My 22T ENO's sealed bearing unit failed for the second time recently after only a few hundred klms, but this was in a freewheel crank setup that sees power transfer in a freewheeling state. Unfortunately the ENO's dont use a radial bearing on both sides of the ratchet/pawls, there is only one bearing taking up half the width of the casing, with the ratchet/pawls taking up the other.
In my case, the problem is that the driven and driving chainrings that are attached to the freewheel via an adapter plate are both alligned slightly laterally offset from the position of the bearing unit within the freewheel casing. This causes angular loading on the bearing which is really only designed for perpendicular/radial loading. This was a potential problem i envisaged when designing my custom freewheel cranks and it has come to light in practice. I got a few hundred Klm's off the first bearing and roughly the same out of the second one which i replaced recently. Dont get me wrong, in a locked state the ENO's are tough as nails but they are not an ideal option for freewheel cranks that see freewheeling power transfer as a jackshaft when the rider is not complimenting the motors power - which locks the pawls for better support of angular loading. For mounting ENOs on freehubs where the bearing only has to support the weight of the sprocket and chain whilst freewheeling you should not have a problem and they should be near indestructable.
The Ironic thing is that the design of cheaper freewheels is better for freewheeling cranks. They have two sets of loose ball bearings riding in case hardened races integral to the inner and outer sections of the freewheel, each on either side of centralised ratchet/pawls. This is a much better setup for angular loading but the problem is that being unsealed, they are more prone to getting dirt in them, they're not rebuildable, and the pawl count/quality/strength is lower than an ENO. What ever sort of one-way bearing or freewheel used with freewheel cranks, it has to support several hundred Nm of torque from the worst case abuse of a rider jumping/landing with their full weight on one crank arm. The ratchet and pawls within ENO's and other quality trials freewheels are much stronger in supporting this abuse which is above and beyond most freewheels design requirements which are only supporting the torque of human power at the rear wheel of a bicycle which is reduced significantly by the gearing up between chainring and freewheel sprocket tooth count at the rear wheel.
ENO outer casing showing ratchet and bearing retaining sections.
I know that Sickbikeparts sell their patented ENO 5-bolt flange freewheel as a "Heavy-Duty" upgrade option for freewheel cranks, but they have the same bearing/ratchet arrangement as any other ENO. I haven't heard of any other premature bearing failures reported on this forum from their intended use on freewheel cranks. I can only assume that most users are able to more optimally/centrally allign the driven and driving chainrings mounted to them, or are simply passing lower motor power through them in a freewheeling state. In my case i had dropout clearance issues so offset mounted chainrings were unavoidable.
Oh and being that ENO's only contain one deep groove radial bearing, I have found that the runout is significant, even with a fresh bearing. They have far more slop than a cheap dual bearing-race freewheel. I can wobble my 59T chainrings side to side by ~3mm at the radius when mounted on my ENO with a fresh bearing.