Tuesday, March 21, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.03

The Dual Conversion Scheme an evaluation of Frequencies.

In the KWM-4 I was presented with a real engineering problem that being how to take advantage of the Collins Mechanical Filter capabilities yet deal with its low frequency of 455 kHz. The path quickly leads to a dual conversion scheme, where you can manage the gain at the higher IF and manage the selectivity at the lower IF.
This is a clue in that at the higher IF the filter does not have to be 2.1 KHz wide but something on the order of 7.5 KHz will certainly keep down the crud ending up in your receiver band pass. The lower IF (Collins Filter) can do all the heavy lifting as now signals in the pass band coming into the filter are very narrow and the mechanical filter is essentially slicing that down to about 25% of the bandwidth -- 2.1 KHz. Below is the frequency scheme for the KWM-4.
Five years ago this is how the frequency mixing was handled. This exemplifies the concept of gain at the higher IF and selectivity at the lower IF. A filter at 10.7 MHz was chosen as the higher IF because given what I had at hand five years ago this worked with the COTS (Commercial Off The Shelf) components that could be easily and readily purchased. This design has a shortfall in that because of the high IF chosen, operation on 30 Meters was questionable. For some this is a deal breaker.
But today were I building this I would have chosen another IF frequency as frequency translation with an Arduino + Si5351 is a far easier task. The uBitx uses a homebrew multi-pole filter at 45 MHz  MHz and that is a good choice of frequency as it avoids the ham bands and does provide for 30M operation. However I would stick with a packaged 45 MHz crystal filter from ECS that does cost about $17 (Digi-Key) but has some really desirable and predictable parameters. It is good for 30 dB of attenuation for +/ 3.75 KHz and the stop band is 80 dB. The Z in/out is 350 Ohms and that is easily matched to 50 Ohms with a 3 Turn to 8 Turn broadband transformer (350:50 = 7:1, 3^2 = 9 and 8^2 = 64, 64/9 = 7.111 --close.). The injection frequency into the 2nd mixer is a matter of what you have and also close examination of the mixing by products.
In the KWM-4 the 10.7 MHz filter with a 7.5 KHz bandwidth resolved the subtractive mixing issue where 10.245 - .455 was out of the filter pass band. So lets say you wanted to use a commercial crystal filter such as the INRAD Model #351 which has a center frequency of 9.0 MHz. Thus the injection frequency would be 45 + 9 = 54 MHz which still puts it outside the ham bands --a bonus. Then the third clock would provide the normal 8.9985 and 9.0015 MHz BFO frequencies --OR a third BFO at 9.000 MHz for CW. The KWM-4 CW scheme would work perfectly here.
The Arduino and Si5351 would provide a far easier approach in a today build of a KWM-4 --but with some noodling I worked with what I had five years ago.
In a recap of a DifX version of a Dual Conversion Transceiver I would use the following:
  • 1st IF at 45 MHz using the ECS packaged crystal filter ~ 7.5 KHz wide
  • CLK0 on the Si5351 would provide injection frequencies above the 1st IF ranging from 46.8 to 75 MHz --all with in the capabilities of the device to give 160-10 Meter all band coverage. A quick math analysis shows that the injection frequencies required for all of the ham band avoids the second mixing frequency. The LO injection frequency for 7 MHz would be 52 to 52.3 MHz and the LO frequencies for 10 MHz would be in the 55 MHz range
  • CLK1 on the Si5351 would provide a 54 MHz fixed inject frequency to convert the signals to 9.0 MHz (54 - 45 = 9 MHz)
  • CLK2 would provide the USB/LSB and CW Carrier Oscillator frequencies
  • The INRAD Model #351 is a 4 pole 2.3 KHz wide filter with a Z in/out of 200 Ohms an easy 4:1 match to 50 Ohms. A word here about homebrew crystal filters. It is not a simple matter to build a good quality crystal filter especially for someone who has never done one. You do need test equipment and you do need to understand what you are doing. There are some excellent tutorials on how to do it --but excellent  results for the neophyte with no test equipment would be like winning the Mega Millions having only played once and having only bought one ticket. The two filters suggested will cost you about $50  and one of them is a surface mount. But their specifications are well known as is their performance. Let me not discourage you from homebrewing a filter --but I am for a sophisticated project like this. A six pole 45 MHz filter may require the purchase of 20 to 30 individual  crystals before you find six that are close in frequency. You can't simply buy 6 crystals and think you are there. If by happenstance you are after buying only six, then I suggest you go out and buy one lottery ticket --for you are indeed a lucky person.
  • For the Bi-Lateral Amps you have a choice of the TIA or the Plessey Amps.
  • For DBM's I would use the ADE-1L (low drive requirements at 3 dBM).
Thus the Dual Conversion scheme offers some real advantages but also brings about the need to be careful in frequency selections.
Pete N6QW