Right, here goes....
Despite having previously heralded the ACS Crossfire model of freewheel being a suitable candidate for duties in FW crank implementation, my recent discovery leads me to unfortunately revise that opinion to labelling them as unsuitable if requiring acceptable service life.
I today opened up one that had been in FW crank service for ~1500km. It was supposed to be to just routine maintenence to replace the bearing grease and check for bearing wear. I was disheartened to find that the ratchet teeth are badly indented and are nearing failure in one indexing position. I expected better of this FW model. The dimensional accuracy, fit & finish and metallurgy are clearly improved over the Dicta type 'cheapies', but obviously still not up to the torque demands of my modest 70kg (154lbs) mass on 170mm crank arms (~130Nm).
I'm concerned that their many not be ANY BMX/SS freewheel with dual angular contact bearings that provides enough torque capability that is required for FW crank duties when coupled with large chainrings. The
'Excess Pro' FW is a recently released model that boasts an 'industry leading' torque-hold rating of '550 LB/in' (62Nm) which is still far short of the required 130Nm for my body weight and crank-arm length, even disregarding shock loading. Unfortunately I can't find a published torque rating for the ACS Crossfire (or the single bearing White Industries ENO for that matter) to compare this to, but if the Excess Pro is above average then its still only half the strength of what I require, and this does not inspire hope of any others being up to the task.
Now you may be thinking, why do Trials bikes have less problems with FW crank related failures? Afterall, FW cranks for e-bike use are an adaptation of their successful use on Trials bikes.
I suspect the reason is that because Trials bikes with FW cranks are geared so low,the torque holding they are subjected to is far lower than in E-bikes with larger chainrings and therefore taller gearing. Because Trials bikes are geared to accelerate so rapidly, the rider can't apply as much loading on the FW in comparison with mid-drive e-bike riders who with wide ranging variable gearing are able to accidentally or deliberately pedal in a relatively tall gearing ratio for the incidental traveling speed. This allows far more FW torque loading than in Trials bike configurations.
If there is a Freewheel model out there that is potentially up to the long term torque holding durability that I require, it's likely to be the White Industries ENO. As far as I'm aware, no-one has reported undue ratchet teeth deformation when used in FW crank duties, however that could be because the run-out issues related to the single bearing in over-running operation lead to their replacement of a dual-bearing unit before significant service life elapses.
Unfortunately the need for a second support bearing requires quite involved fabrication and so I will reluctantly follow Miles'/Fullthrottle's example of how this is done.