I was planning on welding the bung in about 5" from the rear of the header pipe.
For what it's worth, the photo shows where I welded the bung on my extra Gold Star pipe. The sensor shouldn't be too close to the head (because it could get too hot) or too close to the end of the pipe (because of the fresh air. Also, it needs to be toward the top of the pipe (so water condensation at startup won't hit the hot sensor), and somewhere out of harms way. The location I picked satisfies all of those constraints. Although it makes no functional difference, for aesthetic reasons I took a little extra time to make an angled adapter piece between the bung and pipe so the sensor would be vertical.
Since Oldhaven asked about A/F meters and Robert mentioned 'Innovate', I use Innovate's portable LM-1 meter. The company replaced this with the LM-2 model in 2008, but haven't make any changes since then. I mention this because a larger company acquired Innovate c2010 and as best as I can determine there haven't been any improvement to their portable A/F meters in a decade, which is nearly a century in electronics-years. Although I'm happy with the performance of my LM-1, and it's possible the decade-old LM-2 may be as good as anything else currently on the market, if I were in the market for a portable A/F meter today I would research all the alternatives before buying one.
As an aside for any LM-1 readers looking at this post, I hadn't used my LM-1 with my current laptop and when I recently did I discovered to my chagrin that the latest 3.37 version of the Logworks software for it has a glitch that won't let me transfer saved data from the meter (the program crashes when I try). Innovate's technical support was completely useless, leaving me to deal with this myself. Luckily, I had a copy of the installation file for 3.01 on a CD and, despite Vista having been the version of Windows when it was issued in 2009, it installed on Win10 without any error messages and works.
There are several (big) advantages to a unit that has several inputs and also logs the data for later study. I use mine with a tachometer input along with a (universal) throttle position sensor I made based on a potentiometer and AAA battery. Looking at the logs in the comfort of my home, rather than trying to memorize a few numbers as I fly down the road, lets me see see what is happening with the A/F ratio at various throttle positions, as well as see transients when I, for example, slam the throttle open (it records data at 12 Hz).