PA modification by Roelof PA0RDT

Roelof PA0RDT described his PA modification for the QMX+, consisting of doubling up on the PA transistors. The standard configuration has 2 + 2 BS170 transistors, Roelof doubled it to 4 + 4. This allowed increasing the supply voltage to 13.8V, or running the WTST transformer (usually used for 9V builds) at 12V. It approximately doubled the power output.

So Roelof essentially transformed his QMX+ into a 10W transceiver. Some relevent forum topics:

https://groups.io/g/QRPLabs/topic/114750835 
https://groups.io/g/QRPLabs/topic/116307831
https://groups.io/g/QRPLabs/topic/116363153

TL;DR: I modified my QMX+ as per Roelof's modification; results are inline with Roelof's and I did not observe any problems. 

Tests by Hans G0UPL 

I modified a QMX+ Rev 3 running production firmware 1_02_009. All experiments were carried out with a 12V supply. The Receive current was measured as 106mA. For the efficiency calculation the receive current is subtracted from the total measured current. Voltage and current measurements are just the display of my inexpensive variable bench power supply which is set to 12V. Dummy load and 26dB of attenuation feed the Advantest R3361C Spectrum Analyzer for harmonic measurements. The power output is calculated by observation of Peak-Peak sinewave waveforms on my Owon XDS3102A oscilloscope. 

I soldered an extra set of BS170 transistors on top of the current set. Not being entirely brave enough to go without any heatsinking on the new transistors (as Roelof did), I bolted in an extra piece of old PCB to act as a heatsink (it is in fact, a piece of PCB chopped out of an unpopulated QMX board). 

Click image for full-size photo:

Three scenarios were tested:

1. Unmodified QMX+ 2+2 BS170 transistors, RWTST transformer (3:2)
2. Modified QMX+ 4+4 BS170 transistors, still with RWTST transformer (3:2)
3. Modified QMX+ 4+4 BS170 transistors, transformer rewound as 3:3 WTST

Finally SSB calibration was re-run on the final modified QMX+ (scenario 3). SSB calibration completed without incident; a few of the synchronization optimization results look a little strange but then again that sometimes happens even with normal standard builds. 

I did NOT try operation at 13.8V supply. 

Results and conclusions

• Generally I believe the results of my modification and measurement are consistent with Roelof's claims.
• Simply doubling the number of transistors (scenario 2) produced an increase in power output on all bands except 6m where a small decrease in power was observed; most probably this is due to higher load by twice as many MOSFET gates, on the driver chip. Most bands also showed a slight increase in efficiency and in many cases, lower harmonic levels.
• Changing to the WTST transformer (3:3 ratio) provided a further increase in power output and a decrease in efficiency.
• No overheating was observed.
• SSB calibration completed without incident. 
• Following the modification, despite poor 20m conditions for the last few days, a successful WinLink connection was made on 20m with LZ3CB my nearest Winlink node (635 km distance). I also connected with OE5XIR on WinLink (1731 km).

Detailed results are presented below. 

1. Standard configuration (2+2 transistors, 3:2 RWTST transformer)

2. Modified: 4+4 transistors, RWTST transformer (3:2 ratio)

3. Modified: 4+4 transistors, WTST transformer (3:3 ratio)

SSB calibration stored on QMX+ device before the modification

Phase pre-distortion:

USB sync:

LSB sync:

SSB calibration after the modification

Phase pre-distortion:

USB sync:

LSB sync: