Classic Sturmey Archer (pre NIG) AW 3s hub; what to look out for, what breaks.

For discussions about bikes and equipment.
Brucey
Posts: 37282
Joined: 4 Jan 2012, 6:25pm

Classic Sturmey Archer (pre NIG) AW 3s hub; what to look out for, what breaks.

Postby Brucey » 25 Mar 2020, 4:12pm

The A-series hubs appeared in the 1930s, and in actual fact the AW was not the first A-series hub to be mass produced, first appearing in ~1937. However it did go on to become the most common SA A-series gear and is arguably the most common hub gear ever made.

It is also one of the most reliable gears ever made; the Sheldon Brown website describes it as 'the most reliable gear changing mechanism ever' and I would not dispute that assertion. It is also one of the most efficient; measurements suggest that it is just a couple of percent worse than a singlespeed drive and (on average) better than many derailleur systems. The efficiency stems from the use of labyrinth seals, and the clever design inside the hub for the sliding clutch; the clutch spring pressure is borne by a carrier, rather than the clutch itself, so there is no parasitic drag as appears in many other hub gears. Pretty much the only 'easy gain' to be had inside this hub design would be to use rolling element bearings on the planet pinions, which isn't done for cost reasons.

Exact production figures are unknown to me but I am given to understand that they made several million a year, right up until year 2000. Since then the model name 'AW' lives on, but the mechanism inside the hubshell is now different, being a so-called No-Intermediate-Gear (or No-Inbetween-Gear) NIG type. The NIG 3s gear design in essence dates from the late 1980s and was used in various other SA 3s hubs but it was not used in the AW model hub until after 2000.

The AW hub could also be either produced under licence (or was assembled locally using Nottingham-made internal parts) and therefore appeared with a variety of names on the shell, including Sears, Sun Tour, Styria, Scintilla, amongst others.

Over the long production run there were many detailed changes to the design and manufacturing methods used to make the internal parts. Yet you can take a 1937 internal and swap parts with a year 2000 internal, and they will almost all interchange without difficulty. Even today a pre-NIG internal will fit into a NIG AW shell and vice versa.

Unbelievably the AW internal was also designed to be backwardly compatible; the preceding 'K' series hub used the same hubshell, and would accept the AW internal provided the left dog ring (which is screw threaded into the shell in Ks and early AWs) is also changed. Today, if you find what looks like an old 'K' series hub on a bike that is still in regular use, it is odds-on that an AW internal sits within that shell.

Image
the left end of the hubshell is shown screw-threaded in position; this makes the illustration accurate for AW hubs made until some time in the mid 1950s

The AW design also successfully fended off would-be usurpers; first the SW (in the 1950s) and then later the NIG gear, which was available for at least a decade whilst pre-NIG AW production was still in full swing. The NIG AW is a pretty good hub but in practice isn't as reliable as the pre-NIG AW design. Despite this, when SA production resumed in Taiwan, they chose to finally discontinue the pre-NIG AW design, presumably because of the neutral that exists between gear 2 and gear 3.

The 2-3 neutral is at first sight an obvious flaw in the design of the AW hub. However it is not anywhere near as dangerous as a neutral between 1 and 2 might be, and effectively the gear does have NIG functionality between 1 and 2. This means that even if the gear is very badly adjusted, you are not going to lose drive as you (say) pull out into the road, provided first gear is selected. Arguably the 2-3 neutral is in fact the saving grace of the AW gear; should the gear cable go out of adjustment the first thing that happens is almost invariably that the hub slips in gear 2, under load. This isn't that likely to cause immediate damage (or be a major hazard in most cases), yet it does usually penetrate the consciousness of even the least alert rider to fact that 'there is something wrong' and they normally get on and have someone fix it.

There isn't a hub gear made that can't be damaged by being ridden on whilst out of adjustment; the differences lie in how likely that damage is to occur under said adverse conditions, how expensive the hub/repairs might be, and indeed how likely that fault condition is to exist in the first place. The AW scores well on all counts, since it takes rare blind obstinance to carry on riding a gear that is slipping, more to damage it, and in the event that the gear is damaged, it isn't difficult or expensive to repair.

