Shimano Nexus 7; how it works. (long)
Posted: 23 Oct 2013, 3:18pm
Shimano Nexus 7; how it works.
I've not often had good cause to strip one of these hubs down fully; they seem to posess an unusual combination of good reliablity and yet alarmingly high internal complexity.
The few times I have looked inside these hubs in the past I was not been able to see how they work, exactly. Part of the reason for this is that (unlike the Nexus 8 ) the way the sun locking pawls are affected by the shift control sleeve cannot be seen directly; like the Nexus 4, the sun locking pawls are hidden inside each sun pinion, and are sometimes able to bear against shoulders on the axle shift sleeve, and sometimes not.
When the gear centre is assembled, it isn't easy to see what is locked and what is moving; to cap it all the parts that do move can also vary with load, so turning a bare centre may give a false impression. Similarly turning a centre backwards does not do the same thing as turning it forwards. All in all, it is a bit tricky.
After some hours of examination I now think I have worked it out. Whilst the details are not necessarily of general interest, the basic layout of the gear has implications for the way it operates so should be of interest to anyone who is thinking of using this kind of gear.
---------------------------------------
The headlines;
The gear is essentially a 3x3 gear, i.e. would be capable of 9 speeds with small revisions, but with two unused (near duplicate) ratios. There are two gear trains in series; the first is a reduction gear and the second a gear-up gear. Four of the seven ratios use just one gear train for reduction or increase; the other three ratios use both gear trains to achieve the ratio. This has implications for the gear's efficiency.
The gear ratios cover a 244% range (from 0.632 to 1.545) in intervals ranging from +13.8% to +17.3% with an average of +16.1%.
The gears are clutched by locking one or more of four sun pinions. All sun pinions are always free to rotate in one direction, and are clutched in the other direction. In addition to this a further clutch C1 is used in the first gear train.
The clutches are arranged so that the gear can be shifted under modest load without great risk of damage. In mitigation of any slippage that might occur, there are four pawls on most sun pinions, of which only two are used at any one time. In addition to this, in most cases a second ratio is also in mesh, with the clutch pawls freewheeling (hence the typical ticking from the gear). In the event of any gross slippage, the other gear should maintain some kind of drive.
The gear is supplied with (white or black) grease lubrication; this is OK for a year or so but isn't always reliable for longer periods; the grease can be displaced, can dry out, and will not prevent corrosion in the long term if there is any water ingress. It is also a little draggy. An annual shot of fresh lube is a very good idea, as is running the hub in a more mobile lubricant which will find its way where it needs to go.
Hub centres are very easily swapped for new ones, and (AFAIK) any Nexus 7 centre/dustcap assy ( there are about a dozen different versions) will fit into any Nexus 7 hubshell. In the event of a major internal fault a replacement centre is the easiest option; apart from driver assemblies, dustcaps and other external spare parts, the only individual spare parts that are available are cones etc. No other internal gear parts are available separately from whole centres.
Overall; although not perfect by any means, it is a very good gear for a utility bike. Its strengths are its relatively even gear spacing, its reliability and its tolerance to poor shifting. Its weaknesses are its long term weatherproofing/lubrication, its inefficiency in some gears, and the lack of individual repair parts.
------------------------------------------------------------
The Detail
------------------------------------------------------------
Gear efficiency.
As mentioned above, gear efficiency is affected by the number of gear trains in use at any one time; four of the gears use just one gear train, and the other three use two. Arguably the wrong three ratios use two gear trains; these gears are #3,4,5 which are often the most commonly-used gears; low efficiency in these gears will have a disproportionate effect on the overall average efficiency.
In round numbers the efficiency of any given gear stage is usually related to the ratio within that stage; thus a single stage gear with a 10% increase or decrease is likely to be subject to half the losses of a similar gear with a 20% ratio shift.
If we assume that there are 1% losses in each gear train per 10% ratio, then predicted efficiencies are given by multiplying the efficiency of the first gear train by that of the second. This gives predicted losses of 3.7%, 2.6%, 7%, 6%, 8%, 3.3%, 5.5% for gears 1-7 respectively.
The above predictions may be representative of a well run-in gear using a slippery lubricant, and may (assuming a conventional spread of gearing use) average about 7% (in addition to chain and other fixed losses which may be of the order of 3-5%). Thus total losses may be in the region of 10-12%.
