The main problem of the oscillations was the strong VFO energy (6 volts peak to peak, or +7dBm) going through the mixer to the input of a single MOSFET post amplifier. The VFO signal, combined with the second and third harmonic signals, would drive the amplifier to oscillate.

I had tried using the MOSFET single balanced amplifier as the post mixer amplifier, but the excessive use of transformers kept me from seriously considering it. Both the mixer and the amplifier has a 1:4 at the input and output, and putting them together would put two transformers in a row.

While looking at the two sitting in a row, I pondered how this might be made into a quad mixer. This could eliminate the transformers between the circuits. I looked in the 2000 ARRL Handbook and came across Colin Horrabin's, G3SBI, quad FET mixer.

After looking at how he used his transformers, I came up with the idea of tying the single balanced amplifier into the input leads of the output transformer of the mixer. This worked great. It was solid as a rock. Never could get it to oscillate, even with my signal generator in overdrive.

The capacitors (.01s) on either side of the middle transformer are the coupling capacitors to the amplifier. They go to Gate 1, with a 100K resistor to ground. G1 is marked on the mixer and amplifier MOSFETs.

In probing the circuit with an SA, the mixer and amplifier seemed to work as one unit. Any measurements between the mixer and amplifier were never completely trustworthy. The gain measured at the output of the mixer would be the same gain at the output of the amplifier. But when the two were disconnected, the gain of the mixer would be 10 dB less.

The single balanced MOSFET amplifier can handle the excessive VFO energy from the mixer without oscillating and provides 7db of gain.

In the picture above, the first mixer, post mixer amplifier and crystal filter are shown. This is the most intense area for LEDs, with 8 in a short space. Six LEDs and two IR LEDs.

The combination mixer-amplifier gives about 14dB of gain, which is needed to overcome crystal losses, convert through the second mixer (with a little gain here) and provide enough signal to mask the noise of the 455kHz IF strip.

If the S-meter is carefully adjusted, static crashes (band noise) will be moving the S-meter, activating the AGC. Tuning the bandpass control, the S-meter will rise when a band is tuned in.

With the SA, I played with all the resistor values to see how this mixer could be fine tuned. I found several interesting improvements.

Gate 2 Bias

The first interesting improvement was running the Gate 2 bias at .05 volt. This is accomplished by placing a 470 ohm resistor to ground from Gate 2. This increased the gain by 2dB, and the brightness of the LEDs was a direct indication of oscillator injection levels.

When there is no VFO drive, the mixer LEDs are completely off. As you raise the VFO injection level with the trimpot at the first amplifier of VFO amplifier chain, the LEDs will slowly become bright.

The LEDs help when adding a frequency counter or any other appliance to the output of the VFO. Any change in the LED brightness will tell you how much you are loading down the VFO.

The best adjustment of the VFO injection (100K trim pot at the first VFO amplifier) as determined by the SA for best IP intercept point and conversion gain for the mixer is as follows:

Locate the first amplifier of the VFO amplifiers, adjust that trimpot so the LED at this MOSFET amplifier is the brightest it can go. Then back it off until the LED just starts to dim. The LEDs at the mixer should not change in brightness.

Gate 1 Resistors

The second interesting improvement was changing the value of the G1 MOSFET resistors. Normally, this value is 100K in the circuits of this mixer. However, as this value was lowered, dynamic range of the mixer improved, with only a 3 dB decrease in gain.

Maximum sensitivity was found using 100K resistors at Gate 1. Maximum dynamic range was found using 470 ohm resistors at Gate 1.

If the receiver is used for SWL reception, 470 ohm resistors should be used at Gate 1 for best stability and dynamic range. SWL signals are much stronger than Amateur Radio signals, therefore, maximum dynamic range of the first mixer is needed.

In the Amateur Radio version of the receiver, the first mixer uses 100K resistors at Gate 1 for maximum sensitivity.

The second mixer uses 470 ohm resistors at the G1 gate because sensitivity is not an issue there, but having good dynamic range and stability is very important.

Swamping the Middle Bifilar Ferrite Core

A third improvement that has proven very important is placing a 2.2K resistor on the end wires of the bifilar transformer in the middle of the mixer/amplifier combination.

The 2.2K resistor is soldered on the traces underneath the board in the first run of boards and has been incorporated into the Rev 2 boards as shown in the picture below.

This modification is highly recommended. The oscillations have not occurred in all builds of the receiver, but its benefit in stabilizing the mixer/amplifier has been tremendous.

Even though tests with the spectrum analyzer showed no oscillations at any drive level in the mixer/amplifier, on the air usage had popping and oscillations below the 40 meter settings on the bandpass filter.

The best guess for the oscillations was regeneration between the input and output of the mixer/amplifier. Attempts to remove it were centered at the middle bilifar ferrite core.

A 10K trim pot was connected on the end wires of the ferrite core and adjusted till the oscillations disappeared. The oscillations began to subside at 2.8K. After some experimentation, a 2.2k value gave the best stability without decreasing gain.