Recently I was drawn in the direction of a simulation software package called LT Spice, Pete Juliano N6QW uses it in several of his articles so I gave it a shot and not having used it before it was surprisingly easy to use. I modelled my original IF amplifier based on the Hybrid Cascode design by Wes Hayward, W7ZOI, and Jeff Damm, WA7MLH in the 2007 QST article, generally I was pleased with its performance and it worked – however, typical of me I started to overthink the design and decided to re-model it in LT Spice and the test results in the simulation pushed me to “tweak” the original design to my own satisfaction and I ended up with a TODO list of major surgery.
The original design I followed was based on a 9MHz IF frequency, I am using 455KHz so several component values were changed to make it work in the MF range, however when I modeled it in LT Spice I noticed several peaks and troughs over a 1THz sweep – yes 1Hz to 1THz (Tera Hz thats a lot of zeros after the one!) and some quite significant gain peaks in the HF spectrum, I guess this was to expected as the amplifier design is not “tuned”.
I managed to locate a supplier called Mouser who stock thousands of the 455KHz variable IF transformers cans, I modelled the new design using the 42IF102 white cap device which has a primary impedance of 60K Ohms and a secondary impedance of 500 Ohms. I did not use the center tap as in traditional bipolar amplifiers you would connect the collector of the device to the center tap and the top of the transformer to the +V power rail, in the LT Spice model this created broad-band characteristics, with two sharp peaks at 455KHz and 10.96MHz – no good to me as my first IF of the transceiver I am using this amplifier in is 10.7MHz!
The input is matched from 50 Ohm to 3.3K Ohm to the input of the first FET gate and the output is matched from the 500 Ohm secondary on the final IF transformer to 50 ohm using an L Match circuit, reasons for this as follows:
- At 455KHz traditional broadband matching transformers get big and need lots of windings.
- The L Match gives me an extra layer of selectivity at the IF frequency.
The gain plot below tells me I will achieve around @90dB of gain at full AGC voltage of 4V, I stepped the voltage and took not of the gain vs voltage and as expected it is not linear but was not far off. 0V produces around -480dB which is of the scale – even the strongest broadcast station would struggle to make it through, around 1.5V we see around -15dB and at 1.75V we get around +7dB and we get around 3.5dB for each 50mV step up to around 2.5V when we start to see around 10dB for a 100mV step – basically you have an AGC range of 390dB, I guess once this design has been fabricated on a PCB these figures will change slightly.
Full AGC Bias applied.
Various AGC Bias levels.
The LT Spice circuit.
Here is the modified hybrid Cascode circuit, notice I added a resistor across each IF transformer, this was to tame the gain and even out the response of the amplifier over the full AGC range.
Next step is to get this fabricated onto a PCB and see it action for real.