugged me anyway.

My K2 uses a toroidal inductor for L33 and it seems that L33 is what is most
sensitive to the proximity of the bottom cover. There is a standoff toward
the front and left of L33 (as viewed from the top front of the K2) that
spaces the RF board from the bottom cover.  There is another hole in the RF
board to the rear and right of L33 that looks like it once was used for
another standoff.  I drilled a matching hole in the bottom cover and
installed a second standoff.  It considerably reduced the amount of freq
shift caused by pressing on the bottom cover.

Luigi W7DZN
++++++++++++++++++++++++++++++++++++++++++++++++