Input matching transformer
Posted: Wed Jul 15, 2020 10:36 am
I'm interested to know why some push pull LDMOS amplifier designs are using just a ferrite core with copper wire while others are using coax inserted through 2x ferrite core's.
First example using coax through ferrite cores, including an initial piece of 50R coax to act as a balun Second example, the well known Chinese amp which uses just a single ferrite core with 2 turns of copper wire on the primary, also acts as a balun I have tested out both methods of matching the 50R unbalanced input to approx 10R balanced measured gate to gate, I found very little difference.
The device I tested with was the MRFE6VP6300H which is a 300W device, bias set to 2.6v and 250mA with no RF drive.
Using a single ferrite core with 2 turns on the primary, and 10R padding resistors I got a no measurable reflected power on the input, however the 3rd harmonic was sitting at around -19dBc.
Using the coax method with 2x ferrite slugs, the input SWR went up very slightly to around 1.1 to 1.2, however the 3rd harmonic went down by a very small insignificant amount - in the order of 1db or less and is now just below -20dBc. There was no change in the output power and efficiency.
I also have 2x 10K multi turn trimmers to fine tune the gain of each gate independently, this made no difference to the 3rd harmonic in terms of dBc, however I was able to almost cancel out the 2nd harmonic where as I couldn't with just a single bias feed to both gates - this showed there is a slight side to side difference in the device.
One thing I haven't looked at yet is the phase variation between both gates, they are supposed to be fed 180 deg out of phase for push pull operation - I'm assuming using coax instead of a wire wound input transformer reduces possible phase variation on higher frequencies and allows for a more accurate phase shift at 180 deg.
First example using coax through ferrite cores, including an initial piece of 50R coax to act as a balun Second example, the well known Chinese amp which uses just a single ferrite core with 2 turns of copper wire on the primary, also acts as a balun I have tested out both methods of matching the 50R unbalanced input to approx 10R balanced measured gate to gate, I found very little difference.
The device I tested with was the MRFE6VP6300H which is a 300W device, bias set to 2.6v and 250mA with no RF drive.
Using a single ferrite core with 2 turns on the primary, and 10R padding resistors I got a no measurable reflected power on the input, however the 3rd harmonic was sitting at around -19dBc.
Using the coax method with 2x ferrite slugs, the input SWR went up very slightly to around 1.1 to 1.2, however the 3rd harmonic went down by a very small insignificant amount - in the order of 1db or less and is now just below -20dBc. There was no change in the output power and efficiency.
I also have 2x 10K multi turn trimmers to fine tune the gain of each gate independently, this made no difference to the 3rd harmonic in terms of dBc, however I was able to almost cancel out the 2nd harmonic where as I couldn't with just a single bias feed to both gates - this showed there is a slight side to side difference in the device.
One thing I haven't looked at yet is the phase variation between both gates, they are supposed to be fed 180 deg out of phase for push pull operation - I'm assuming using coax instead of a wire wound input transformer reduces possible phase variation on higher frequencies and allows for a more accurate phase shift at 180 deg.