However gearing losses may be nearer doubled in a gear that is poorly lubricated or not run in yet, with similar fixed losses. In this case average total losses of 17-19% might be expected.
In contrast measurements of (well-run in, well lubricated) simple single stage epicyclic gears with chain drives can show total losses of ~4-7%, with an average of nearer 4% if the direct drive gear is favoured. So other gears (e.g. most five speed and some other 7 speed gears) might well be noticably more efficient than a Nexus 7. It should however be noted that 10% losses don't usually mean a 10% speed reduction; about 5% reduction might be more likely at modest speeds.
-----------------------------------------------------------------
How the gear ratios are achieved:
The first (reduction) gear train comprises a single ring gear, with three dual-stepped planet pinions, each step having its own sun pinion. It has three ratios;
D = Direct (1.0)
L = Low (0.741)
B = Bottom (0.632)
The second (increase) gear train has similar parts, again with three ratios;
D = Direct (1.0)
H = High (1.335)
T = Top (1.545)
The clutching gives the ratios in combinations thus;
Gear 1 = Ratios BD = 0.632
Gear 2 = Ratios LD = 0.741
Gear 3 = Ratios BH = 0.843 = 0.632 x 1.335
Gear 4 = Ratios LH = 0.989 = 0.741 x 1.335
Gear 5 = Ratios LT = 1.145 = 0.741 x 1.545
Gear 6 = Ratios DH = 1.335
Gear 7 = Ratios DT = 1.545
unused near-duplicate ratios;
BT = 0.976
DD = 1.000
--------------------------------------------------------------------
How the gears are configured:
From right to left in the hub;
Gear train 1
Ring 1 = 72T
Sun 1 = 42T
Planet 1/2 = 14T/20T
Sun 2 = 36T
Gear train 2
Ring 3 = 66T
Sun 3 = 36T
Planet 3/4 = 14T/19T
Sun 4 = 30T
-------------------------------------------------------------------
Ratio calculation:
In each case the gear ratio is given by the formula;
Vr = Vpc (1 + (Ns x Npr)/(Nr x Nps))
for stepped planet pinions which simplifies to;
Vr = Vpc (1 + Ns/Nr)
for simple single step sun/planet/ring configurations, where;
Vr = ring gear speed
Vpc = planet cage speed
Ns = teeth on sun pinion
Nr = teeth on ring gear
Nps = teeth on planet in mesh with sun
Npr = teeth on planet in mesh with ring
So Vr/Vpc or Vpc/Vr gives the ratio depending on the direction of drive.
E.g. in gear 2 the ring gear is the input and the planet cage is the output, using ring 1/planet 1, and sun 2/planet 2 so the gear ratio is
Vpc/Vr = (1 + (36x14)/(72x20))^-1 = 0.741
------------------------------------------------------------------
Clutching configurations:
Each sun pinion can be locked in one direction and is permanently free in the other. An additional clutch C1 can transfer drive from the sprocket/driver Ring 1 to PC1. PC1 is permanently engaged with PC2. Drive is transferred to the hubshell from PC2 directly (gears 1 and 2) or from Ring 3 (all other gears).
Configurations are as follows:
Clutch.....C1....S1...S2...S3...S4
default...AFB...AFF...AFF..AFB..AFB
Gear1.....---...LBa...--...--...---...
Gear2.....---...LB....LBa..--...---...
Gear3.....---...LBa...--...--...LFa...
Gear4.....---...LB....LBa..--...LFa...
Gear5.....---...LB....LBa..LFa..LF....
Gear6.....LFa...LB....LB...---..LFa...
Gear7.....LFa...LB....LB...LFa..LF....
Where:
AFB = Always Free Backwards
AFF = Always Free Forwards
LB = Locked Backwards
LF = Locked Forwards
a = active clutch (i.e. .loaded, not overrun)
S1 can back up S2, and S2 can in turn back up C1.
S4 can back up S3, and the PC2 pawls can back up S4.
Backup clutches are overrun if engaged at the same time as another clutch in the same train.
Total loss of drive is theoretically possible but would require the failure of multiple pawl sets.
