H-Mode mixer

Fairchild switches

The following Fairchild bus/video/analog switches are tested in the H-Mode mixer circuit:

All mixers are constructed on small single sided PCB's with a ground plane of about 1.5 by 2 inches. All components are SMD. The following configuration is used for all mixers

Transformer: Mini-circuits TT4-1A. The same 3 transformers at the same positions are used for all mixers. The transformers are mounted on DIP6 socket for easy transfer between the different mixers.
The squarer is a divide by 2 squarer with the Texas Instruments 74AC74M.
The PCB layouts are as compact as possible and very similar to each other, although some switches allow for a more compact layout than others.

In earlier tests the FST3125 showed considerable degradation in symmetry on the higher bands leading to bad spur reduction. With the PCB's used in this test the difference between the FST3125 and the FSAV332 is not so profound. But still the FSAV332 seems a better choice. Compact construction appears to play an important role with the H-Mode mixer circuit.

Some common observations for all mixers:

Conversion loss was always around -4.9dB to -5.0 dB for all mixers when using TT4-1A transformers. Not much difference there. Conversion loss seems to be determined by the transformers rather than the switches.
The SPDT type mixers (FSAV430, FSAV450, FSAV330, FSA3157) show considerably less 9MHz RF-IF isolation than the FSAV332 or FST3125 when using the fixed divide by 2 74AC74 squarer. The SPDT type mixers need a slightly adjusted squarer, a little bit off 50% duty cycle for better RF-IF isolation.
All mixers are found to follow a more or less 1:3 IMD behavior, but not exactly.
All mixers show a steadily decreasing IIP3 on higher frequencies. Some mixers are more affected than others. The transformers are very important.
Devices that allow up to 7V Vdd are to be preferred for highest IP3.

Next follows an overview of the results of each mixer that was investigated. The following can be noted with regard to all the IP3 measurements:

IP3 measurements are done with 20 kHz spaced tones at around 0dBm on all HF bands. On 80M measurements at levels ranging from -6dBm to +5dBm are carried out to check the validity of the results. (1:3 3rd order behavior)
I have measured the FST3125 at 10 KHz, 20 KHz, 50 KHz and 100 KHz and the results are very similar. No big surprises at different spacing, so I measured all mixers at 20 kHz to keep the report compact.
The IP3 has also been determined separately for the low and the high IMD product. Most of the time the differences are minimal, rarely exceeding 1 dBm, usually are the same. For compactness and accuracy the average of the 2 values is reported.
The RF termination of the mixer sometimes influences the intercept point measurement considerably. After it was observed that a step attenuator set at 0dB between the hybrid combiner and the mixer made a noticeable difference to the IP3 (usually less IMD, but not always…) all IMD levels were recorded with no attenuator between the hybrid and the mixer. Some mixer configurations are more susceptible to this effect than others.

The test result of each mixer configuration is presented in a table with a row per amateur band and the following columns:

Band.
The amateur band the measurement was made on.
BPF-IL.
Band pass filter insertion loss of the filter in front of the mixer.
Mixer-CL.
Mixer conversion loss measured around 0dBm input level.
MDS - BPF.
MDS level observed without the BPF in front of the mixer.
MDS + BPF.
MDS level observed with the BPF in front of the mixer.
IF-Rej.
The mixers RF-IF isolation at 9MHz IF frequency.
Spur-Avg.
The average level above MDS of 132 spurs on 15M.
IIP3 @ -6dBm.
The IIP3 measured at -6dBm input level.
IIP3 @ -3dBm.
The IIP3 measured at -3dBm input level.
IIP3 @ 0dBm.
The IIP3 measured at 0dBm input level.
IIP3 @ +3dBm.
The IIP3 measured at +3dBm input level.
IIP3 @ -5dBm.
The IIP3 measured at +5dBm input level.
Bias.
The bias point voltage with the best IP3.
IIP3.
The IIP3 at 0dBm input level at that best bias point.

Column 2 to 4 may need some extra explanation. These columns are added to show an interesting discrepancy between the MDS level with and without the band-pass filter in front of the mixer. One would expect that the difference in MDS should match the IL of the filter. With most mixer configurations especially on the higher bands this was certainly not the case at all. MDS would drop severely with the BPF in front on some bands. The cause of this is excess spurs that are actually forming an increased noise floor! One can actually hear a crowd of weak spurs at or around MDS level flowing into each other.

Back to H-Mode mixer Introduction

Back to the TOC