Another day, another wheel (and so on).
A customer brought in a wheel from a brand whose name I cannot speak aloud.
Apparently something pretty serious is going on with it and they want it fixed.
The wheel is laced 2:1, and this confirmed
that yeah, this brand really doesn't understand wheels
all that well.
But if I just said that, I'd be throwing stones, so
I'm going to write out the suspected causes and countermeasures too.
Be grateful. ←(condescending tone)
For the sake of future photo shoots, just discretion wasn't enough, so
I taped over the brand name.
I found that if you stick it on tight, the logo shows through in relief,
so I decided to apply it loosely instead.
So the "pretty serious" issue is this:
the tire sidewall was rubbing against
the frame's left chainstay, stripping the paint.
Shoe rub, but for the frame. "Frame rub."
The customer said the bare carbon is showing where the paint is stripped,
and they only noticed it when they tried swapping to their race wheels before a race.
From what I've seen, there have been cases—not many—where someone put
a really slack wheel on a frame with a super-tight chainstay and it rubbed.
But frame rub usually happens on a wheel with a broken spoke.
I investigated the rubbing marks. I taped a reference point at the start of the rub.
→
→→
→→→
→→→→
The rub marks end around here.
There were rub marks on the tire sidewall spanning roughly 150°.
The wheel had no lateral runout.
The customer's frame is an Anchor RS9.
The chainstay is a bit tapered, but when I showed a colleague an actual RS9
with a 25C tire on it, the spacing between the tire sidewall and chainstay
wasn't abnormally tight.
The RS9's rear dropout is what's called a "straight dropout."
Straight dropouts don't allow the rear wheel to be mounted at an angle.
Classic steel road frames with slotted ends, or track bikes and some TT bikes
with straight-blade ends can end up with the rear wheel mounted at an angle.
If the customer somehow installed the rear wheel at an angle relative to the frame,
the rub marks would be continuous all the way around the tire sidewall.
Since we only see partial rubbing (though it's about half the circumference),
this is a wheel problem.
As for how the customer was riding, they weren't
twisting the wheel extremely on flat ground or anything like that—
it looks like the rubbing happened during normal climbing, standing on the pedals.
The customer's weight isn't particularly heavy either.
The tire is a Schwalbe Grand Prix 4000 S II in 23C, but
this tire grows roughly 2mm with age and use.
Actually measuring it, I got 25.1mm.
The fact that it's effectively a 25C now is less due to aging growth and more because
it's mounted on a wide rim in a stretched-tire configuration—
that's the main reason.
But as I noted before, even if the tire is effectively a 25C,
that doesn't make it incompatible with the RS9,
so I can't blame the frame or the tire.
It's the wheel's fault.
I removed the tire and checked the wheel's lateral centering.
It's dead-center, but honestly, being off would've been more encouraging.
A tightening pass might've fixed it then.
The spoke tension, given these conditions,
is sufficiently tight though not screaming-tight,
and there isn't much more room to tighten it further.
With this rim, this hub, these spokes, and this lacing pattern,
if we want to do anything else to help,
spoke tension adjustment is the only variable left.
If from here I tightened the nipples while keeping lateral runout at zero
and maintaining wheel centering, the tire rub contact area would hopefully shrink
like a tide going out: 150°...120°...80°...50°...0°!
But the wheel is so slack that tightening alone won't fix it.
It'd be nice if the story was just "this one unit was abnormally slack,
tighten it this much and it'll never rub again!"
But it's not that simple.
Which means this wheel rubbing the frame on an RS9 isn't something that happens
in 1 out of 100 units—it's not that rare.
Unless we eliminate the root cause, the problem won't be solved.
This wheel is laced 2:1, but
every complete wheel manufacturer doing 2:1 lacing
has noticed a certain principle that this one hasn't.
I think that principle is sufficient to explain the frame rub.
"Don't mess with left-right uneven lacing if you're not smart enough for it."
That cause is doing four-spoke-pattern lacing, and
24-hole four-spoke (12+12H) and
24-hole four-spoke (16+8H) are completely different things.