----------------------------------------------------------
Shifters, cable pulls and clutching movements:
Most shifts require a ~5.00mm cable pull and actuate one or two sun clutches. However the 6-5 shift requires clutch C1 to be moved as well as S3. This shift has a longer cable pull (almost 7mm), in part because of the C1 mechanism which requires a longer stroke and a higher force in one direction. Note that no single shift requires more than two clutches to be moved.
It will be noted that there is an unused DD ratio that would give a lossless 1.000 gear. Cynics have suggested that this is not used because it would stand out like a sore thumb vs the adjacent lossy ratios. This may be so, but it didn't stop them from putting the most and least efficient gear ratios adjacent to one another in the Nexus 8 gear.
It seems to me that it is just as likely that the clutching could have been made too complicated had they done this; both the 3-4 and the 4-5 shifts would have been C1 shifts also moving a total of two or three clutches.
The shifter cable pull pattern is unique to the Nexus 7 gear and there is no exact substitute shifter; although having said this, the S-clutches are not quite as sensitive to cable adjustment as in N8/A8 gears. It is said by some that a SRAM X-series gripshift for 7/8s derailleur gears ( 7 clicks of ~5.0mm cable pull) will operate an N7 gear, but it leaves the adjustment incorrect in at least two of the seven gears, which will make exact setup critical, and will doubtless lead to trouble down the line.
Shimano shifters include single-action, single lever pods, and various gripshifter types. The shift is reverse pattern, i.e. the hub defaults to gear 1 and the cable is pulled to access higher gears.
Sun Race make an N7 compatible shifter. In addition some Gazelle models come with a Gazelle-branded part that appears not to be a shimano shifter, but has the correct shift pattern for an N7 hub.
---------------------------------------------------
Common faults include;
Slippage damage; bad adjustment that leads to slippage will damage pawls; even if this doesn't by itself result in persistent further slippage, chips from damaged pawls will get to places they shouldn't and this will further damage the gear.
Water ingress - usually damages the ball ring first, and the clip can collapse; this sometimes necessitates a new hubshell.
RH cover damage- an unshipped chain can damage the cover. Replacements are available; be sure to get the right one for your internals (NB. a replacement internal may use a different cover from the original).
RH hub bearing damage; ball-clip can collapse, eventually leading to a failed driver and RH cone. Only a mobile lubricant will migrate internally to this bearing, so many forms of gear relubrication are not effective here.
Dried/inadequate grease; leads to water damage, slipping, and other general wear, e.g worn planet pinions.
Dirty cassette joint; leads to bad shifting and eventually other hub damage.
Internal debris; gear clearances are not large. Quite small pieces of debris (<0.5mm in size) can interfere with gear meshing and cause the gear to perform badly.
Shift sleeve damage; forcing the shifter on loaded upshifts can twist the shift sleeve, potentially making the S2 and S4 locking out of phase and/or overly sensitive to adjustment (typically gear 2-3-4 shifts affected). It is best to shift under light load or no load whilst pedalling forwards with this hub.
Assembly fault: longitudinal clearance in the gear assy (retained by a snap ring on the left end of the Centre) is often not large; comparatively small pieces of debris can lead to seizure.
Assembly fault: S2 and S3 form a subassembly where the parts are retained by an internal snap ring, but are free to rotate WRT to one another. The snap ring can be damaged during factory assembly, leading to a (worse in some gears, but OK in gear 5) fault (the suns won't rotate freely) and many other faults if pieces of damaged snap ring are floating around in the gear.
Assembly fault: if a Centre has slack longitudinal clearances, and the RH ball clip is in any way imperfect, individual balls from the RH bearing can come free when the centre is installed. This can lead to rapid and total gear failure.
Assembly fault: when the Centre is installed in the hubshell, it is very easy to dislodge one or more of the hubshell drive pawls on PC2 or R3. This leads to a variety of immediate faults. It is best to rotate the Centre backwards as it is inserted into the hubshell, and/or insert the R3 pawls into gaps in the hubshell dog ring.
Assembly fault; some 3/32" sprockets have offset drive lugs; it is a bad idea to install these with the lugs to the right; this can wear the driver splines, which are quite short.
Assembly fault: if the clip for the main ball-ring is in any way imperfect, individual balls can escape from it during centre installation. It is best to install the RH cover as a final operation, having adjusted the hub bearings and verified that the ball ring is fully populated.
