# Grip Shift teardown

No, Grip Shift is not a new assembly language operation. This article is about bicycles and mechanics.

So what *is* Grip Shift? It's a kind of gear shifter I *really* like. Twist the handle and the gear magically changes:

A demo of the grip shift

See how nice it is? Low profile, no parts sticking out, nicely conforms to the shape of the handlebars. Compare it to a trigger shifter:

This looks like a piece of slag from a crashed spaceship. (This one even works about as well as a piece of slag.)

Grip Shifts were (are?) popular on mountain bikes because you can keep a strong grab on the handlebars while changing gears. It's also nice that I can shift several gears in one smooth move (limitations apply).

However this format also has some downsides. If you ever tried to open a jar with wet hands, you know what I'm talking about. The round shape makes it harder to turn the shifter when it's raining and the surface gets slippery. From personal experience, wrapping a sock around the rotating part helps, but how often do you ride around with spare socks?

The rubber on the handle helps increase friction, but it becomes a problem of its own. It wears out and smoothens! That's doubly bad news when you managed to take good care of it, and it wears out 10 years after manufacturing stopped, when finding a replacement rubber is next to impossible.

But those are not the problems that pushed me to do all the work you'll read about below. You see, I wanted to ride with 9 gears instead of my old 7. All other things constant, those 9 are packed tighter on the shifter than the 7:

I started accidentally switching 2 gears at a time when I only meant to switch one! Turns out people complaining about the precision of grip shifters online weren't making it up.

What now? If you think I'm going to put up with this forever, you're wrong. I'll solve the problem or I'm not a hacker. So here's the plan:

1. Read up about twist shifters and derailers.
2. Get a shifter that works for me.

Who am I kidding? In reality, it went something like that:

1. Take my shifter apart to see how it works (easy because it disintegrates when you replace the cable).
2. Make up my mind to manufacture my own, with the exact gear spacing I like.
3. Jump into the rabbit hole of pull distances, hubbub, shifting systems, and cog pitches.
4. Get a lot of Grip Shifts across generations and learn how all of them work, to choose the best design.
5. ???
6. Transcend the mortal form.

Today I'm at step 4 and that's why you have this writeup. But before we proceed to the teardowns, let's get up to speed with the basics of bike gears and shifting.

## Shifter frontend and backend

The job of the gear shifter – the API contract it exposes to the user – is that you twiddle with something with your hand and then the pace at which you need to pedal changes. For grip shift, the twiddling is the rotation of the handle.

That's the "front end". There's also the "back end": what does the twisting action translate into?

The interface on the other side of the shifter is exposed by the metal cable that disappears in the gear mechanism. The cable is meant to be pulled across a range of lengths and held in position. It's also tensioned: it pulls back automatically to one end when no force counteracts it:

The extreme positions of the cable correspond to the lowest and highest gear.

This is the generic API for controlling gears. This is how it works with a derailer (both front and back), and, as far as I can tell, also with hub gears. I could imagine continuous transmissions also expose this interface. It turns out that pulling on a metal cable is an elegant abstraction!

## Derailer

Abstractions are leaky. The above contract doesn't yet describe everything you need to understand the upcoming teardowns. Think about it: switching gears across a 7-gear cassette will be different than a 11-gear one. If only because there are more stops.

The derailer is the consumer of the cable pull contract. It then interfaces with the cogs, which are arranged from smallest to biggest (black); and the chain. To do this, it uses a small guiding cogwheel (red) that pushes – **derails** the chain from one cog to the next. The cogweel moves – you guessed it – because the cable changes position:

When we add the chain to this simplified model, it turns out that we need one more cog. The chain should leave on the same side it came in:

So what leaky abstractions are there?

### Indexing

Gears in this setup are discrete. You can't use cog 3,5. It's either 3 or 4, no in-betweens! If you pull the cable to position the guiding sprocket between 3 and 4, you'll hear telltale grinding noises, and your confused chain will start skipping between them. Go get your bike and try it. Actually no, don't try. The thought of a casette abused by a skipping chain gives me almost physical pain.

This abstraction travels back up the cable and appears in the shifter as *indexed shifting*. Meaning: the shifter will not hold the cable in invalid positions. Only in those which correspond to gears – unless you mistweaked the adjustment rings. Or bent the derailer. *Shudder.*

### Pull distance

There's another leaky abstraction here. How much cable to pull to move from gear 3 to 4? The cogs have a defined distance between them, called *cog pitch*. This is **not** the amount of cable to pull.