I've touched on this before, but let me write it again.
Take a standard left-right even 24-hole wheel and lace it four-spoke.
Given that rim and hub dimensions are the same,
when Y holes laced X-spoke gives Y/X the same ratio,
even if the number X changes between lacing patterns,
the spoke trajectory becomes perfectly similar.
Since they're similar, the final crossing angle is also the same.
From there, the final crossing point rotates and shifts by an angle
unique to each spoke count
(360° / the number of final crossing intersection points on one side),
and at the new location, a similar spoke pattern pops up—
repeat that pattern all the way around and you've got a wheel.
As a thought experiment, a 1440-hole 240-spoke wheel
would have spoke trajectories similar to the above diagram,
with final crossing points shifting by 1° each time.
The point I'm making is that "four-spoke" or "six-spoke" isn't an absolute standard
that determines the final crossing angle by itself. Even with the same rim and hub,
if the hole count conditions change, it's possible for a four-spoke and a six-spoke
to have the same crossing angle.
So 16+8H is 24 holes total, but
looking at just the right flange, it's no different from one side of a 32H hub.
The rim is 24H, not 32H, so technically there's a difference, but
this wheel, while appearing to be laced four-spoke on a 24H rim,
is actually laced in what's essentially a 32H four-spoke pattern.
This is the wheel's biggest design weakness.
Here, let me draw a flat diagram of a 12+12H spoke-reverse-side
radial lace rear wheel in hub-rim-hub format.
If I set the valve hole phase to 0°, it gets hard to draw, so
I'm setting the first rim hole clockwise from the valve hole to 0° phase.
I'm drawing the drive-side spokes and rim holes in blue,
and the spoke-reverse side in red.
I'm showing tangent lacing as one crossing (two-spoke) just to keep the diagram
from getting too messy with more crossings.
Next, I'll draw the 16+8H diagram too.
The rim hole phase on the rim line is the same as the 12+12H,
but because it's laced 2:1, the pattern is blue-red-blue repeating.
Why the valve hole neighbor is blue is because
on the actual wheel too, both neighbors of the valve hole
are drive-side spokes.
For zero-spoke lacing where the spoke trajectory lies on a radial line
(normal radial lacing is like this), it becomes perpendicular to the rim line.
First, I drew the spoke-reverse-side radial lace.
This wheel's 2:1 lacing is arranged so that,
viewed from the side,
the spoke-reverse-side radial spoke passes through the drive-side's final crossing. So
if I extend the red line to the drive-side, it should hit the blue final crossing.
Then just draw the evenly-spaced right-flange holes for a 32H hub equivalent,
and I've completed the flat diagram for a 16+8H 24H rear wheel.
I've put the 12+12H and 16+8H diagrams side by side.
When doing the same "drive-side four-spoke" lacing pattern,
the 16+8H spokes angle closer to radial.
In other words, even though both are "24H drive-side four-spoke,"
the 16+8H will have lower tension.
Plus, the spoke-tension-variance correction that comes with left-right uneven lacing is so extreme that
the spoke-reverse side can sometimes end up higher tension than the drive side.
I actually checked the H1ST.
One spoke-reverse side read 157.
One drive-side read just under 154, so I'll call it 154.
An important point I forgot to mention: this wheel uses
left-right identical-diameter spokes. But
if it were left-right different-diameter spokes, I couldn't determine
the second ST difference just by comparing H1ST.
Measuring with the wheel viewed drive-side clockwise for drive-side,
then flipped measuring counter-clockwise for spoke-reverse-side:
the "Yama" on the drive side refers to one side,
and "Han-Yama" (spoke-reverse) the other—
but that's not very meaningful.
If I number the rim holes clockwise from the top,
the drive-side reads 1, 3, 4, 6, 7, 9, 10, 12,
13, 15, 16, 18, 19, 21, 22, 24, and
the spoke-reverse side reads 2, 5, 8, 11, 14, 17, 20, 23.
