All the work reported on the forum recently on improving performance of RH4 heads is exciting and makes for some very interesting reading. I applaud all the contributors involved in the various aspects of this undertaking.
Because I have a stock ‘74 CDO with an RH10 head, I have no context for the recited performance improvements, e.g., improved low/mid-range torque, more power through the higher rpm range, better top end, etc, reported after retrofitting an RH4 head with an intake port sleeve variant. I’m not sure but believe that such performance enhancements are intended to make an RH4 headed bike equivalent to an RH10 headed analog, all other things being equal. Or maybe I’ve got it wrong and the goal is to make an RH4 head exceed the performance of an RH10 head. Or maybe it’s just to improve the performance of an RH4 head, but with no specific performance goal in mind.
Regardless, all the insert approaches (machined, printed aluminum, etc) are providing qualitative performance improvements (seat of the pants dyno, need for re-jetting, etc) which makes me all the more curious to know quantitatively (flowbench, dyno) how much performance actually improved (1?, 3?, 5? HP). I personally think it might be quite challenging to decode 1-3 HP changes via seat-of-the-pants experience with 2 good running bikes. As the various approaches continue, they might even exceed the airflow or HP achieved via a stock RH10 head? But how will we know when we’ve hit that milestone?
I think it would be a really cool aspect of the entire undertaking if it could all be tied together with quantitative information, i.e., see and understand all the pieces of the puzzle, where each individual piece is congruent with all other pieces. I suspect in such an undertaking that the probability of being surprised by some aspect of it - surprised good or surprised bad, but nevertheless surprised - would likely occur, which is the rewarding part of iterative experimentation and learning.
It’s easy for me to suggest quantifying the subject information, but it’s a serious, time consuming, and spendy undertaking to generate airflow and dyno results for each variant. But as progress continues here, at some point performance information has to be quantified if improvements are to continue. One much less expensive approach might be to use the “real” road like Worntorn does with dyno hill, but we don’t all live in the mountains. Perhaps some onboard electronic data acquisition device could identify performance differences?