FPM5 ~ How to save from pulling out your hair!
12/27/2016 ~ Revised Simulation
This is the value of LT Spice as you can modify the design without soldering one wire. Many options and iterations can be evaluated and the bonus --it can be done quickly. In chasing the elusive problem of too much gain at lower frequencies and not enough on the higher bands, I am being convinced that it now is not the specific circuit I thought the problem. But at the same time there is an opportunity to optimize that circuit. So that is what I did at 4:30 AM this morning. My daylight hours are taken up with caregiver duties so I have found working in the early morning hours does provide some opportunity. I am trading sleep for ham radio design work.
Noteworthy in the new design is forcing the impedance matching by going back to the original primary winding (8 turns) and changing the secondary to two turns. The matching is now 64/4 so that it is a 16 to 1 transformation. The secondary is terminated in 50 ohms and that reflects back as 800 Ohms in the primary. I now placed a 820 Ohm resistor across the primary. I also played around a bit with emitter feedback. These changes are intended to "flatten the gain" yet provide a stable gain over a wide frequency range.
In looking at the gain plot below, the delta change at either end of the intended operational range (75 to 15 Meters) from the center range is about 0.4 dB so this is relatively flat. I will make those changes and then evaluate the real circuit performance versus the predicted performance. This plot looks to be more in line with the desired "flat" performance over the operating range.
Stay tuned and I will report the results.
So now that still leaves us with the problem of too much gain at the lower end and not enough at the higher range. My attention is now focused on the prior stage and that can be done by looking at the input to this amp and see the signal levels. Ahead of this is the band pass filters and ahead of that the SBL-1 Rx Tx Mixer stage. One other area is the signal level coming out of the Si5351. I have not looked at the signal levels from the LO. The LO is operating over the range from 12.5 MHz (75M) to 30 MHz (15M). The output should be fairly constant but that may be a bad assumption. I do know that is a problem with the AD9850. Once the prior stages have been evaluated and a conclusion reached then the circuit elements upstream can likewise be evaluated.
Happy New Year!
12/26/2016 ~ Late Update! Synopsis --that did not work too well.
Despite having the LT Spice simulation shown below the change in hardware did not resolve the higher gain at the lower frequencies and anemic gain on 20/15 Meters. So the good news is that the issue is elsewhere. What I didn't do is to plot the actual gain coming out of just this pre-driver stage and then see what the plots look like in comparison to the LT spice plot. But I am now suspecting the issues are elsewhere. The positive news is that we are eliminating possibilities!
Well it has been nearly a week since getting the FPM5 in a box and I can truly say this has been some ride. Firstly the adage "if it ain't broke don't fix it" rings true. The rig worked great "al fresco" hay wired into place but once getting it into a box with proper grounding and short leads it is then that the warts show up. Thus I have been working down a list of issues and one by one these are being resolved.
Finally I was at the point of why there was way too much intermediate RF gain at 75 Meters and the transmitted signals were anemic on 20 & 15 Meters. This is where we can be tripped up by thinking we know something, discounting it and then keep coming back to the same point with nothing being changed.
A long time ago I developed a circuit that is a relay switched bidirectional amp consisting of two legs and employing the ubiquitous 2N3904 NPN transistor. This device is usually available in quantities for a price around 4 cents. You can beat that! Typically my bidirectional design had one leg be the receiver RF amp on Rx and on transmit the other leg was the pre-driver for transmit. This has served me well and is typically my "go to" circuit. I guess my experience with the KWM-4 should have been a large blinking sign but some how my brain ignored it until now.
Finally today I decided to put my circuit to the acid test using LT Spice. About five years ago I hay-wired it together --it worked and I moved on. But I had never subjected the design to a rigorous evaluation. That was my first mistake! Simulating the circuit with new values provides a rigorous look at a real gain plot and this looks much better.
Here is what I observed. First there was way too much inductance on the transformers. By empirically reducing the turns on the primary to 5 turns and the secondary to 2 turns the impedance transformation was 25/4 = 6.25. The secondary side always sees 50 Ohms because of the fixed resistive load of the 100 Ohm fixed and 100 Trim pot in parallel. The output is taken off of the center wiper through a 10NF cap. This says the primary load is reflected back into the secondary as a load of 300 ohms. The next thing I did was to look at the emitter of the 2N3904. Initially I had a 50 Ohm resistor in parallel with a 100 NF cap. I tried adjusting that value and with 1K in parallel with the 100 NF this seemed to provide a flatter curve and shifted the center hump past 10 MHz which now aligns with the intended operation on 75, 40, 20 and 15 Meters.
The modified circuit is presented below: Note the two 50 Ohm resistors in series simulate the 100 Ohm trim pot with the center wiper at midpoint.