The tire was rubbing around the opposite side of the valve, so
I thought maybe holes 10–15 or so had low tension,
but that's not the case.
Besides, if tension were markedly slack, there'd be lateral runout.
What stands out is that the spoke-reverse side is higher tension.
Why is higher spoke-reverse tension a problem?
I can hear the objection coming, but having the spoke-reverse side determine
the spoke-tension limit seems odd.
So there's a high-level technique where you deliberately use
left-right uneven-diameter or uneven-spoke-count lacing
with lower variance correction than left-right uneven count lacing,
to partially counteract the extreme correction of the latter.
You can't use this without truly understanding wheels,
and you won't even think of it.
As an example of left-right uneven count + reverse uneven diameter:
Colima made a 12+8H three-spoke-pattern wheel where
the spoke-reverse-side spokes are thicker (higher spoke specific gravity)
and I nearly fell over when I first saw it.
I wondered how many people besides me understood what that meant.
The Colima people might not be using "nomunlab terminology" to grasp wheel properties,
but they're seeing wheels with the same or better understanding as me.
↑This is it.
It recovered from the broken digital camera's SD card.
It was a point-check, not something we sold here—actually, it's rare we did sell this wheel.
↑Left-right reverse uneven-diameter lacing.
It's not an extreme spoke specific-gravity difference,
which gives the sense of calculated intent.
Also that quick release is so cool (not that it matters).
Anyway, back on track.
As a counter to left-right uneven lacing, choosing left-right
uneven diameter versus uneven spoke count—
uneven diameter is the harder path.
↑This is a DT Champion conversion table.
↑And a DT Competition conversion table.
Both are 14mm thread pitch, with
the vertical axis showing D1ST (the first tension reading on DT's gauge),
and the horizontal showing spoke tension (the second tension reading).
They look similar, but
the Champion table has D1ST from 1 to 3.5,
while Competition goes from 0 to 2.5,
so they're not actually the same table.
Technical archive
A customer brought in a wheel from a brand whose name I cannot speak aloud.
Apparently something pretty serious is going on with it and they want it fixed.
The wheel is laced 2:1, and this confirmed
that yeah, this brand really doesn't understand wheels
all that well.
But if I just said that, I'd be throwing stones, so
I'm going to write out the suspected causes and countermeasures too.
Be grateful. ←(condescending tone)
For the sake of future photo shoots, just discretion wasn't enough, so
I taped over the brand name.
I found that if you stick it on tight, the logo shows through in relief,
so I decided to apply it loosely instead.
So the "pretty serious" issue is this:
the tire sidewall was rubbing against
the frame's left chainstay, stripping the paint.
Shoe rub, but for the frame. "Frame rub."
The customer said the bare carbon is showing where the paint is stripped,
and they only noticed it when they tried swapping to their race wheels before a race.
From what I've seen, there have been cases—not many—where someone put
a really slack wheel on a frame with a super-tight chainstay and it rubbed.
But frame rub usually happens on a wheel with a broken spoke.
I investigated the rubbing marks. I taped a reference point at the start of the rub.
→
→→
→→→
→→→→
The rub marks end around here.
There were rub marks on the tire sidewall spanning roughly 150°.
The wheel had no lateral runout.
The customer's frame is an Anchor RS9.
The chainstay is a bit tapered, but when I showed a colleague an actual RS9
with a 25C tire on it, the spacing between the tire sidewall and chainstay
wasn't abnormally tight.
The RS9's rear dropout is what's called a "straight dropout."
Straight dropouts don't allow the rear wheel to be mounted at an angle.
Classic steel road frames with slotted ends, or track bikes and some TT bikes
with straight-blade ends can end up with the rear wheel mounted at an angle.
If the customer somehow installed the rear wheel at an angle relative to the frame,
the rub marks would be continuous all the way around the tire sidewall.
Since we only see partial rubbing (though it's about half the circumference),
this is a wheel problem.