At the heart of the mechanism lies a planetary gear train, with sun and planet pinions having 20T, in which the ring gear (60T) always turns at 4/3 the speed of the planet cage. The hubshell can be driven by the planet cage or by the ring gear; thus by driving the planet cage or the ring gear from the pedals, but choosing where the output is taken, you get three gears ratios, 3/4, 1.0, 4/3. The only part that moves during a gearshift is a cruciform sliding clutch, which is permanently meshed with the 'driver' to which the sprocket is fixed.

In gear 1 the ring gear is driven by the clutch, and the drive is taken from the planet cage to the hubshell via the low gear pawls. The high gear pawls (which are built into the ring gear) are unable to drive because the clutch interferes with the back of the high gear pawls, thus inhibiting them; they cannot spring outwards and transfer the drive to the hubshell.

In gear 2 the ring gear is again driven by the clutch, but now the high gear pawls are no longer inhibited, so transfer the drive to the hubshell. The hubshell is turning faster than the planet cage so the low gear pawls are overrun, and usually make a 'tick-tick' sound as you ride.

In gear 3 the clutch drives the planet cage and the (faster turning) ring gear drives the hubshell, much as it does in gear 2, only now you have a 4/3 increase. With the control cable slack, the gear defaults to gear 3.

Between gear 2 and gear 3 the clutch doesn't engage with anything; if it were part-engaged with both the ring gear and the planet cage, there could be extremely high forces generated, and damage or breakage would be possible, even with a ramped clutch.

