Thoughts, Concerns and Considerations-Stop for a Minute to Consider Frequency Schemes.Addendum 12/02/2015 See the video at the end of this post on Signal Handling Tests.
Part 16 was jam packed with information BUT when homebrewing amateur rigs it is always best to take two steps back and think about what ALL can happen when you make changes. I added information to Part 16 in response to inquiries about circuit information that would enable the Simpleceiver to be placed on 20 Meters.
Initially the request was for an RF amplifier stage and then I provided the Band Pass Filter constants thinking that I would receive requests for that information as well. In passing in Part 16 I mentioned you would have to change the LO frequency so that you would receive USB instead of LSB. This was done so that the same BFO frequency would be used for either band. Sometimes that is not a good solution and let us examine the WHY. Problems show up both on receive and transmit with the transmit being more serious with out of band signals when a frequency scheme has not been thoroughly evaluated!
The IF frequency of 12.096 MHz for this project was arbitrary to some degree --I had a bunch of crystals in the junk box and quickly saw that a VFO at 5.0 MHz in an additive mix (up convert)would net you the 40 Meter band. For those that have the ARRL Publication QRP Power you can see a project from W1VT, (Zack Lau) that uses that same approach for a 40M CW transceiver. So it has been done before. But this is where the professional designer (I am not one of those) would look at all mixing products and how the circuit topology can adequately handle all the possible frequency outcomes.
So in traveling down our circuit path we have the RF amp stage that was designed using LT Spice as a broader band amp versus a narrow band amp. That could or could not be a problem based on the frequency scheme so that is the first thing to evaluate. In our case the RF amp is centered on 7 MHz and drops off by 3 dB at around 14 MHz. The addendum input in Part 16 centered the RF amp on 14 MHz.
Next we have the 40M Band Pass Filter and that has a fairly good response curve where we have covered the SSB portion of the Band and all but the lower 40 kHz of the CW band are dropped off. C'mon this will be a SSB transceiver when we are done! So that part looks OK.
Now on to the mixer stage and this can be where the 1st problem is encountered. In the Simpleceiver design we are up converting the incoming frequency to the IF of 12.096 MHz. Thus our Local Oscillator PLUS the incoming signal are mixed so that the result is always 12.096 MHz. If our incoming signal was at 7.200 MHz then our LO would be tuning 4.896 MHz. If the incoming signal was at 7.15 MHz our LO would be at 4.91 MHz. Do you see a pattern here of reverse or backward tuning LO? The lower in received frequency the higher the LO. This was the same situation in the W1VT CW rig. With a digital LO the math in the software enables you to have the display read properly. You can even switch the encoder leads so that Clockwise is up in frequency or if you are left handed then CCW is up in frequency. Lots of flexibility here.
Now with this scheme the BFO frequency must be set so that you will copy Lower Sideband. An alternate way for this frequency scheme would be to set the Local Oscillator above the incoming signal so that the LO - the Incoming Signal = the IF. To receive 7.2 MHz using this approach would entail having the LO at 19.296 MHz (19.296 - 7.2 = 12.096). But when you use this approach you have a sideband inversion so the result would be USB with a BFO crystal frequency set ABOVE the filter center frequency. To receive Lower Sideband (LSB) the BFO frequency would have to be set BELOW the filter center frequency.
In our design for 40 Meters because we are not subtracting frequencies but up converting then there is no sideband inversion so that the BFO frequency ABOVE the filter Center Frequency is used for LSB and a BFO below the Center Frequency would give USB.
But in our mixer stage (the SBL-1) there are two mixing products that result (this is where there can be problems if you do not chose frequencies wisely). For our scheme we have the sum product of the incoming plus LO which results in an IF of 12.096 MHz. But we also have Incoming - the LO. Going back to 7.2 MHz our LO was at 4.896 MHz so their difference frequency is 2.304 Mhz. Our IF amplifier will easily reject that component as there is some 10 MHz of "space between these signals. Plus don't forget the Crystal filter and its ability to reject out of band signals.
Here is a plot of the 1st IF amplifier response to demonstrate this point. So our 2.304 MHz signal is about 34 dB down which is OK for our purposes. You would probably like to see this be greater than 40 dB (ideal). For such a simple design this is adequate and don't forget that the BPF ahead of the SBL-1 is not passing any 2 MHz energy and our crystal filter's ability to reject out of band signals.