---------------------------------------------------------------------
NB I may add pictures to this thread at some point.
cheers
I've not often had good cause to strip one of these hubs down fully; they seem to posess an unusual combination of good reliablity and yet alarmingly high internal complexity.
The few times I have looked inside these hubs in the past I was not been able to see how they work, exactly. Part of the reason for this is that (unlike the Nexus 8 ) the way the sun locking pawls are affected by the shift control sleeve cannot be seen directly; like the Nexus 4, the sun locking pawls are hidden inside each sun pinion, and are sometimes able to bear against shoulders on the axle shift sleeve, and sometimes not.
When the gear centre is assembled, it isn't easy to see what is locked and what is moving; to cap it all the parts that do move can also vary with load, so turning a bare centre may give a false impression. Similarly turning a centre backwards does not do the same thing as turning it forwards. All in all, it is a bit tricky.
After some hours of examination I now think I have worked it out. Whilst the details are not necessarily of general interest, the basic layout of the gear has implications for the way it operates so should be of interest to anyone who is thinking of using this kind of gear.
---------------------------------------
The headlines;
The gear is essentially a 3x3 gear, i.e. would be capable of 9 speeds with small revisions, but with two unused (near duplicate) ratios. There are two gear trains in series; the first is a reduction gear and the second a gear-up gear. Four of the seven ratios use just one gear train for reduction or increase; the other three ratios use both gear trains to achieve the ratio. This has implications for the gear's efficiency.
The gear ratios cover a 244% range (from 0.632 to 1.545) in intervals ranging from +13.8% to +17.3% with an average of +16.1%.
The gears are clutched by locking one or more of four sun pinions. All sun pinions are always free to rotate in one direction, and are clutched in the other direction. In addition to this a further clutch C1 is used in the first gear train.
The clutches are arranged so that the gear can be shifted under modest load without great risk of damage. In mitigation of any slippage that might occur, there are four pawls on most sun pinions, of which only two are used at any one time. In addition to this, in most cases a second ratio is also in mesh, with the clutch pawls freewheeling (hence the typical ticking from the gear). In the event of any gross slippage, the other gear should maintain some kind of drive.
The gear is supplied with (white or black) grease lubrication; this is OK for a year or so but isn't always reliable for longer periods; the grease can be displaced, can dry out, and will not prevent corrosion in the long term if there is any water ingress. It is also a little draggy. An annual shot of fresh lube is a very good idea, as is running the hub in a more mobile lubricant which will find its way where it needs to go.
Hub centres are very easily swapped for new ones, and (AFAIK) any Nexus 7 centre/dustcap assy ( there are about a dozen different versions) will fit into any Nexus 7 hubshell. In the event of a major internal fault a replacement centre is the easiest option; apart from driver assemblies, dustcaps and other external spare parts, the only individual spare parts that are available are cones etc. No other internal gear parts are available separately from whole centres.
Overall; although not perfect by any means, it is a very good gear for a utility bike. Its strengths are its relatively even gear spacing, its reliability and its tolerance to poor shifting. Its weaknesses are its long term weatherproofing/lubrication, its inefficiency in some gears, and the lack of individual repair parts.
------------------------------------------------------------
The Detail
------------------------------------------------------------
Gear efficiency.
As mentioned above, gear efficiency is affected by the number of gear trains in use at any one time; four of the gears use just one gear train, and the other three use two. Arguably the wrong three ratios use two gear trains; these gears are #3,4,5 which are often the most commonly-used gears; low efficiency in these gears will have a disproportionate effect on the overall average efficiency.
In round numbers the efficiency of any given gear stage is usually related to the ratio within that stage; thus a single stage gear with a 10% increase or decrease is likely to be subject to half the losses of a similar gear with a 20% ratio shift.
If we assume that there are 1% losses in each gear train per 10% ratio, then predicted efficiencies are given by multiplying the efficiency of the first gear train by that of the second. This gives predicted losses of 3.7%, 2.6%, 7%, 6%, 8%, 3.3%, 5.5% for gears 1-7 respectively.
The above predictions may be representative of a well run-in gear using a slippery lubricant, and may (assuming a conventional spread of gearing use) average about 7% (in addition to chain and other fixed losses which may be of the order of 3-5%). Thus total losses may be in the region of 10-12%.