See the next part for more information about design variations in the AW gear.
Last edited by Brucey on 25 Mar 2020, 4:30pm, edited 5 times in total.
-------------------------------------------------------------------------------
~~~~~~~~~~~~~~~~~~~~~~Brucey~~~~~~~~~~~~~~~~~~~~~~~~

Brucey
Posts: 37282
Joined: 4 Jan 2012, 6:25pm

Re: Classic Sturmey Archer (pre NIG) AW 3s hub; what to look out for, what breaks.

Postby Brucey » 25 Mar 2020, 4:12pm

Design variants; nearly all AW hubs are clearly marked with the date of manufacture. In most cases if you find an AW hub in an old bike, it will still have the original internal inside the shell. However it isn't guaranteed to be the case, so it can be helpful to be able to identify the internal parts as well as the hubshell. Below is a (probably incomplete) list of known design changes in the parts, with approximate dates where possible.

The hubshell; initially it was designed with a LH threaded insert in the LH side, a feature that harked back decades to even older designs. You can tell these hubs because they have painted (rather than plated) LH end cap to the hub and furthermore that cap has two large flats on it so that the LH end cap can be replaced (eg if the LH bearing cup or dog ring needs to be replaced). Later hubs have a splined LH end cap, that is a push-fit into the hubshell. The revised arrangement gives very little trouble, since the LH dog ring and bearing cup are the rarely damaged in service. The change in design occurred some time in the 1950s.

The hubshell. AW hubshells come in several different styles, which don't affect the fit of any AW internal (including the current NIG type) into any hubshell. However they do affect the possibility of fitting other internals. In the drawing in the post above, the LH end cap has a 'scalloped upstand' in it, that protrudes into the hubshell, but is itself internally scalloped. This will allow fitment of an FW internal or an S5, but not some later internals. In the late 1970s and the 1980s AW hubshells often had 'solid upstands' and thus will only accept AW internals, or others which have an unobstructed interior to the planet cage. This persisted until some time in the noughties; current AW shells and SRF3 shells have practically no upstand.

The hubshell. Initially 40h was the most common option, with 36h for export markets. However with the rise in popularity of the Moulton, SA deigned to make 28h hubshells. Hence the first moultons had 36 spoke wheels, but after a year or two they got 28 spoke ones instead. 40h hubshells disappeared some time in the 1960s leaving 36h and 28h options, which is pretty much what you can still buy now (although I perhaps have not seen 28h steel from Taiwan, only 28h alloy). If you want to spoke a SA AW hub with a 32h rim, this is most easily accomplished with a 40h shell; two spoke lengths are required.

Low gear pawls. Originally these were made symmetric (i.e. with the pivot central) and could be installed two (although it appears to be four, it isn't) ways round, but at some time (exact date unknown) this design was changed so that the pawl has a short side and a long side. This ought to give more positive engagement. There may also be changes in the tip radius of these pawls.

Ring gear. In early AW hubs they are an 'interrupted' pattern i.e. there were separate, staggered, drive dogs internally which the sliding clutch engaged with in first and second gears. At some point (before 1950 I think) the design was changed to four continuous internal ribs (splines). This gives a slightly draggier shift between 2 and 1 but it also makes it nigh-on impossible for the gear to slip when '1' is selected, even if the adjustment is dreadful.

Ring gear. The four ribs inside the ring gear which the clutch engages with are not as they appear to be. They appear to be regularly (evenly) spaced, at exactly 90 degrees, but in reality they are usually not positioned evenly; they are in two opposite pairs set at angle intervals of slightly over and slightly under 90 degrees. This means the cruciform clutch is never driving on all four fingers, only ever two. I think that this arrangement may be to reduce wear between the clutch tips and the ribs; certain forms of non-concentricity are better tolerated by two-finger drive.

Sliding clutch and spring carrier. These parts have received several minor changes and ideally they should be used as a matched set. In particular the carrier sleeve ought to be a match for the clutch itself; the flange on the carrier can be of different widths and likewise the recess in the back of the clutch. The earliest illustrations of AW gears show a ramped clutch but I've never seen one of these in the flesh. The clutch was machined from (overthickness) solid to start with but this was changed to a much simpler/cheaper component, pressed from sheet steel. There isn't much to choose between them in how they work, but on balance I prefer the spring carrier to be of the sort that inhibits the end of the spring splaying out.

The clutch spring; This appears to exist in LH and RH wound forms. The current NIG AW uses a RH wound spring, but I don't think it makes any real difference in a pre-NIG AW.

Spring cap. There is a small cap over the RH end of the clutch spring. The spring (unlike in a NIG-AW) shouldn't ever rotate in service; the only time it sees a twisting load is when the RH cone is tightened. However the cap needs to be intact to prevent the end of the spring from splaying; it doesn't have to splay far before it is on the sloping part of the RH cone and this soon mangles the RH hub bearing. Some steel spring caps seem not to want to fit between the balls in the RH bearing when being fitted from right to left, which is rather irritating when reassembling a hub. Occasionally a plastic spring cap will crack, so check for this when servicing a hub.

