开云体育

A low pass filter element that has gone awry?? I think the 20 and 17 use the same filter?

Gordon


On Mon, Aug 21, 2023 at 8:20?AM John Terrell, N6LN <N6LN@...> wrote:

My V2 is working well but I can only get 3 watts out on 17 meters.
All other bands have higher power. I've done the latest ./update
of the software and also the txcal command. I have adjusted the
scale coefficients on all bands to have no more than 25 watts out
on CW to be conservative. Despite increasing the scale coefficient
for 17 meters it tops out at 3 watts. This is into a dummy load. I
have even tried a new sd card image and there's no difference.
Here are the power outputs (Drive = 100, with hw_settings.ini
scales adjusted, and on 10, 12 and 17 M maxed out):
10 M???? 10 Watts
12 M???? 13 W
15 M???? 25 W
17 M?????? 3 W
20 M????? 24 W
30 M????? 25 W
40 M????? 24 W
80 M??? ? 22 W

I would think it might be a PA transistor or driver problem but
why would it be on one band only? And I used to get much more
power out on 17 meters. Again it doesn't seem to be a software
problem because I even swapped out sd cards to run a different
image from a file saved that worked before. Could it have
something to do with the LPF circuits? But 17M doesn't have its
own dedicated LPF line and all other bands work.

Suggestions?
Jack, N6LN

On Aug 20, 2023, at 05:42, Rafael Diniz <rafael@...> wrote:

?Did you tested for spurs down in the frequency bands?

I can see I can easily put a good amount of power at least down to 1MHz. Different than uBitx, I could pipe power from 1 to 30 MHz without changing the LPF settings.

- Rafael

On 8/20/23 07:35, Gordon Gibby wrote:
Very good. I did not narrow mine down to make quite as accurately a measurement as you did.

Mine appears to be the 11th harmonic of the 80 m signal, much worse at the lower end of that band, subtracting from the (40 MHz plus operating frequency) local oscillator signal. On 80meters that ends up being within the 3 to 5 MHz range.

On my unit, it wasn’t so bad towards 4 MHz ; but it got much worse towards 3.5 MHz where the spur actually crossed the operating frequency around 3.64 megahertz. If I can reduce this spur, and then do my little filter on the 20 m I think I’m done with fixing emissions! With the change to 2 mA on the S I 5351 and other changes it’s gotten a lot lot better. I’m wondering if I miss set the idling current but haven’t had a chance to get back to it.

Where did the spurs from the second harmonic of the BFO show up?

73
Gordon Kx4z

On Aug 20, 2023, at 02:11, Ashhar Farhan <farhanbox@...> wrote:

?
Gordon,
Here is a trick that I use.
First, I switch to a resolution of 300 hz or less on the spectrum analyzer, keep the span to 10 khz.
Next, center the spur and tune the LO away by 1 khz, find out how much the spur has moved and in which direction. That will telp you which harmonic of LO it is and if it is subtracting or adding to another signal.
We already know the spur freq and with rhe shift we also know which harmonic of the LO it is. We can easily back calculate the other signal.
For me, the other signal was the second harmonic of the bfo (40.035 x 2 = 80.070)
- f

On Sun, Aug 20, 2023, 10:56 AM Gordon Gibby <docvacuumtubes@...> wrote:

I've been terribly busy for the last couple week, starting up AP
classes in precalculus and Chemistry at a local private high
school, and this weekend, we are running a big ham radio license
class.... but I think I can now explain the mathmatics and thus
the interactions that are causing the spur that I've been
tackling on 80 meters.

Here are the observed data from before and now some math that
explains it:

Operating Spur 11th harmonic Local
Oscillator Implied Bandpass
Frequency Frequency of operating that goes
with Filter (should be constant)
(MEAS)URED) (MEASURED)frequency making that spur
3.500 5.008 MHz 38.5 MHz 43.508 MHz
40.008 MHz
3.510 4.912 MHz 38.61 43.522 MHz
40.012
3.520 4.816 MHz
3.600 4.012 39.6 MHz 43.612 MHz
40.012 MHz
3.620 3.808 MHz 39.82 43.628 MHz
40.008 MHz
3.640 3.640 MHz

When you measure a frequency with a spectrum analyzer, it isn't
exactly like a ferquency counter. The spectrum analyzer creates
"bins" of frequencies and you just find which "bin" had the
highest amount, so the accuracy of your measurement has to do
with your chosen BANDWIDTH....and using a very narrow bandwidth
has the advantage of increased precision, but slows things down
greatly.

The data above are nicely in sync with the 11th harmonic of the
TRANSMITTER OUTPUT (an odd harmonic, so favored by the push-pull
architecture over the even harmonics) which is present is good
amount before the low pass filter.....subtracting (mixing) from
the (tuned) LOCAL OSCILLATOR SIGNAL coming from the Si5351, and
producing a spur that is within the 80 meter band, and exactly of
the right frequency to match the observed value, given the value
of the IF filter (which is corrrectly calculated for each spur in
good agreement).

So I don't yet know HOW the transmitter output is being coupled
into some mixer -- likely the real mixer, since the signal shows
up at the very next 2N2222 transistor -- but some possibilities
include:

a) passed along through the +12V line despite capacitive bypassing
b) possibly magnetically introduced into the transformers of the
mixer
c) capacitive coupling into a pcboard trace from the Si5351
d) something else I haven't guessed yet.

So possible solutions for this?
a) check my biasing of the power amplifier which may have
inadequate idling current. Increasing the idling current may
reduce the 11th harmonic and decrease the spur
b) Add more capacitive bypassing or RC decoupling to the mixer
and/or Si5351 stage
c) Add some sot of series "trap" tuned to roughly 38- 41 MHz to
the output amplifier to try and reduce this specific harmonic?
That might be difficult because it will move across lots of
frequencies
d) Work on physical shieldng and mu-metal shielding of the mixer?