However gearing losses may be nearer doubled in a gear that is poorly lubricated or not run in yet, with similar fixed losses. In this case average total losses of 17-19% might be expected.
In contrast measurements of (well-run in, well lubricated) simple single stage epicyclic gears with chain drives can show total losses of ~4-7%, with an average of nearer 4% if the direct drive gear is favoured. So other gears (e.g. most five speed and some other 7 speed gears) might well be noticably more efficient than a Nexus 7. It should however be noted that 10% losses don't usually mean a 10% speed reduction; about 5% reduction might be more likely at modest speeds.
-----------------------------------------------------------------
How the gear ratios are achieved:
The first (reduction) gear train comprises a single ring gear, with three dual-stepped planet pinions, each step having its own sun pinion. It has three ratios;
D = Direct (1.0)
L = Low (0.741)
B = Bottom (0.632)
The second (increase) gear train has similar parts, again with three ratios;
D = Direct (1.0)
H = High (1.335)
T = Top (1.545)
The clutching gives the ratios in combinations thus;
Gear 1 = Ratios BD = 0.632
Gear 2 = Ratios LD = 0.741
Gear 3 = Ratios BH = 0.843 = 0.632 x 1.335
Gear 4 = Ratios LH = 0.989 = 0.741 x 1.335
Gear 5 = Ratios LT = 1.145 = 0.741 x 1.545
Gear 6 = Ratios DH = 1.335
Gear 7 = Ratios DT = 1.545
unused near-duplicate ratios;
BT = 0.976
DD = 1.000
--------------------------------------------------------------------
How the gears are configured:
From right to left in the hub;
Gear train 1
Ring 1 = 72T
Sun 1 = 42T
Planet 1/2 = 14T/20T
Sun 2 = 36T
Gear train 2
Ring 3 = 66T
Sun 3 = 36T
Planet 3/4 = 14T/19T
Sun 4 = 30T
-------------------------------------------------------------------
Ratio calculation:
In each case the gear ratio is given by the formula;
Vr = Vpc (1 + (Ns x Npr)/(Nr x Nps))
for stepped planet pinions which simplifies to;
Vr = Vpc (1 + Ns/Nr)
for simple single step sun/planet/ring configurations, where;
Vr = ring gear speed
Vpc = planet cage speed
Ns = teeth on sun pinion
Nr = teeth on ring gear
Nps = teeth on planet in mesh with sun
Npr = teeth on planet in mesh with ring
So Vr/Vpc or Vpc/Vr gives the ratio depending on the direction of drive.
E.g. in gear 2 the ring gear is the input and the planet cage is the output, using ring 1/planet 1, and sun 2/planet 2 so the gear ratio is
Vpc/Vr = (1 + (36x14)/(72x20))^-1 = 0.741
------------------------------------------------------------------
Clutching configurations:
Each sun pinion can be locked in one direction and is permanently free in the other. An additional clutch C1 can transfer drive from the sprocket/driver Ring 1 to PC1. PC1 is permanently engaged with PC2. Drive is transferred to the hubshell from PC2 directly (gears 1 and 2) or from Ring 3 (all other gears).
Configurations are as follows:
Clutch.....C1....S1...S2...S3...S4
default...AFB...AFF...AFF..AFB..AFB
Gear1.....---...LBa...--...--...---...
Gear2.....---...LB....LBa..--...---...
Gear3.....---...LBa...--...--...LFa...
Gear4.....---...LB....LBa..--...LFa...
Gear5.....---...LB....LBa..LFa..LF....
Gear6.....LFa...LB....LB...---..LFa...
Gear7.....LFa...LB....LB...LFa..LF....
Where:
AFB = Always Free Backwards
AFF = Always Free Forwards
LB = Locked Backwards
LF = Locked Forwards
a = active clutch (i.e. .loaded, not overrun)
S1 can back up S2, and S2 can in turn back up C1.
S4 can back up S3, and the PC2 pawls can back up S4.
Backup clutches are overrun if engaged at the same time as another clutch in the same train.
Total loss of drive is theoretically possible but would require the failure of multiple pawl sets.