Cones. These have a pressed on labyrinth seal part, which through manufacture (or wear) often ends up with a razor-sharp edge. I have lost count of the number of times I have lacerated my fingers when screwing the RH cone down against the clutch spring.

The ball-ring and dust shield. The original design for this part uses a lipped bearing race and a pressed-on metal cap which retains ~24 loose 3/16" balls. This version of the part also has an external upstand so that there is a full labyrinth between the dust cover and the ball-ring. In the 1970s this was redesigned so that the outer lip was deleted, and the steel dust cap was also redesigned with a different lip shape. This ball ring part was in turn redesigned (in the 1980s I think) so that the bearing cup isn't lipped and there are fewer 3/16" balls which are set into an acetal retainer. There is an additional groove in the OD of this ball ring which retains a plastic dust shield. You can use either type of pressed steel dust shield with the second type of ball ring but (IIRC) only the second type of steel dust shield can be used to replace the plastic dust shield on later hubs. You need to use a steel dust shield if you want to fit two 1/8" thickness sprockets to an AW driver, for which there is enough length available (in contrast to the NIG-AW design which has shorter driver splines).

The dog rings. Originally they (both left and right) were a 10T design, which gave lots of backlash but very consistent two-pawl drive. In the 1980s they were redesigned to 20T type, which gives less backlash but also increases the chances of one-pawl drive (bad). This change is also accompanied by a shape change in the dogs themselves. I think this means that it may be bad to use older high gear drive pawls in a 20T dog ring, i.e. the pawls which best match the newer dog rings ought to have a certain tip radius, whereas older pawls had something closer to a knife-edge profile.

The high gear drive pawls. These vary between types for 20T and 10T dog rings. There are also changes in the profile of the backside of the pawl (which affects how the high gear pawls are inhibited in gear 1). At the least you should use these pawls in matched pairs (of identical length) and ideally you should only use early pawls in 10T dog rings.

Driver. Tolerances on the spacing of the four fingers of the driver vary, such that it is possible to get a quite sloppy fit between the clutch and the driver. Even so this doesn't seem to greatly affect the working of the gear.

Driver. Early drivers were screw threaded, but at some point (around 1950) the familiar three-lug splined design appeared. It remained unchanged thenceforth. The same sprocket fitting was used by SRAM and is still used by shimano in many of their IGHs; it is ( I believe) the subject of an ISO standard. The screw threaded driver allows easy fitment of a (narrow) multiple freewheel, but requires that the driver is removed from the hub whenever the sprockets are changed.

Driver. Early drivers had a straight taper on the four fingers. This was later changed to a stepped taper.

Locknuts. These were 15mm AF for many years and then changed to 17mm at some point (late 1980s?). This means better loading into the NTWs/dropouts, but 17mm is a less common cone spanner size so this change is arguably slightly irritating; SA cones use a 16mm spanner and if a suitable tabbed washer were used on the LH side with a 16mm locknut, you would only need one cone spanner to make a bearing adjustment.

Lubricator. There are at least two types of metal lubricator. This was replaced by the plastic type in about 1960 (or slightly earlier) before the lubricator was deleted altogether in about 1988/89, which coincides with the adoption of grease lubrication instead of oil ( so old fashioned, doncha know....). You can add oil (or SFG) to any hub via the hollow axle, having removed the indicator rod.

Many internal parts. Have been changed from 'machined from solid' to 'made via powder metallurgy'. This includes parts like planet pinions. Very occasionally there are signs that the powder met processing (or hardening perhaps) have gone wrong but these are few and far between.

Many internal parts are slightly different weights. Examples include hubshells, ball rings, ring gears and planet cages. By selectively choosing these parts (from hubs of different vintages) you can end up with a 'hot rod' 3s hub which is significantly lighter in weight than an average AW. The current (protrusionless) SRF3 hubshell is the lightest type (apart from some 1950s hubshells which have a thin-walled aluminium shell; however these hubshells are not very strong so I don't recommend their regular use). It is also possible to lighten many of the parts still further; eventually some parts can be made so weak they may fail, but there is plenty to come off I reckon; this is a topic of ongoing examination.

Axle. For the entire duration of pre-NIG production the axle was one with a riveted sun pinion. However at present you can buy the axle which is more like the one you really wanted; SA in Taiwan make an inherently stronger 'all solid' axle to fit the obsolete pre-NIG AW. Asides from a question mark over the hardening on the sun pinion, this is a significant improvement. There are two lengths available, ~148mm and ~163mm.

more to follow
Last edited by Brucey on 25 Mar 2020, 9:43pm, edited 11 times in total.
-------------------------------------------------------------------------------
~~~~~~~~~~~~~~~~~~~~~~Brucey~~~~~~~~~~~~~~~~~~~~~~~~

