Topics

uBITX 6 oscillators and filters


Mark Erbaugh
 

I did as Ashhar suggested in the hardware writeup and used pencil and paper to work out the mixing details and sideband inversion.

The 45 MHz IF is filtered with a  45 MHz bandpass filter. I'm assuming that this is on the order of +/- 15 kHz. I haven't found the bandpass specs for the 11.059 MHz crystal ladder filter but I assume it's around +/- 1.2 kHz for a 2.4 kHz bandpass. This is wider than some CW operators would like and the obvious solution is some sort of narrow filter at audio frequencies. However, as a thought experiment, would it be possible to offset the 45 MHz and 11.059 MHz IF signals against each other in such a way to significantly narrow the final bandwidth (kind of like my old Kenwood TS940 did?

Let me try and put some numbers to it:

Take a desired CW signal at 3.525 MHz and an interfering signals at 3.5245 and 3.5255. With the radio tuned to 3.525. MHz, the first mixer is at 48.525 MHz and the desired signal is at 45 MHz, the interfering ones at 44.9995 and 45.0005 MHz. The 45 MHz BP filter passes everything from 44.985 to 45.015 MHz, which includes all signals. If instead I lower the 1st mixer to 48.510 MHz, the desired signal would be at 45.985 (which would be passed by the 45 MHz BP filter), one interfering signal would be at 44.9845 MHz which would be out of the passband), the other would be at 45.9855 (which would be passed). We now have just the desired signal and only one interferer.

If the second IF filter is at 11.059 MHz +/- 1.2 kHz, that is 9.859 to 12.259 MHz. If I mix the first IF with a 33.726 2nd IF, I end up with the desired signal at 12.259 MHz (passed) and the interferer at 12.2595 (not passed). I then mix this IF with an 11.659 BFO to get a 600 Hz audio tone with no (or reduced) interferers.

I suspect that the filter skirts may not be sharp enough to eliminate the interfering signals as my discussion above assumed brick wall skirts, but would it possible to narrow the bandwidth using this scheme?

73,
Mark, N8ME


Evan Hand
 

Mark,
The short answer to your speculation is that it does work to some extent.  However, as you stated the skirts are not that sharp for the 45MHz crystal filter.  Dr. Lee, KD8CEC wrote software for the v3 through v5 uBITX that had the capability of adjusting both IF frequencies dynamically using a Nextion screen.  He called it Att and IF Shift.  Here is a link to the software:
http://www.hamskey.com/search?q=att

Check it out.
73
Evan
AC9TU 


Jerry Gaffke
 

Mark,

The 45mhz filter skirts are not sharp enough for this.
However the those 45mhz skirts have been successfully used for transmit ALC,
the local oscillator is moved around to a spot where transmit power to the antenna is as desired,
the skirts are gentle enough that the entire SSB passband goes through the 45mhz filter unmolested.

An audio filter for CW is the easiest solution, but the audio preamp at Q70 is the first to overload
from a nearby signal (by 30 or 40db as I recall) if the filter is placed in the usual spot around
the volume control.  So good enough for most of us, but not exactly a contest rig.

If you want a choice of narrow and wide crystal filters, one possible solution is described here:
    https://groups.io/g/BITX20/message/36549
This was originally suggested by Farhan, and has been successfully implemented.
The rest of that thread is worth a read.
I have not tried adding a second crystal filter.

Jerry, KE7ER


 c  On Tue, Dec 1, 2020 at 12:17 PM, Mark Erbaugh wrote:

I suspect that the filter skirts may not be sharp enough


Jerry Gaffke
 

Mark,

Here's where PA1FOX spoke of successfully implementing a CW crystal filter in parallel with the parallel SSB filter.
https://groups.io/g/BITX20/message/53927
Here's another thread about crystal filters of various sorts:
https://groups.io/g/BITX20/topic/7705048#36947

VK2ETA is the one that came up with the transmit ALC scheme using a skirt of the 45mhz filter,
this could also be used during receive:
https://groups.io/g/BITX20/message/51909

Jerry, KE7ER




On Tue, Dec 1, 2020 at 12:47 PM, Jerry Gaffke wrote:

The 45mhz filter skirts are not sharp enough for this.
However the those 45mhz skirts have been successfully used for transmit ALC,
the local oscillator is moved around to a spot where transmit power to the antenna is as desired,
the skirts are gentle enough that the entire SSB passband goes through the 45mhz filter unmolested.

An audio filter for CW is the easiest solution, but the audio preamp at Q70 is the first to overload
from a nearby signal (by 30 or 40db as I recall) if the filter is placed in the usual spot around
the volume control.  So good enough for most of us, but not exactly a contest rig.

If you want a choice of narrow and wide crystal filters, one possible solution is described here:
    https://groups.io/g/BITX20/message/36549
This was originally suggested by Farhan, and has been successfully implemented.
The rest of that thread is worth a read.
I have not tried adding a second crystal filter.

Jerry, KE7ER


Curt
 

Interesting discussion guys.

The shifting of oscillators relative to filter skirts reminds me of using the IF shift on my lowly ts140 to copy cw when I didn't have a narrow cw filter, it worked some. My SK friend left behind an sx110 with a single xtal and phasing control that behaves similar.

Remarkable that the Dutch builder found that the parallel 4 MHz narrow filter he crafted did not require a switch when paralleled with the original 12 MHz filter, their skirt rejection sufficient in itself like 2 dipoles for different bands on a common feed. 3 or 4 xtals, capacitors and some programming equals a cw filter. Maybe it can be included if there is a v7.

I am pretty content with the variable bandwidth audio filter I integrated in my v4, a nescaf.

73 curt