Topics

BFO calibration


Rafael Diniz
 

Hi,
I'm sorry that you probably read 1000 times this question, anyway, when
adjusting the BFO, I see two "sweet spots" - when I can get signal in
the spectrum, one in a lower frequency, other in the higher one. Which
one is the correct? Ubitx v6.

Rafael


Bob Lunsford
 

I remember a tuneup video saying that the tuning required the tuning knob to be turned 'counter-clockwise' (to the left) but I am not sure if this is pertinent. If so, perhaps someone here can clarify this and/or let us know if it can be of help. It might be a simple way of proceeding from this point N'é?

Bob — KK5R

On Friday, January 15, 2021, 9:59:06 PM EST, Rafael Diniz <rafael@...> wrote:


Hi,
I'm sorry that you probably read 1000 times this question, anyway, when
adjusting the BFO, I see two "sweet spots" - when I can get signal in
the spectrum, one in a lower frequency, other in the higher one. Which
one is the correct? Ubitx v6.

Rafael







Evan Hand
 

On Fri, Jan 15, 2021 at 08:58 PM, Rafael Diniz wrote:
I'm sorry that you probably read 1000 times this question, anyway, when
adjusting the BFO, I see two "sweet spots" - when I can get signal in
the spectrum, one in a lower frequency, other in the higher one. Which
one is the correct? Ubitx v6.
The short answer is the BFO setting closest to 11.056MHz

73
Evan
AC9TU


Rafael Diniz
 

tks Evan, so 11.056.600 for me.
; )

On 1/16/21 1:33 AM, Evan Hand wrote:
On Fri, Jan 15, 2021 at 08:58 PM, Rafael Diniz wrote:

I'm sorry that you probably read 1000 times this question, anyway,
when
adjusting the BFO, I see two "sweet spots" - when I can get signal in
the spectrum, one in a lower frequency, other in the higher one. Which
one is the correct? Ubitx v6.

The short answer is the BFO setting closest to 11.056MHz

73
Evan
AC9TU


Mick
 

FYI,
incase anyone else finds this interesting
I have a v6 and currently have my BFO set at 11.055.605 

I produced a graph (attached) showing the relationship between BFO setting, input audio tone, and resulting watt output.
I used a tone generator from my computer through my digital input card (from elekitorparts) adjusted the audio input volume to eliminate possible saturation. Tune the radio to 40 m 7.163 MHz and tx’d into a dummy load.
Some observations:
1) for every 100 increase in BFO setting the bandwidth decreased  about 50Hz.
2) once in a favourable range it came down to how I heard my voice (from tinny to bassy) on another receiver.
3) the response graph was not a normal bell curve as I expected, it has significant peaks and valleys ( Does anybody know why?)
--
 

73
Mick VA3EPM 


Jerry Gaffke
 

Mick,

That is interesting.

What you are measuring is the frequency response of the 11.059mhz crystal filter,
I would not expect a bell curve, for a given BFO setting it should ramp up, have a flat top, then ramp down.
What you are seeing is some ripple in in that filter passband, which is not unusual.

Here is a webpage that shows filter response, I'd say your filter is similar to that of the second image:
    https://www.giangrandi.org/electronics/crystalfilters/xtaltuning.html
You have a linear vertical scale of power in Watts, filter response curves are usually shown
with a vertical scale in dB, so it is a plot of the logarithm of the power ratio.
If we figure your ripple to be varying between 10 Watts and 6 Watts, the trough is down from the peak by 10*log(6/10) = 2.2 dB.
That's a reasonably good figure for ripple on a crystal filter that has not been hand tuned.
Getting it closer to flat would involve adjusting the eight 68pf caps, and the impedance seen by the filter from T5 and T6.
Those adjustments would be somewhat different for each set of eight matched crystals that we might build a filter with.

Bottom line:
If you can find a BFO setting for which you get good sound quality when speaking into the mike, that's plenty good enough.

Jerry, KE7ER




On Sat, Jan 16, 2021 at 06:18 AM, Mick wrote:

FYI,
incase anyone else finds this interesting
I have a v6 and currently have my BFO set at 11.055.605 

I produced a graph (attached) showing the relationship between BFO setting, input audio tone, and resulting watt output.
I used a tone generator from my computer through my digital input card (from elekitorparts) adjusted the audio input volume to eliminate possible saturation. Tune the radio to 40 m 7.163 MHz and tx’d into a dummy load.
Some observations:
1) for every 100 increase in BFO setting the bandwidth decreased  about 50Hz.
2) once in a favourable range it came down to how I heard my voice (from tinny to bassy) on another receiver.
3) the response graph was not a normal bell curve as I expected, it has significant peaks and valleys ( Does anybody know why?)


Jerry Gaffke
 

Correction:  There are five 68pf caps (and eight crystals)

Attempts to adjust those caps and the impedance seen from T5 and T6 for minimum ripple will also
affect filter bandwidth (we are shooting for around 2khz), and how much the filter attenuates the signal.

