got consistent results with a M8 nut brazed in back side of pipe about 12" from port
my welding was a bit scabby
As discussed below, even a little air leaking into the pipe is too much air.
and don’t bother trying to feed me anymore crap about it being my rabbit hole.
I hate it when someone else's rabbit hole merges with mine. Sigh...
Progress here is on a temporary (I hope) hold because the garage A/C failed yesterday afternoon and we face predicted record-breaking heat through Friday. The repairman is due in a few hours but then I/we/he has to figure out how to find space amidst the clutter... er, I mean, extremely well-organized contents to plant a 12-ft. ladder to reach the condenser in the 13-ft. ceiling.
It's important when navigating this rabbit hole to keep in mind that "reversion" refers to two separate, but interrelated, phenomena. In the case of those dyno tests, it refers to the pressure wave that reaches the end of the pipe and then bounces back just in time, and with the right positive or negative amplitude, to either keep all the exhaust from escaping from the cylinder, or to help suck additional charge into the cylinder. Either way, performance can either suffer or benefit in the rpm range where such "pressure reversion" takes place. Here, essentially no physical transport of air/CO2 within the pipe has to take place. The reason is, pressure applied at one end of a long tube that's sealed at the other end also raises the pressure at the sealed end without transporting any of the additional gas molecules that were applied at the open end to raise the pressure.
For AFR measurements, the issue is the actual transport of O2 molecules from the open end of the pipe up some distance into the pipe (albeit, not nearly as far up the pipe as to reach the cylinder). In "reversion" of this type there
is physical transport of O2 molecules, and if any of them make it as far as the AFR sensor they will screw up the measurement.
---- sidebar ---
The stock, open exhaust pipe on my Catalina's is 1.65"ID and 51" long, for a total volume of 1.79 liters. This would be enough to fill the cylinder more than three times over if all of it "reverted." The modified pipe and silencer I have on it now for AFR testing has the sensor 18" from the end, so if a volume of 647cc were uniformly drawn back into the cylinder then fresh air would reach the sensor, which it doesn't. Interestingly, the sampling tip of my unmodified Innovate was 9.67" long, for a volume between it and the exit of the exhaust pipe of 339cc, and there were issues of reversion with it. However, that's not to say that at least 339cc is sucked back into the cylinder on each cycle. But it does that that enough is sucked far enough back in that it, along with mixing, results in at least some fresh air reaching ~9.7" into the pipe.
Keep in mind that an AFR meter doesn't actually measure the AFR. It measures the amount of free O2 (or deficit of free O2) from which the controller then calculates the AFR. If the sensor detects no O2 whatever the controller displays that as '14.7'. Because the controller displays results calculated from deviations from 'zero' it takes only a tiny amount of wayward free O2 to generate a bogus result.
------ end sidebar ---
Turning to the video, the "major dip in the midrange" they found to one extent or other always was accompanied by an AFR that flatlined at or below 10:1, i.e. very rich. Sticking bolts of various kinds had an effect on the h.p. in the ~2500-3500 rpm range, but always the mixture was too rich. This means the "reversion" the narrator talked about only refers to pressure waves, not to O2 being sucked back up the pipe. When they changed the jet to lean the mixture the AFR indeed showed a leaner mixture at higher rpm, but it still flatlined at 10:1 in the midrange.
The video shows unlabeled red and blue curves on most of the plots so it's not always possible to know what's what. But, as far as I can tell, within experimental error the various bolts had a small effect on the h.p. in the midrange but no effect on the AFR.
OK, back to Cyborg's branch of the rabbit hole. His question is
"How much “observer effect” is caused by the probe...."? In the interests of maintaining cordial international relations with our neighbors to the north, after testing the new Mark II instrumentation system I'll shove a probe up the exhaust and see what effect the disruption causes. The answer could turn out to be "a lot," given what the attached photographs show.
The first photograph is a side view of the sampling probe in a 1-3/4" OD pipe, and the second is an end-on view. The paired 3/8"-OD sampling tubes block a little over 10% of the exhaust pipe over a considerable length, and the fumes that manage to get past that obstacle then encounter the fat mount for the sensor at the end of the tube. As the third photograph indicates, I can at least make the life of the exhaust fumes a little easier if I attach a new mounting point to the probe a few inches upstream from the current mounting point.
In the time it has taken to write this, interrupted by lunch and the visit of the A/C repairman, I again have life-giving cool air in the garage. However, the day had a 100 oF head start on the cooling so it is only now getting down to a bearable temperature. As an aside for people interested in magneto capacitors that fail, it was the capacitor in my four-year old A/C unit that had failed. The repairman told me they service units made in the 1950s that still have their original capacitors, but when the mandated change was made to environmentally friendly materials the lifetimes dropped to anywhere from 1 to 10 years. Replacing failed A/C capacitors is the largest category of their work.