I went ahead and made the mod as suggested by Allard to improve the CW Carrier Waveform shape on the BITX40.
The 2.2uF capacitor value to ground wasn't correct (I think Allard based this value on an earlier incorrect comment from me).
I calculated that the required capacitance to achieve a > 5mS rise time on the waveform as about 270 nF so I ended up
using both a 220 nF and a 47 nF disc capacitor wired in parallel to approximate this.
Previously I was using a 4K7 resistor between Raduino D6 and C107 for the carrier mod. I checked output power on CW
as well as residual carrier on LSB with PTT closed with no mic audio, both before and after the mod and noticed no significant changes
so I don't think that there are any undesirable side-effects from this mod.
I don't currently have a working scope so I used my FLEX-1500 to monitor the transmitted CW signal from the Bitx40, connected to a dummy load.
PowerSDR has a scope view which allowed me to monitor the shape of the received BITX40 waveform with sufficient resolution to demonstrate a
noticeable improvement in the the waveform shape.
I have attached a couple of .PNG files that show the before and after cases. You can see that before the mod the CW waveform was very square looking, whereas
after the changes it has much more of a desirable trapezoidal shape. The received post-mod signal also sounds more pleasing to ear.
P.S. Next on my list is to try providing an uninterrupted supply voltage to pin 6 of the LM386 by tombstoning R111 and feeding the top end of it with +12V.
My hope is that this might reduce or even eliminate the remaining BITX40 audio thump on the TX -> RX transition.