Brucey
Posts: 37282
Joined: 4 Jan 2012, 6:25pm

Re: Classic Sturmey Archer (pre NIG) AW 3s hub; what to look out for, what breaks.

Postby Brucey » 25 Mar 2020, 4:13pm

What breaks; the short answer to this is 'not much, not often'.

IME it takes sustained abuse/neglect to break one of these hubs. However even if well-maintained, eventually the axle will probably break but most of these hubs will do half a lifetime's cycling before that is likely to happen. You can now get a stronger 'all solid' axle to replace the original type with its riveted sun pinion.

If all you do is oil it occasionally, keep it well adjusted, and ride your bike normally the hub may in fact last a lifetime.

Image
20T dog rings and a steel dust cover over the ring bearing means this is probably a late 1980s hub

Some things to look out for include

- that the track nuts are (deliberately) soft; if you overtighten them and/or don't use grease on them, the threads can be stripped. However this is almost invariably not accompanied by axle damage; the track nuts are intended to be consumables.

- that the indicator rod threading is not a slack fit in the toggle key; best to measure the parts and make sure that you have at least 0.2mm engagement on the thread diameter.

- that the gear doesn't slip too hard, too often, in second gear. It will only start to do this if the adjustment is bad, but such slippage does yank the gear cable very hard indeed, which can cause damage to the cable/control rod

- that the axle locking washers (NTWs in modern parlance) fit the dropout width properly. Ideally the torque load on the axle is shared between both sides, and this will only happen if the NTWs are in good shape and fit the dropouts properly.

- that (like a NIG hub) it is best you get in the habit of pedalling without full force for a couple of turns of the pedals after you shift from first into second gear or resume pedalling. The risks of one-pawl- drive are much less (eg with 10T dog rings) but are still present.

- when reassembling the hub be sure to start the ball ring on whichever thread start allows it to screw on furthest.

- when refitting a sprocket be sure that you use an undeformed snap ring and that it is well seated. To seat the snap ring, it should be tapped down using a small drift.

- that you use the correct shifter with this hub; practically speaking this means a trigger shifter is the best choice. Most later/current SA 3s shifters may pull too much cable even when new (which only gets to be more as time goes on with some shifters eg the aluminium thumbshifter). IME using a later model 3s shifter with an older hub gives less security in second gear.

- NB you can replace the hub bearings with loose balls instead of clipped ones. This ought to be 9/7 as strong, and furthermore it avoids the annoyance of removing (and usually damaging) the labyrinth seal; the old clip can be destructively removed, leaving the labyrinth seal undisturbed. Using loose balls does require more care when assembling the hub though.

- it is (unlike the NIG hub) usually not possible to remove the toggle key (with the ball ring in situ) and therefore also not possible to move the axle through the planet cage or remove the sliding clutch in a pre-NIG AW. However some strategic grinding of the holes in the clutch carrier and the pawl pockets in the ring gear can make this possible. If you plan to use this hub in situations where better access to these parts might be beneficial, such a modification may be worthwhile.

- if you re-use a steel shell, it is best to build the wheel with spoke washers, and to follow the old spoking pattern in the hubshell exactly. This will give the spoke elbows the easiest time of it.

3s gears like the pre NIG AW are not for everyone; they have widely spaced ratios, lots of backlash and some folk feel they are too compromised to use. However people get on well with a singlespeed, and favour its virtues of simplicity and reliability; a 3s hub is perhaps better viewed as a 'singlespeed with choices' (that you can get used to) rather than a competitor to a modern derailleur gear in terms of gear ratios. Choosing the middle gear on a 3s setup is something that requires at least as much attention as in a singlespeed. The simplicity and reliability of a SA 3s gear are not much different from that of a singlespeed, yet it is usually a considerably easier and faster ride with a 3s gear than a single freewheel.