Of those, (a) and (b) are the easiest to start with. When I get
some more TIME I'm going to work on those.

I saw somewhere that some other transceiver was claiming < 50dBc
spurious responses. It would be great to have sBitx's doing
that routinely also. Mine may well be an outlier with more
problems than most, but that gives manufacturing tolerance
information to make the entire system better.

This is progress.
73
Gordon KX4Z

Very good. ?I did not narrow mine down to make quite as accurately a measurement as you did.

Mine appears to be the 11th harmonic of the 80 m signal, much worse at the lower end of that band, subtracting from the (40 MHz plus operating frequency) local oscillator signal. ? On 80meters that ends up being within the 3 to 5 MHz range.

On my unit, it wasn’t so bad towards 4 MHz ; but it got much worse towards 3.5 MHz where the spur ?actually crossed ?the operating frequency around 3.64 megahertz. ? If I can reduce this spur, and then do my little filter on the 20 m I think I’m done with fixing emissions! ?With the change to 2 mA on the S I 5351 and other changes it’s gotten a lot lot better. ? I’m wondering if I miss set the idling current but haven’t had a chance to get back to it.

Where did the spurs from the second harmonic of the BFO show up?

73
Gordon Kx4z

On Aug 20, 2023, at 02:11, Ashhar Farhan <farhanbox@...> wrote:

?
Gordon,
Here is a trick that I use.
First, I switch to a resolution of 300 hz or less on the spectrum analyzer, keep the span to 10 khz.
Next, center the spur and tune the LO away by 1 khz, find out how much the spur has moved and in which direction. That will telp you which harmonic of LO it is and if it is subtracting or adding to another signal.
We already know the spur freq and with rhe shift we also know which harmonic of the LO it is. We can easily back calculate the other signal.
For me, the other signal was the second harmonic of the bfo (40.035 x 2 = 80.070)
- f

On Sun, Aug 20, 2023, 10:56 AM Gordon Gibby <docvacuumtubes@...> wrote:

I've been terribly busy for the last couple week, starting up AP
classes in precalculus and Chemistry at a local private high
school, and this weekend, we are running a big ham radio license
class.... but I think I can now explain the mathmatics and thus
the interactions that are causing the spur that I've been
tackling on 80 meters.

Here are the observed data from before and now some math that
explains it:

Operating? ? ? ? Spur? ? ? ? ? ? ? ? 11th harmonic Local
Oscillator? ? ? ?Implied Bandpass
Frequency? ? ? ?Frequency? ? ? ?of operating? ? ? ?that goes
with? ? ? ? ?Filter (should be constant)
(MEAS)URED) (MEASURED)frequency? ? ? ? ? making that spur
3.500? ? ? ? ? ? ? ?5.008 MHz? ? ? ? ? 38.5 MHz ?43.508 MHz? ? ?
? ? ? 40.008 MHz
3.510? ? ? ? ? ? ? ?4.912 MHz? ? ? ? ? 38.61 ?43.522 MHz? ? ? ? ?
? ?40.012
3.520? ? ? ? ? ? ? ?4.816 MHz
3.600? ? ? ? ? ? ? ?4.012? ? ? ? ? ? ? ? ? 39.6 MHz 43.612 MHz? ?
? ? ? ? ? 40.012 MHz
3.620? ? ? ? ? ? ? ?3.808 MHz? ? ? ? ? 39.82 43.628 MHz? ? ? ? ?
? ? 40.008 MHz
3.640? ? ? ? ? ? ? ?3.640 MHz

When you measure a frequency with a spectrum analyzer, it isn't
exactly like a ferquency counter.? ?The spectrum analyzer creates
"bins" of frequencies and you just find which "bin" had the
highest amount, so the accuracy of your measurement has to do
with your chosen BANDWIDTH....and using a very narrow bandwidth
has the advantage of increased precision, but slows things down
greatly.

The data above are nicely in sync with the 11th harmonic of the
TRANSMITTER OUTPUT (an odd harmonic, so favored by the push-pull
architecture over the even harmonics) which is present is good
amount before the low pass filter.....subtracting (mixing) from
the (tuned) LOCAL OSCILLATOR SIGNAL coming from the Si5351, and
producing a spur that is within the 80 meter band, and exactly of
the right frequency to match the observed value, given the value
of the IF filter (which is corrrectly calculated for each spur in
good agreement).

So I don't yet know HOW the transmitter output is being coupled
into some mixer -- likely the real mixer, since the signal shows
up at the very next 2N2222 transistor -- but some possibilities
include:

a)? passed along through the +12V line despite capacitive bypassing
b)? possibly magnetically introduced into the transformers of the
mixer
c)? capacitive coupling into a pcboard trace from the Si5351
d)? something else I haven't guessed yet.

So possible solutions for this?
a)? check my biasing of the power amplifier which may have
inadequate idling current.? ?Increasing the idling current may
reduce the 11th harmonic and decrease the spur
b)? Add more capacitive bypassing or RC decoupling to the mixer
and/or Si5351 stage
c)? Add some sot of series "trap" tuned to roughly 38- 41 MHz to
the output amplifier to try and reduce this specific harmonic??
?That might be difficult because it will move across lots of
frequencies
d)? Work on physical shieldng and mu-metal shielding of the mixer?

Of those, (a) and (b) are the easiest to start with. ?When I get
some more TIME I'm going to work on those.

I saw somewhere that some other transceiver was claiming < 50dBc
spurious responses.? ? It would be great to have sBitx's doing
that routinely also.? ?Mine may well be an outlier with more
problems than most, but that gives manufacturing tolerance
information to make the entire system better.

This is progress.
73
Gordon KX4Z