As for how the customer was riding, they weren't
twisting the wheel extremely on flat ground or anything like that—
it looks like the rubbing happened during normal climbing, standing on the pedals.
The customer's weight isn't particularly heavy either.
The tire is a Schwalbe Grand Prix 4000 S II in 23C, but
this tire grows roughly 2mm with age and use.
Actually measuring it, I got 25.1mm.
The fact that it's effectively a 25C now is less due to aging growth and more because
it's mounted on a wide rim in a stretched-tire configuration—
that's the main reason.
But as I noted before, even if the tire is effectively a 25C,
that doesn't make it incompatible with the RS9,
so I can't blame the frame or the tire.
It's the wheel's fault.
I removed the tire and checked the wheel's lateral centering.
It's dead-center, but honestly, being off would've been more encouraging.
A tightening pass might've fixed it then.
The spoke tension, given these conditions,
is sufficiently tight though not screaming-tight,
and there isn't much more room to tighten it further.
With this rim, this hub, these spokes, and this lacing pattern,
if we want to do anything else to help,
spoke tension adjustment is the only variable left.
If from here I tightened the nipples while keeping lateral runout at zero
and maintaining wheel centering, the tire rub contact area would hopefully shrink
like a tide going out: 150°...120°...80°...50°...0°!
But the wheel is so slack that tightening alone won't fix it.
It'd be nice if the story was just "this one unit was abnormally slack,
tighten it this much and it'll never rub again!"
But it's not that simple.
Which means this wheel rubbing the frame on an RS9 isn't something that happens
in 1 out of 100 units—it's not that rare.
Unless we eliminate the root cause, the problem won't be solved.
This wheel is laced 2:1, but
every complete wheel manufacturer doing 2:1 lacing
has noticed a certain principle that this one hasn't.
I think that principle is sufficient to explain the frame rub.
"Don't mess with left-right uneven lacing if you're not smart enough for it."
That cause is doing four-spoke-pattern lacing, and
24-hole four-spoke (12+12H) and
24-hole four-spoke (16+8H) are completely different things.
I've touched on this before, but let me write it again.
Take a standard left-right even 24-hole wheel and lace it four-spoke.
Given that rim and hub dimensions are the same,
when Y holes laced X-spoke gives Y/X the same ratio,
even if the number X changes between lacing patterns,
the spoke trajectory becomes perfectly similar.
Since they're similar, the final crossing angle is also the same.
From there, the final crossing point rotates and shifts by an angle
unique to each spoke count
(360° / the number of final crossing intersection points on one side),
and at the new location, a similar spoke pattern pops up—
repeat that pattern all the way around and you've got a wheel.
As a thought experiment, a 1440-hole 240-spoke wheel
would have spoke trajectories similar to the above diagram,
with final crossing points shifting by 1° each time.
The point I'm making is that "four-spoke" or "six-spoke" isn't an absolute standard
that determines the final crossing angle by itself. Even with the same rim and hub,
if the hole count conditions change, it's possible for a four-spoke and a six-spoke
to have the same crossing angle.
So 16+8H is 24 holes total, but
looking at just the right flange, it's no different from one side of a 32H hub.
The rim is 24H, not 32H, so technically there's a difference, but
this wheel, while appearing to be laced four-spoke on a 24H rim,
is actually laced in what's essentially a 32H four-spoke pattern.
This is the wheel's biggest design weakness.
Here, let me draw a flat diagram of a 12+12H spoke-reverse-side
radial lace rear wheel in hub-rim-hub format.
If I set the valve hole phase to 0°, it gets hard to draw, so
I'm setting the first rim hole clockwise from the valve hole to 0° phase.
I'm drawing the drive-side spokes and rim holes in blue,
and the spoke-reverse side in red.
I'm showing tangent lacing as one crossing (two-spoke) just to keep the diagram
from getting too messy with more crossings.
Next, I'll draw the 16+8H diagram too.
The rim hole phase on the rim line is the same as the 12+12H,
but because it's laced 2:1, the pattern is blue-red-blue repeating.