Adding a 3s gear to a singlespeed machine 'costs' about 1-1/2 lbs or so; hardly a big deal. BITD it was reckoned that uphill and downdale, the average touring speed with a 3s gear was about 2mph faster than with a singlespeed. This may well be true of commuting with 3s gear vs a singlespeed today (for the same effort).

cheer
-------------------------------------------------------------------------------
~~~~~~~~~~~~~~~~~~~~~~Brucey~~~~~~~~~~~~~~~~~~~~~~~~

Brucey
Posts: 37282
Joined: 4 Jan 2012, 6:25pm

Re: Classic Sturmey Archer (pre NIG) AW 3s hub; what to look out for, what breaks.

Postby Brucey » 26 Mar 2020, 5:32pm

regarding part selection for a lighter weight hub, I have weighed some parts from my stash of spares;

Image01352.jpg
pre-NIG planet cages

the weight variation in the above planet cages is ~29g

Image01353.jpg
pre-NIG ring gears

the weight variation in the above ring gears is ~35g

So 64g between these two parts alone.

Image01365.jpg
hubhells

There is a significant weight variation in hubshells too, from memory in the region of 75g.

So about 140g is up for grabs just with part selection on these three parts alone.

With some, uh, 'attention' from a grinder, I don't doubt that further weight reductions are possible.

cheers
-------------------------------------------------------------------------------
~~~~~~~~~~~~~~~~~~~~~~Brucey~~~~~~~~~~~~~~~~~~~~~~~~

Sid Aluminium
Posts: 93
Joined: 26 Feb 2019, 7:38pm
Location: Beyond the edge of the wild

Re: Classic Sturmey Archer (pre NIG) AW 3s hub; what to look out for, what breaks.

Postby Sid Aluminium » 27 Mar 2020, 3:12am

Oh, chapeau! Brilliant.

Sid Aluminium
Posts: 93
Joined: 26 Feb 2019, 7:38pm
Location: Beyond the edge of the wild

Re: Classic Sturmey Archer (pre NIG) AW 3s hub; what to look out for, what breaks.

Postby Sid Aluminium » 27 Mar 2020, 8:08pm

I knew about the Rollo Elf lathe which used a Sturmey-Archer AW for a multi-speed gearbox, but last week whilst educating myself on medical ventilators, I ran across this surprising entry:

In the United Kingdom, the East-Radcliffe was an early example. It used a Sturmey-Archer bicycle hub gear to provide a range of speeds.

Screen Shot 2020-03-27 at 2.55.24 PM.png
Screen Shot 2020-03-27 at 2.55.24 PM.png (117.55 KiB) Viewed 99 times


The East-Radcliffe apparently used an FW:

For the past seven years Radcliffe ventilators have been widely used in the treatment of patients suffering from ventilatory inadequacy. The original positive pressure pump (Russell and Schuster, 1953; Mushin, Rendell-Baker and Thompson, 1959) consisted of an Oxford Inflating Bellows (Macintosh, 1953) which was compressed intermittently by a falling weight. During the expiratory period the weight was lifted by an arm attached to the hub of a Sturmey-Archer four-speed gearbox. This, in turn, was driven by an electric motor. The inspiratory pressure could be altered by varying the number of weights and the frequency of respiration by manipulating the gear change or by altering the sprocket ratio on the chain drive.

Hoodibaba!

Brucey
Posts: 37282
Joined: 4 Jan 2012, 6:25pm

Re: Classic Sturmey Archer (pre NIG) AW 3s hub; what to look out for, what breaks.

Postby Brucey » 27 Mar 2020, 9:16pm

that is interesting; where's the pedals....?.... :wink:

A couple of days ago I found a rare thing; a worn sun pinion in a pre-NIG AW.
Image01627.jpg
a rare thing; a worn axle from a pre-NIG AW, compared with an unworn example

This is only about the third one I have ever seen, and the cause was not hard to ascertain; the hub in question had been run for a long time without the benefit of any oil in it; much as you might see in a fretting wear condition, the wear particles had been produced as a 'dust of rust' which had liberally coated everything inside the hub. The sun pinions were also worn internally (where they run on the pinion pins) but (similarly to the worn NIG AW sun pinion I photographed in another thread) the teeth on the planet gears were not badly worn.

cheers
-------------------------------------------------------------------------------
~~~~~~~~~~~~~~~~~~~~~~Brucey~~~~~~~~~~~~~~~~~~~~~~~~