Turns out, there are multiple standards for how much cable to pull in order to get a certain movement of the guiding wheel. That's called the *pull ratio*. Honestly, I don't know why the pull ratio wasn't standardized to be constant. There are probably reasons related to how much cable can fit inside a shifter. We'll see echoes of that in teardowns.

Cog pitch from the cassette and pull ratio from the derailer both leak into the shifter: you need to pull the cable by *X*mm in order to get the next gear. That is called the pull distance. It depends both on the cassette you have installed and the derailer model.

Replace the derailer and you need a new shifter. Replace the cassette with one having a different number cogs and you need a new shifter.

Or at least this is the mathematical principle. Bike equipment manufacturers, driven by the usual consumerist forces, prefer to not make this too complicated, and they just sell all three together as a set. So if you try to find the last part for your custom matched set, you may get lucky or… not. Either way, you're on your own. Here's a bunch of tables to help.

Do you still wonder why I thought making my own shifter is a good idea? If I could design a parametric one, I would only need to tweak it to match whatever random cassette and derailer I already have.

But what does it take to design one? Let's take a careful look at existing designs and find out.

## Grip Shift MRX – 2000's?

[mrx blown apart]

The closest to the Platonic ideal of a Grip Shift out of my samples. You could make it simpler, but I don't believe SRAM ever went simpler. I'll go deeper into that topic around the SRT-300 design.

For now, let's stick to the simplest operational model we have:

The cable wraps as its end is pulled around the handlebars in the perfect mathematical space. What does it look like in real life?

(The spring is visible on both pictures. It sits in the black, fixed part, but I decided to keep it on the white part to show how it meshes with the notches.)

There's lots of plastic, plus some metal, rubber, and grease, all to:

• translate the rotating motion of the hand into cable pull,
• guide the cable along the desired path,
• give a different feel at gear positions,
• hold the cable in selected gear,
• make clicky noises, and
• make cable replaceable.

And that's before ergonomics like a grippy rubber are even considered. But let's break it down in a different way. It's easier to consider two main blocks of functionality separately: the **cable path** and the **ratchet**.

### Cable path

At risk of sounding silly, if you want to wrap one thing around another, one needs to move and the other needs to be fixed. The handlebars are fixed here. Also, one half of the MRX is fixed: the one where the cable enters. There's a screw on the side which goes through the plastic piece, through the metal ring, and bites into the handlebars.

You can also see it from the inside.

That screw will come back to bite *us* later.

The white piece is the rotating one. It's where the cable end is attached. When you twist it, you pull the cable away from entrance in the black fixed piece and spool it on the white piece.


The white ring is about the same distance from the handlebars as the rubber grip (28mm diameter), meaning that if you wrapped the cable around the grip directly, you'd pull it about the same amount.

The cable end disappears behind the lifted arc.

It can be seen when you lift the corner of the rubber. Pull it out to replace it.

Now that the cable is out of the way…

### Ratchet

This is the part that goes clicky-click, scaring people on combined bike-footpaths who haven't heard my bell when I slow down to avoid frightening them.

Look at the shape of the metal piece. Remember it because you will look for it on the floor as soon as you open the shifter. It's sitting in the piece attached to the handlebars, but to actually do anything, it needs the second half. Go ahead and scroll back to the disassembly picture.

The spring rubs against the wavy pattern. Each notch is shaped to fit the top of this metal spring neatly. When the spring falls into the notch, it locks into a gear. When you twist the white piece, it moves relative to the spring, and the spring falls into the next notch. There are 8 notches on mine, because the shifter is 8-gear. Makes sense!

The shape of the wavy part defines the feel of moving from one gear to another, but the spring also has a say. It doesn't sit tightly in its spot, but can actually move back and forth a little bit. That gives a little bit of play when a gear is fully engaged.

But why? I think it's to make shifting more reliable. When you shift from one gear to another, you don't just want to dial in the exact position. Mechanical elements have tolerances and are elastic, and aiming at the exact place may not be enough to overcome the resistances. Instead, you should overshoot your goal. The play gives the necessary hysteresis. It doesn't change how much the cable will move on a click, but it means you start from a higher (or lower) baseline.

### Shifter notes

This is not the oldest design I have. But it's the simplest starting point.

Unlike all other shifters I took apart, this one is held together by a latch that is easy to undo. So (foreshadowing) it does not unexpectedly come apart into a thousand pieces.

