AW_hub wrote:Brucey, excellent description and photographs of the SRAM Automatix.
I have a few comments:
Sealing against winter weather was never one of SRAM (nee Sachs) strong points for hub gears. You'd think it never rained in Schweinfurt.
I have dismantled several Sachs T3 3 speeds ruined by water ingress (balls rusted, cage corroded and broken up.
Never seen that extent in the many Sturmey Archer AW's I have dismantled - so I have to disagree that the Automatix sealing is no worse than the AW. The AW uses a very ingenious double labyrinth seal on all 3 ball bearings, they knew about rain in 1936..
Sachs/SRAM on the other hand always leave the bearing cone exposed - you can even see the balls if you look closely from the outside. I dismantled an Automatix on the weekend after reading your review, it was the freewheel version. I was impressed with the overall design, and having your review helped me understand its operation. But it was very disappointing on the weather sealing - people buy a hub gear over a derailleur for weather-proof operation, amongst other things.
Older AWs have two shields for the ball-ring, and as you say labyrinths on the cones. This arrangement is not 100% waterproof! It functions OK with oil lubrication in part because the oil film heals itself as soon as the hub is used, older type oils also decompose over time to give a 'varnish' that helps to protect the surfaces, and any water that gets into the bearings is expelled as the oil (inevitably) comes out in use. If such hubs are not used regularly and still see rain, water usually penetrates the ball-ring and causes corrosion. I am experimenting with an upgraded seal for this part, which includes a contact seal. Modern AWs have a plastic retainer in the ball-ring, a single shield, and one of the lips on the ball-ring has been deleted. All of this significantly degrades the sealing vs the older design. Despite often not having a labyrinth on the main cones (just a close-fitting dustcap) SRAM hubs are also most vulnerable at the ball-ring, as you say; modern ones should be fitted with a shield at this point, behind the sprocket, similar to SA hubs, although it is often missing, presumably either overlooked or removed following damage in service. The net result is that a modern (greased) AW has ball-ring sealing that is comparable to modern SRAM hubs. Even the SA labyrinth seals don't work as well as they used to; the use of grease rather than oil means that the seal surfaces can lose their coating of oil/grease and it doesn't heal itself (although chain lube overspill might help). Once the surfaces in the labyrinth are no longer hydrophobic (as they usually are because of the oil coating), water gets in more easily.
I agree that the sealing on many modern IGHs leaves plenty to be desired, and that the mess and routine topping up required with a total loss hub might well be a more acceptable compromise in the long term.
I am experimenting with the use of a generous fill of a thixotropic semi-fluid grease in some SA hubs at present. This requires that the ball-ring seal is upgraded to retain the lubricant in use, and on brake hubs, the left side seal should be upgraded to a full contact seal also to prevent brake contamination. On non-brake hubs it isn't necessary to revise the labyrinth seals on the main comes because there is some 'total loss' of the SF grease. Contamination of the SF grease is less of a worry than with standard grease; for a start there is more of it (the way I'm doing it) so there is more dilution of any crud and in addition there is still some total loss to remove it. So far the mess from the total loss is comparable with that from the chain lube anyway, so it is looking good to date.
I agree the lube inside was sparse. There is little doubt it would benefit from having an oiler tapped into the hub shell, say a nylon Sturmey Archer AW one (readily available). This would have the added benefit of flushing metal swarf and dirt out from the hub over time, and keeping water out despite the poor sealing. Thin SAE 20 oil of course.
The question is - where along the hub? I have an old Sachs Duomatic from 1968 with its OEM hub oiler that should give some clues.
Possibly nowhere; you may find that in hubshells that are hardened locally for ballraces -and I think the SRAM may be one such- the hubshell is hardened all over, and cannot be drilled easily. Do let us know how you get on if you try this.
Dyno studies done near Seattle in 2001 by Frank Berto showed oil lubed Sturmey Archer AW's and Sachs T3's were the most efficient epicyclics, comparable to deraileurs. The grease-lubed Shimano 4 speed hub tested was terrible.
IIRC others have questioned the methodology of these tests, although I thought -with a few caveats- that it looked OK for what it was. The three-speed gears were, unlike all the other gears tested, very well-used. I think the doubters suspect that in these tests the average torque was much lower than the peak torque seen in real-world use; this tends to favour gears with low LIL (load independent losses) rather than gears with appreciable LIL even if the LDL (load dependent losses) are possibly not as good in the low LIL gears. Grease-lubricated hubs, and hubs with contact seals, drag springs etc have higher LIL, and a four-speed shimano hub also has a double gear train for poor LDL in some gears. If it wasn't run in either, the net result was predictable. Note that Rohloff tests gear efficiency at a high torque, comparable to peak torque when pedalling. (Although this aspect of their testing is more representative, I can cheerfully pick holes in their test data in other ways).
As torque increases, at some point the use of low viscosity oil will allow metal-to-metal contact in bushings and the LDL figures will probably creep up. That there is some contact is obvious; the parts wear and continue to wear for the life of the hub. I believe that the use of slightly thicker lubricants with solid lubricant and extreme pressure additives (proven to reduce wear in highly loaded gear trains) may also reduce LDL at high torque values, although I have no means of demonstrating this.
BTW one obvious point is that IGH's with a 'normal' direct drive gear are intrinsically efficient overall, even if the other gears are not efficient per se; this is because most people ride mostly in the 'normal' gear. In extremis, a (say) ridiculous 25% loss in a gear you only use 5% of the time would only contribute a 1.25% loss overall, for example. The SA three-speed became popular in the first place in part because of this simple fact, and I still set my gearing on my SA-five speeds to exploit this, even though arguably there is a very high top gear and another lower gear might be more useful. Both current SA and SRAM two-speed gears, as well as Shimano 4-speed etc are 'gear-up' hubs and ignore this possibility; they leave gear trains in operation to sap more power than is necessary in 'normal' gears (assuming that the gear range is set to favour climbing gears over 'higher than normal' gears).
The Rohloff at least has the direct drive gear set where it will do most good for most people.