Jerry, KE7ER


On Sat, Jan 16, 2021 at 07:45 AM, Jerry Gaffke wrote:
Getting it closer to flat would involve adjusting the eight 68pf caps, and the impedance seen by the filter from T5 and T6.


Jerry Gaffke
 

Let's assume the capacitors are truly all of equal value, the crystals are identical, 
and we are happy with the bandwidth of the filter.
Minimizing ripple is then mostly a matter of adjusting the impedance seen by the filter.

The transformers at T5 and T6 have a 2:1 winding ratio, so a 4 to 1 impedance ratio.
The BiDi amps have input/output impedances of 50 ohms
and the 220,22,220 ohm pad at R114,115,116 establishes a 50 ohm termination at the modulator.
So the filter should see a nominal 50*4 = 200 ohms at both ends.

If you want to play with adjusting those impedances, you would first want to set up a way to continuously plot
the filter passband so you can see what your tweeks are doing.
You have a 50% chance that the 200 ohms presented to the cyrstal filter must be increased, not decreased.
This would be easy, add series resistance somehow at both the input and output (though this does slightly increase filter attenuation).

Reducing impedance at the demodulator might be accomplished by messing with the pad at R114,115,116 (just reduce R114?).
The other end is slightly more difficult: on receive by reducing the 47 ohms at R37, on transmit try reducing the 10 ohms at R43.
Increasing these resistances is one way to effectively "add series resistance somehow".

Could be fun.
But if the rig sounds good, that's probably good enough for most of us.
 
Jerry, KE7ER


On Sat, Jan 16, 2021 at 07:54 AM, Jerry Gaffke wrote:

Attempts to adjust those caps and the impedance seen from T5 and T6 for minimum ripple will also
affect filter bandwidth (we are shooting for around 2khz), and how much the filter attenuates the signal.


Mick
 

Jerry,
Thanks for the explanation and article, very interesting!
--
 

73
Mick VA3EPM 


Rafael Diniz
 

Thanks a lot Jerry, really enlightening your emails. Did you already put
all this information in a website or somewhere?

Flat response and a wider bandwidth is a dream, together with the
GPSDO-disciplined Si5351. Then I'll have my perfect ubitx v6 for digital
multi-carrier modulations.
; )

uBitX rocks!

73's,
Rafael PU2UIT

On 1/16/21 1:32 PM, Jerry Gaffke via groups.io wrote:

Let's assume the capacitors are truly all of equal value, the crystals
are identical, 
and we are happy with the bandwidth of the filter.
Minimizing ripple is then mostly a matter of adjusting the impedance
seen by the filter.

The transformers at T5 and T6 have a 2:1 winding ratio, so a 4 to 1
impedance ratio.
The BiDi amps have input/output impedances of 50 ohms
and the 220,22,220 ohm pad at R114,115,116 establishes a 50 ohm
termination at the modulator.
So the filter should see a nominal 50*4 = 200 ohms at both ends.

If you want to play with adjusting those impedances, you would first
want to set up a way to continuously plot
the filter passband so you can see what your tweeks are doing.
You have a 50% chance that the 200 ohms presented to the cyrstal
filter must be increased, not decreased.
This would be easy, add series resistance somehow at both the input
and output (though this does slightly increase filter attenuation).

Reducing impedance at the demodulator might be accomplished by messing
with the pad at R114,115,116 (just reduce R114?).
The other end is slightly more difficult: on receive by reducing the
47 ohms at R37, on transmit try reducing the 10 ohms at R43.
Increasing these resistances is one way to effectively "add series
resistance somehow".

Could be fun.
But if the rig sounds good, that's probably good enough for most of us.
 
Jerry, KE7ER


On Sat, Jan 16, 2021 at 07:54 AM, Jerry Gaffke wrote:

Attempts to adjust those caps and the impedance seen from T5 and
T6 for minimum ripple will also
affect filter bandwidth (we are shooting for around 2khz), and how
much the filter attenuates the signal.


Jerry Gaffke
 

As the article points out, reducing the impedance seen by the can be had
by adding a resistor to ground at each end of the filter.
Far easier than all the fiddling around that I had suggested.

Jerry, KE7ER


On Sat, Jan 16, 2021 at 08:33 AM, Mick wrote:
Jerry,
Thanks for the explanation and article, very interesting!


_Dave_ AD0B
 

When I first got a V3 in 2018, I noticed the variation in power output (exactly as Mick showed in his graph) using ft8. If wanting the most power, the frequency had to be chosen carefully.

Didn't realize why, but was seeing ripple in the filter.  So learned something today.
Thank you Jerry.

For various reasons, mainly boredom, I don't use ft8 anymore 

Generally get good audio reports. Although apparently not all frequencies are treated the same. 
Makes me wonder if something like this could be made to ignore some audio frequencies and prioritize others to give higher power by choosing only narrow bands of the audio spectrum. Similar to the way phone companies used to multiplex telephone calls.  
--
73
Dave
ADOB
Raduino bracket and Ham_Made_Keys