Increasing output power from lftx with resistor mod gives saturated output


I have attempted to increase the output power from my Rev 2.2 LFTX as
per instructions from Matt E. on this list dated Aug 3 2013 in which
he suggests reducing the values of resistors R3 and R5, but I am seeing
a clipped output at peak values much less than expected.

Using the LFTX schematic as a guide, I left the resistors on
the board untouched, but soldered wires to either side of R3 and R5 to
add additional resistors in parallel. My first attempt was with
resistors of value 220 ohms; I figured this would about halve the value
at R3 and R5 which are originally 225 ohms. I do see some increase in
output power, but very minimal. Specifically, whereas before I was
observing 1 V P-P, or about 2.5 milli-watts on an oscilloscope that is
50-ohm DC coupled, I am now seeing 1.34 V P-P, or about 4.6 milli-watts
with a 1 kHz sine-wave. To obtain this without clipping I had to reduce
the amplitude on the USRP sink block from 1 to about .77, othwerwise the
tops of the sinusoids flatten anywhere above the 1.34 V P-P level. So I
appear to be saturated.

I am using a USRP1 Rev. 3. I verified that the LFTX output op-amp in
question is supplied with +/- 3.3 volts, so I don’t see why it is
clipping at half of this.

I am hoping to use the LFTX in an HF Ham radio application. I have
mini-circuits ZHL-1A TX pre-amplifier for this purpose, but was hoping
to obtain a little more output from the LFTX than I am currently getting
to drive the ZHL-1A. Any thoughts?