The 20 Meter Case
So the above validates our choice of IF and LO frequencies for 40 Meters. While this may border on the anal retentive, let us examine what took place with the simple request to move everything to 20 Meters. That is the real danger with LT Spice --5 minutes of time and you change a few parameters and Boom a new design. But what is the impact of those changes?
So we started with the RF Amplifier and we have a plot of the response of that change to 20 Meters shown below. Our 3 dB points makes this amp work from about 8 MHz to 24 MHz. Initially we would say hey this will work for 30, 20, 15 and 15M. Did we strike Gold? Maybe not as a lot of energy across a broad spectrum is being passed on to the two section Band Pass Filter. In extreme cases it may be necessary to have a three stage BPF. So Broader is not always better!
Our next change was the 20M Band Pass Filter design. Again another 5 minutes with LT Spice and we have a new Band Pass Filter which is shown below.
This plot shows that about 1 MHz above and below the center frequency that the signal drop off is about 35 dB, which makes this acceptable for our use. Mind you the Band Pass Filter is not transceiver specific --it is simply 50 Ohms in and out and this one has a nice range.
So now on to our SBL-1 mixer stage (again this could be a problem with mixing frequencies). So now our incoming signal is at 14.2 MHz and the IF is at 12.096 MHz. We get the sum and difference frequencies at the output of the incoming and +/- the LO. Thus if the LO is 26.296 MHz and the incoming is at 14.2 MHz the difference result is 12.096 MHz --our IF. This would work well as the sum result of the Incoming plus the LO would be 40.496 MHz which is way outside any of our tuned networks. This is an important point about a term called high side mixing. By placing the LO above the incoming that neatly takes care of any sum mixing products as they are well outside the ranges of the tuned networks.
Because of high side mixing we have a sideband inversion and to receive USB the BFO has to be set above the filter Center Frequency. So the BFO frequency we used for the 40 Meter version will work on USB with the LO above the incoming frequency. There will be no need to shift the BFO frequency.
Now lets see what happens if you simply moved the LO below the incoming frequency. If we have the incoming at 14.2 MHz and our IF is at 12.096 MHz presumably we could put a 2.104 MHz LO signal into the SBL-1 and the results would be 12.096 MHz and 16.304 MHz. That 16.304 MHz component is just a few dB down from our 1st IF amplifier response curve and hopefully we have pretty good crystal filter that would fully reject that component. But more space between frequencies is always better. So low side mixing is not a good choice for the 20 Meter Band. In Part 16 where we spent 5 minutes looking at moving the LO frequency to 2.10 MHz, our further detailed look tells us to avoid that LO frequency range.
This becomes even more critical on the transmit side and improperly choosing Crystal Filter frequencies and mixing schemes could let unwanted mixing products slip through the Band Pass and Low Pass Filters. In the case where higher power is being run "traps" (tuned networks) designed for the filter IF frequency by necessity would have to be included in a low pass filter design. Take our 20 Meter example for instance 12.096 MHz will pass through a 20 Meter Low Pass Filter. But would have a tougher time with a 40 Meter filter especially if the cutoff is at 8 Mhz.
Thus having a 12.096 MHz filter was by convenience (had them in the junk box) and does enable using a 5 MHz VFO or VXO versus a digital DDS or PLL, but there are significant concerns about simply taking 5 Minutes with LT Spice and designing new bands and networks without due consideration of the possible shortcomings and unwanted signal byproducts.
A word or two here about the Simpleceiver actual hardware build. I have received an email asking me "WHY" did you do this and that which I promised I would answer those questions in a blog entry.
- The hardware layout was done with some thought based on that the sections were built sequentially starting from the back end. Get the audio amp working first and then move forward through the project so that as each module is built and tested it then becomes a part of the overall test system. A recent paper I read has another view --start with the hardest module first and get that working. I guess I don't subscribe to that approach as a progressive build lets you identify immediately when a stage is not working because everything behind that stage IS working!
- The use of coax cable and how did I decide when to use coax and when not to. I guess if anything is carrying RF plus if you keep all the ins and out at 50 Ohms then coax interconnect is the answer. In the case of the output from the product detector to the audio amp --that is not coax but miniature microphone cable since the output of the Product Detector is Audio. Each board that has power to the board also has a ground return to a common point supplying the minus power. When his gets done all circuit boards will be mounted in a metal box and so all circuit boards will share a common ground plane. In addition I connected a short direct ground from the IF Amp/Filter Board to the Product Detector Board.
Signal Handling video taken 12/02/2015
Enough words for now. Hopefully this has been helpful.