Why the valve hole neighbor is blue is because
on the actual wheel too, both neighbors of the valve hole
are drive-side spokes.
For zero-spoke lacing where the spoke trajectory lies on a radial line
(normal radial lacing is like this), it becomes perpendicular to the rim line.
First, I drew the spoke-reverse-side radial lace.
This wheel's 2:1 lacing is arranged so that,
viewed from the side,
the spoke-reverse-side radial spoke passes through the drive-side's final crossing. So
if I extend the red line to the drive-side, it should hit the blue final crossing.
Then just draw the evenly-spaced right-flange holes for a 32H hub equivalent,
and I've completed the flat diagram for a 16+8H 24H rear wheel.
I've put the 12+12H and 16+8H diagrams side by side.
When doing the same "drive-side four-spoke" lacing pattern,
the 16+8H spokes angle closer to radial.
In other words, even though both are "24H drive-side four-spoke,"
the 16+8H will have lower tension.
Plus, the spoke-tension-variance correction that comes with left-right uneven lacing is so extreme that
the spoke-reverse side can sometimes end up higher tension than the drive side.
I actually checked the H1ST.
One spoke-reverse side read 157.
One drive-side read just under 154, so I'll call it 154.
An important point I forgot to mention: this wheel uses
left-right identical-diameter spokes. But
if it were left-right different-diameter spokes, I couldn't determine
the second ST difference just by comparing H1ST.
Measuring with the wheel viewed drive-side clockwise for drive-side,
then flipped measuring counter-clockwise for spoke-reverse-side:
the "Yama" on the drive side refers to one side,
and "Han-Yama" (spoke-reverse) the other—
but that's not very meaningful.
If I number the rim holes clockwise from the top,
the drive-side reads 1, 3, 4, 6, 7, 9, 10, 12,
13, 15, 16, 18, 19, 21, 22, 24, and
the spoke-reverse side reads 2, 5, 8, 11, 14, 17, 20, 23.
The tire was rubbing around the opposite side of the valve, so
I thought maybe holes 10–15 or so had low tension,
but that's not the case.
Besides, if tension were markedly slack, there'd be lateral runout.
What stands out is that the spoke-reverse side is higher tension.
Why is higher spoke-reverse tension a problem?
I can hear the objection coming, but having the spoke-reverse side determine
the spoke-tension limit seems odd.
So there's a high-level technique where you deliberately use
left-right uneven-diameter or uneven-spoke-count lacing
with lower variance correction than left-right uneven count lacing,
to partially counteract the extreme correction of the latter.
You can't use this without truly understanding wheels,
and you won't even think of it.
As an example of left-right uneven count + reverse uneven diameter:
Colima made a 12+8H three-spoke-pattern wheel where
the spoke-reverse-side spokes are thicker (higher spoke specific gravity)
and I nearly fell over when I first saw it.
I wondered how many people besides me understood what that meant.
The Colima people might not be using "nomunlab terminology" to grasp wheel properties,
but they're seeing wheels with the same or better understanding as me.
↑This is it.
It recovered from the broken digital camera's SD card.
It was a point-check, not something we sold here—actually, it's rare we did sell this wheel.
↑Left-right reverse uneven-diameter lacing.
It's not an extreme spoke specific-gravity difference,
which gives the sense of calculated intent.
Also that quick release is so cool (not that it matters).
Anyway, back on track.
As a counter to left-right uneven lacing, choosing left-right
uneven diameter versus uneven spoke count—
uneven diameter is the harder path.
↑This is a DT Champion conversion table.
↑And a DT Competition conversion table.
Both are 14mm thread pitch, with
the vertical axis showing D1ST (the first tension reading on DT's gauge),
and the horizontal showing spoke tension (the second tension reading).
They look similar, but
the Champion table has D1ST from 1 to 3.5,
while Competition goes from 0 to 2.5,
so they're not actually the same table.
Technical archive
Products Mentioned in This Article
* Amazon affiliate links — prices may vary
Original Japanese post: のむラボ日記 #3760