Great series. I have the receive part all built, just trying to fault find it - the most fun part lol. My xtal filter looks a bit wider than I wanted but be ok to watching FT8 traffic as a test - there's always something transmitting! I suspect I haven't got the alignment setup up between VFO and BFO and xtal filter so need to get pencil to paper out and work out offsets for LSB and USB. 40M band predominantly LSB SSB but darn FT8 is always USB. Keep up the great vids - they are proving invaluble!
Put in a LSB/USB switch and have the software place the BFO either above or below the crystal filter passband. Same offset, i.e., start with 300Hz away from the -3dB point. I started with centre freq + (2700Hz -3dB passband/2) + 300Hz. That should get you in the ballpark. Charlie
good morning Charlie, Are you 100 % sure its bleeding through the mixers. I suspect due to the output level of the si5351 that it may be picking up directly from it on your radio. The outputs are reasonably high on the si5351 boards, This is done to maximise it's flexibility and usefulness. you did the plots on the filters so you basically know they are approximately right , the sbl1 mixer gives you basically the same results. The difference between your mixer and the sbl1 is this... The sbl1 has a screening around it and your mixer doesn't. The SBL1 is very similar to your mixer and is allowing your oscillator to bleed through ( just the best way to describe it), what if its not bleed through and more your radio's dummy load picking up the oscillator directly ( this may give the impression its going through the mixer). The output of the SI5351 is reasonably high at 3V and would transmit across that 18 inches or so on your desk. There are things that can be done to reduce it like attenuator pads ETC. you can try screening the oscillator with something metal to see if the level drops (if its bleed through it wont make any difference, if its transmitting over the short distance it will reduce in level). I hope this is of some help Best regards and 73 Simon (M6FJW) P.S. No I don't have time to do the intermediate and the full licence right now, Maybe sometime in the future.
Thanks Simon and Stephen. I have a few things to try here. If I scope the RF out of the balanced modulator then I do see some bleed through. Not a lot, but some. I was thinking about trying a single balanced mixer which has both the balance pot and a trim cap. I prefer the DBM, but could be an option as an experiment. I think for a start I'll remote mount the Si5351 in a shielded box and see how that works. I know it has to be done, but I love seeing the electronics out in the open! I'm also reluctant to reduce the output level of the Si5351 as the levels are spot on for driving the mixer (if I assume the levels are the same as a SBL-1). That being said, I have noting top loose and the ability to do so is baked into the library. I do know the commercial radio will pick up the Si5351 output directly, but it's at a much lower level. What I didn't film was transmitting a podcast. It's sounds great. As for the filter response, yes i did plot that and it didn't look too bad. Clearly not a good as a commercial 8 pole filter, but ok. It's another area I'd like to look at in due course. Again thanks to you both for the input. Charlie
Hi Charlie , A quick question, what was the width of the filter at its base and the width at the -3dB point. The difference between the two cannot exceed 300 Hz (or you get the ability to transfer the carrier through). I am sure you used the center frequency of the filter + half the filter width + 300 Hz as your BFO frequency. If the -3dB point and the base of the filter are greater than 300 Hz then yes it could bleed through, but this shouldn't be by a great deal due to the attenuation of the skirts on the filter. One other note you could try extra capacitors on the power supply right were the power enters each board just in case the power rails are the problem and carrying the RF through to multiple sections. keep up the great video's Best regards Simon ( M6FJW)
Cheers Simon. I'll look at it again after work. I have a few ideas to split things out as i note the Si5351 is very close to both the balanced mod and the IF amp. 73s Charlie, ZL2CTM
Simon. I have done some comparison experiments between the homebrew crystal filter and a Yaesu XF-92A filter. The commercial filter has significantly steeper skirts and greater stopband attenuation, which is what I thought. I'm tempted to rebuild the filter using-say-six crystals. That may not be the best approach as i suspect the crystals themselves are not high quality given the price I paid (I'm not complaining as it was always going to be an experiment). The other option that i may look at is use the XF-92A and continue. All good fun! As an aside, there is a small amount of bleed though, but it's not too bad. Moving the BFO up 300Hz also helped (as expected). The filter is the low hanging fruit.
Gidday Simon. I had a good play with varying the Si5351 output levels. It did make a difference, but at the expense of the RX and TX audio quality. 8mA seems to be the best.Thanks again. Charlie
I think you would get more from improving the diode balance than by improving the bandpass filter. Just the same, it is just not possible to completely cancel your carrier, and you're probably good with what you've got.
Regarding your use of the geometric mean of the typical and minimum specs, I don't think this is sufficient. If your biasing depends on the actual beta, then you need to check your operating point (the quiescent Vc in this case) using both the minimum and maximum possible beta. This is a bit problematic, since max beta is almost never specified, but I use 3 x the typical value, since on the rare datasheets that DO give a max number, that's usually about what it is. Since you use voltage divider bias, beta will have very little effect on the operating point, though, and will actually depend more on the junction temperature. You already fudge this by using Vbe = 0.7, which is reasonable and has always worked for me.
If I built a circuit with a gain of about 5, and then ran its output through a 1/1000 voltage divider, I would probably be asking myself if I needed an amplifier.