## Sachs Plasma – 2000s?

The Plasma looks like an improvement over the MRX. The rubber grew fancy, and it's 9-gears instead of only 8. Apart from that, it attaches to handlebars using a tightening ring rather than a grub screw.

But I really don't know which one came first. The MRX can still be bought on SRAM website. Someone on the Internet referred to the rubber as a 1990's thing. If anyone can **help me date those designs**, please get in touch. I'll credit you in the updated article.

The ratchet is pretty much the same. 8 gears, 9 gears, same difference.

The line wraps around a center in pretty much the same way. But the cable is threaded in a different way. The rubber flap moved to the fixed piece. To take out the cable end, you first need to dial in the last gear.

I'm using this shifter now and I often accidentally switch two gears at the same time. I'm not impressed, and I join the crowds of online complainers. That didn't happen with my old

## SRT-300 – 1990's

which I've been using for over 20 years. Indeed it's the oldest one in my collection.

I personally vouch that it's built like a Soviet car. Annoying at times, but indestructible and with nothing hidden from you. It survived all my serious crashes, I only greased it maybe 3 times, I wore off the rubber (took longer than a decade), and 1000km a year for me is little. If it had 9 gears I'd still be using it. Alas, for mechanical reasons it's more complicated than it seems.

The previous two are basicaly using the same ratchet design as this one, with minor modifications. The spring on this one is a bit smaller, and I think that's it.

But this one is legendary. Because of

### Cable routing

If not for this shifter, I would not have gotten interested in the topic. You see, when you try to replace the cable, this shifter literally opens itself up to you, showing you all it has inside, sharing all it has to give (the floor spring and grease).

And what you find inside is not straightforward. The cable end is not sticking in the moving part, but… somehow it ends up back in the fixed piece again. And it does a full turn inside the mechanism, too.

Why? And what actually pulls the cable if it both starts and ends in the same piece of plastic? Does the cable actually reverse inside like the thread in a sewing machine? (Spoiler: it doesn't.)

Trace the cable: from the entrance, to a ramp, then down off the ramp and on the inner ring, and finally in the end holder.

When you spin the shifter, you lift some angle of it from the inner ring up on to the ramp. Something that used to be wrapped on a small circle is now wrapped on a large circle. Large circles have larger circumference, meaning more cable is needed to cover the same angle! The pull comes from lifting the cable up on the ramp, because of the difference in circumference. And that pull is *less than if the cable was pulled directly*. It's like a mechanical subtraction engine!

Again, why? This looks so unnecessarily complicated!

I measured the diameters of both circles: 42mm and 26mm. How much cable does a quarter turn pull?

And how much does a direct solution like the MRX pull? The diameter was 28mm, so it's pretty straightforward:

The MRX pulls 7/4 as much cable, even though the cable is wrapped on the smallest circle possible already. Turn it around: to pull the same amount of cable, the MRX must be pulled 4/7 as much as the SRT.

That doesn't sound significant until you realize that a quarter rotation is enough to shift across (/me carefully eye-measures the shifter) *the entire gear range* on the MRX. That's a lot of gears in one rotation! Squeeze an extra gear in like on the Plasma and you get about 10° angle per gear.

Meanwhile, the SRT lets you rotate a little bit more per gear, a little more precisely. The 7- and 8-gears systems are slightly different, but a back of the napkin calculation gives me almost 20° per gear, with 130° total spin range. And remember the picture from the beginning:

That would explain why the SRT felt a lot more precise in my hands.

SRAM was clearly aware of this issue. All 3 shifters so far were designed for Shimano derailers with 1.7 pull ratio. Other shifter models are meant for SRAM's own derailer system called "1:1" which has a 1.1 pull ratio (naming is hard also outside of computing). Meaning: to shift the same amount at the cogs, you need to pull more cable at the derailer. Which is great if you've been pulling too much per index.

I found a picture of a 1:1 equivalent of the Plasma:

This version is so much less cramped they could spare some dots between gear numbers!

## Grub screw

In all previous designs, the half which is kept in place is being held by a ring and a screw. Together, they interrupt the insides of the mechanism, going from the handlebars all the way outside. You can't slide anything completely around this space, which is bad news for the range of rotation of the moving half.

Believe my word for now that the best position for the ratchet spring is opposite of the screw. On the following diagram, I mark both extreme positions of the notches on the moving half.

As you see, the screw section means you can never exceed 180° between the extreme positions. On one side, the notches must always overlap the ratchet, but on the other, they must never overlap the screw.

