I would be genuinely interested to learn if I had any of that wrong, and I'm willing to accept that I might! I almost certainly don't have as much experience with dynos as some people on here, but what I do know, I was taught by the Dyno Dynamics guy who sold my uni a dyno and taught us how to use it. Maybe it was just our software that worked like this, or maybe he just didn't know what he was talking about...
Mike, nice guy that he is, is a very good PR man for DD type rollers and how they are "the best" etc etc....... He was of course selling them, so they would'nt be anything other than that.
DD, are a pet hate dyno, as when you really dig into these, how they work, what they measure, how they measure etc etc, you get past the sales pitch and see the truth.
DD only measure atw. atw figures, much as US cousins and DD dyno fans are not the "defacto" fool proof ONLY Proper way of dynoing..
DD measure their atw figure, and depending on the transmission type told to the dyno, "adds" its factor to report a "flywheel figure". This is no better than adding x% to the atw figure to get your "estimate" of fly figure.. And this is said to be accurate? Really?
Atw figures, will alter with temps, tyre pressures, make of tyres, speeds run (gear used for example), and will change.
My dyno measures atw, and also coastdowns... and you see the atw figures change with temps and runs.... as of course you would expect as things warm up in the drive train. If only measuing atw and adding the % factor, this simply cannot be as accurate, and rolling losses are not measured, and a % factor is applied to "estimate" tge fly figure.. So to differ in opinion to you, coastdown losses, are meaningfull, and the more questionable figures are those dynos which only measure atw with no account of coastdowns.
2 cars recently on my dyno.. both gave 355atw figures, but differing fly losses, and the drivetrains/tyres etc were the difference... If on a DD they would have been given the same @fly figure, but were quite different overall.
Where DD rollers, in their own promo video's they show you, show cars riding up onto the front roller, halving the contact patch, yet will report the atw figures as being "ACCURATE" when they clearly only had half the frictional loss when riding onto the one roller, typifies the scenario I describe.
My own dyno is a Dastek/SunRam hybrid.. mongrel of sorts, but measures std cars at std power, and remapped ones at Realistic, not BS high figures, and often correlates mass airflow to predicted power with a few %.
Its a tuning tool at the end of the day, and Dyno Lottery of course applies..
To describe dyno's as you gave with your very broad brush tho, is based on the DD sales pitch and not the wider picture. DD are'nt all that.. I was looking to buy one, and the more I asked the more I did'nt want to buy into their sales pitch. They went bankrupt anyhows...