----------------------------------------------------------
Shifters, cable pulls and clutching movements:
Most shifts require a ~5.00mm cable pull and actuate one or two sun clutches. However the 6-5 shift requires clutch C1 to be moved as well as S3. This shift has a longer cable pull (almost 7mm), in part because of the C1 mechanism which requires a longer stroke and a higher force in one direction. Note that no single shift requires more than two clutches to be moved.
It will be noted that there is an unused DD ratio that would give a lossless 1.000 gear. Cynics have suggested that this is not used because it would stand out like a sore thumb vs the adjacent lossy ratios. This may be so, but it didn't stop them from putting the most and least efficient gear ratios adjacent to one another in the Nexus 8 gear.
It seems to me that it is just as likely that the clutching could have been made too complicated had they done this; both the 3-4 and the 4-5 shifts would have been C1 shifts also moving a total of two or three clutches.
The shifter cable pull pattern is unique to the Nexus 7 gear and there is no exact substitute shifter; although having said this, the S-clutches are not quite as sensitive to cable adjustment as in N8/A8 gears. It is said by some that a SRAM X-series gripshift for 7/8s derailleur gears ( 7 clicks of ~5.0mm cable pull) will operate an N7 gear, but it leaves the adjustment incorrect in at least two of the seven gears, which will make exact setup critical, and will doubtless lead to trouble down the line.
Shimano shifters include single-action, single lever pods, and various gripshifter types. The shift is reverse pattern, i.e. the hub defaults to gear 1 and the cable is pulled to access higher gears.
Sun Race make an N7 compatible shifter. In addition some Gazelle models come with a Gazelle-branded part that appears not to be a shimano shifter, but has the correct shift pattern for an N7 hub.
---------------------------------------------------
Common faults include;
Slippage damage; bad adjustment that leads to slippage will damage pawls; even if this doesn't by itself result in persistent further slippage, chips from damaged pawls will get to places they shouldn't and this will further damage the gear.
Water ingress - usually damages the ball ring first, and the clip can collapse; this sometimes necessitates a new hubshell.
RH cover damage- an unshipped chain can damage the cover. Replacements are available; be sure to get the right one for your internals (NB. a replacement internal may use a different cover from the original).
RH hub bearing damage; ball-clip can collapse, eventually leading to a failed driver and RH cone. Only a mobile lubricant will migrate internally to this bearing, so many forms of gear relubrication are not effective here.
Dried/inadequate grease; leads to water damage, slipping, and other general wear, e.g worn planet pinions.
Dirty cassette joint; leads to bad shifting and eventually other hub damage.
Internal debris; gear clearances are not large. Quite small pieces of debris (<0.5mm in size) can interfere with gear meshing and cause the gear to perform badly.
Shift sleeve damage; forcing the shifter on loaded upshifts can twist the shift sleeve, potentially making the S2 and S4 locking out of phase and/or overly sensitive to adjustment (typically gear 2-3-4 shifts affected). It is best to shift under light load or no load whilst pedalling forwards with this hub.
Assembly fault: longitudinal clearance in the gear assy (retained by a snap ring on the left end of the Centre) is often not large; comparatively small pieces of debris can lead to seizure.
Assembly fault: S2 and S3 form a subassembly where the parts are retained by an internal snap ring, but are free to rotate WRT to one another. The snap ring can be damaged during factory assembly, leading to a (worse in some gears, but OK in gear 5) fault (the suns won't rotate freely) and many other faults if pieces of damaged snap ring are floating around in the gear.
Assembly fault: if a Centre has slack longitudinal clearances, and the RH ball clip is in any way imperfect, individual balls from the RH bearing can come free when the centre is installed. This can lead to rapid and total gear failure.
Assembly fault: when the Centre is installed in the hubshell, it is very easy to dislodge one or more of the hubshell drive pawls on PC2 or R3. This leads to a variety of immediate faults. It is best to rotate the Centre backwards as it is inserted into the hubshell, and/or insert the R3 pawls into gaps in the hubshell dog ring.
Assembly fault; some 3/32" sprockets have offset drive lugs; it is a bad idea to install these with the lugs to the right; this can wear the driver splines, which are quite short.
Assembly fault: if the clip for the main ball-ring is in any way imperfect, individual balls can escape from it during centre installation. It is best to install the RH cover as a final operation, having adjusted the hub bearings and verified that the ball ring is fully populated.
---------------------------------------------------------------------
NB I may add pictures to this thread at some point.
cheers