Appreciate you showing your calculations and explanations of what you are doing
Thanks Kenneth. I hope you find the videos useful. At least for showing what not to do!Charlie
Great series. I have the receive part all built, just trying to fault find it - the most fun part lol. My xtal filter looks a bit wider than I wanted but be ok to watching FT8 traffic as a test - there's always something transmitting!
I suspect I haven't got the alignment setup up between VFO and BFO and xtal filter so need to get pencil to paper out and work out offsets for LSB and USB. 40M band predominantly LSB SSB but darn FT8 is always USB.
Keep up the great vids - they are proving invaluble!
Put in a LSB/USB switch and have the software place the BFO either above or below the crystal filter passband. Same offset, i.e., start with 300Hz away from the -3dB point. I started with centre freq + (2700Hz -3dB passband/2) + 300Hz. That should get you in the ballpark.
Charlie
good morning Charlie,
Are you 100 % sure its bleeding through the mixers. I suspect due to the output level of the si5351 that it may be picking up directly from it on your radio. The outputs are reasonably high on the si5351 boards, This is done to maximise it's flexibility and usefulness.
you did the plots on the filters so you basically know they are approximately right , the sbl1 mixer gives you basically the same results. The difference between your mixer and the sbl1 is this... The sbl1 has a screening around it and your mixer doesn't. The SBL1 is very similar to your mixer and is allowing your oscillator to bleed through ( just the best way to describe it), what if its not bleed through and more your radio's dummy load picking up the oscillator directly ( this may give the impression its going through the mixer). The output of the SI5351 is reasonably high at 3V and would transmit across that 18 inches or so on your desk.
There are things that can be done to reduce it like attenuator pads ETC. you can try screening the oscillator with something metal to see if the level drops (if its bleed through it wont make any difference, if its transmitting over the short distance it will reduce in level).
I hope this is of some help
Best regards and 73
Simon (M6FJW)
P.S. No I don't have time to do the intermediate and the full licence right now, Maybe sometime in the future.
Thanks Simon and Stephen. I have a few things to try here. If I scope the RF out of the balanced modulator then I do see some bleed through. Not a lot, but some. I was thinking about trying a single balanced mixer which has both the balance pot and a trim cap. I prefer the DBM, but could be an option as an experiment. I think for a start I'll remote mount the Si5351 in a shielded box and see how that works. I know it has to be done, but I love seeing the electronics out in the open! I'm also reluctant to reduce the output level of the Si5351 as the levels are spot on for driving the mixer (if I assume the levels are the same as a SBL-1). That being said, I have noting top loose and the ability to do so is baked into the library. I do know the commercial radio will pick up the Si5351 output directly, but it's at a much lower level. What I didn't film was transmitting a podcast. It's sounds great. As for the filter response, yes i did plot that and it didn't look too bad. Clearly not a good as a commercial 8 pole filter, but ok. It's another area I'd like to look at in due course.
Again thanks to you both for the input.
Charlie
Hi Charlie ,
A quick question, what was the width of the filter at its base and the width at the -3dB point. The difference between the two cannot exceed 300 Hz (or you get the ability to transfer the carrier through).
I am sure you used the center frequency of the filter + half the filter width + 300 Hz as your BFO frequency. If the -3dB point and the base of the filter are greater than 300 Hz then yes it could bleed through, but this shouldn't be by a great deal due to the attenuation of the skirts on the filter.
One other note you could try extra capacitors on the power supply right were the power enters each board just in case the power rails are the problem and carrying the RF through to multiple sections.
keep up the great video's
Best regards
Simon ( M6FJW)
Cheers Simon. I'll look at it again after work. I have a few ideas to split things out as i note the Si5351 is very close to both the balanced mod and the IF amp.
73s
Charlie, ZL2CTM
Simon. I have done some comparison experiments between the homebrew crystal filter and a Yaesu XF-92A filter. The commercial filter has significantly steeper skirts and greater stopband attenuation, which is what I thought. I'm tempted to rebuild the filter using-say-six crystals. That may not be the best approach as i suspect the crystals themselves are not high quality given the price I paid (I'm not complaining as it was always going to be an experiment). The other option that i may look at is use the XF-92A and continue. All good fun! As an aside, there is a small amount of bleed though, but it's not too bad. Moving the BFO up 300Hz also helped (as expected). The filter is the low hanging fruit.
Gidday Simon. I had a good play with varying the Si5351 output levels. It did make a difference, but at the expense of the RX and TX audio quality. 8mA seems to be the best.Thanks again.
Charlie
I think you would get more from improving the diode balance than by improving the bandpass filter. Just the same, it is just not possible to completely cancel your carrier, and you're probably good with what you've got.
what type of antenna used here for CW rig??
Regarding your use of the geometric mean of the typical and minimum specs, I don't think this is sufficient. If your biasing depends on the actual beta, then you need to check your operating point (the quiescent Vc in this case) using both the minimum and maximum possible beta. This is a bit problematic, since max beta is almost never specified, but I use 3 x the typical value, since on the rare datasheets that DO give a max number, that's usually about what it is. Since you use voltage divider bias, beta will have very little effect on the operating point, though, and will actually depend more on the junction temperature. You already fudge this by using Vbe = 0.7, which is reasonable and has always worked for me.
If I built a circuit with a gain of about 5, and then ran its output through a 1/1000 voltage divider, I would probably be asking myself if I needed an amplifier.
976 to 1000
Thanks! I hope the videos are useful/encouraging.