That screws you up if you want to pack 10 or 11 or 12 gears at the same angular spacing as you had in the 7-gear model – no matter the pull ratio.

Yes, you could get around this limitation. Push the screw farther out, and out of the way. Oh no, the shifter is now longer by a few mm! Acceptable trade-off? Not to SRAM, their many-geared models go in a different direction.

(TODO: get an old, SRT-era ESP. Is the design different? They are already on 1:1 so they don't need a ramp, and they can't possibly fit 7 wide-spaced 1:1 gears within 180° on the same design.)

### Notes

You would think that the cable sliding up and down the ramp would rub it through. 20 years of gear shifting says you don't need to worry. Better start thinking about DIY rubber grips though.

Another thing that went wrong is the missing shield protecting the cable. it's gone on all the shifters I have and most I came across. But it doesn't open much of a gap and doesn't seem to affect longevity, at least not on human scale.

## X.O 9-gear (before 2012)

This is a 1:1 model. But the pull ratio is not the most interesting thing about it.

Every time I touch this one, the sticky rubber makes a dirty mess on the desk.

From the outside, it's the flimsy transparent gear indicator which is damaged on both shifters I collected. We moved from Soviet car build quality to something that's flimsy and that can be repaired, badly, to some extent.

Inside, the story changes. While the cable is routed like on the Plasma, the ratchet is more like a complete redesign than an evolution. Even then, you have to look for a spring on the floor.

### The ratchet

What struck me first is a long, coiled spring. It literally did, trying to escape its life under pressure inside the mechanism. Let's look at the mechanism first and find the spring later.

There are 3 pieces now. There's the usual piece with the handle. In this model, it lost both the cable and the ratchet notches. The fixed piece now carries the notches, and there's also another moving piece – with the actual ratchet spring, as well as the cable – attached to the fixed one via magical means I was unable to overcome. You can turn this new part, but it only clicks one way. The old ratchet spring is now straight, and in this position, it holds the ratchet notches in place, blocking the possibility to release the cable.

X.O spinning in the easy direction

Pay attention to the screw driver above. It presses on just some piece of plastic and the whole thing rotates - but only in one direction.

Look at the rotating half, and it doesn't seem to even do much. Without the ratchet and the cable, its job is to mesh with the ratchet spring on the other side. Press the spring gently like the rotating half would, and you can rotate the whole thing in the release direction.

X.O only easily spins in one direction when pressing the spring

Actually, it looks like the moving half and the moving piece attached to the fixed half need to stay at similar angles all the time. But they are not attached perfectly: there is still some play – just like before! Except before it was the spring that was allowed to move inside its box. But thereat's not all: this time, the cable is not attached to the handle, so you don't control it directly. The cable follows the internal piece, not your hand.

This means SRAM removed direct control of the cable! Now when you turn to release the cable, nothing happens at first, until it suddenly goes all at once. The goal being crispy upshifting, I guess? I'll have to try to see if it feels good.

he lack of direct control can be seen really well on the next shifter. Here, the inside moving piece is white, and it also holds the cable. It can be seen from the outside through a little cable hole. See how the cable moves in steps?

X.0 shifting up and down

Also, now there's another spring. It fits between the moving half and the semi-moving piece, tensioning one against the other. When I try to release without a cable installed, I feel resistance! It's actually easier to pull on this one than release. My guess is that this design change also exists to solve the slippery grips problem by giving you extra force to counteract the spring in the derailer. Kudos to them. Gonna have to try it out, too. Still, I think this can be achieved without introducing an extra moving piece. If there was a spring pushing the loosely held other spring... anyway.

### Notes

Flimsy. Bulky. I have no idea how to disassemble all the moving parts. Spring annoying to stick back in. But 9 widely spaced indices means I will at least try it out soon.

The rubber flap again. Worse than before. I had to disassemble it to remove the cable on one of mine.

The indicator window is not needed for the correct operation of the shifter, so you might think that even when it's broken, the shifter can still live for a long time. Except then dust has a nice way to the internals, which will probably get clogged with some regularity, degrading the experience of using the shifter.

The front shifter also has 9 positions, even though the markings are for 3 gears. It's advertised as a "micro-ratchet". Instead of shifting into the one position that hopefully corresponds to your gear, you can shift into a position that actually works.

It's true that it's hard to calibrate the front shifting to actually match the chainring position. And it's also true that I often turn the shifter by a fraction of a gear to get rid of the grinding. But I also like that 90% of the time I just pull it and I know that it lands approximately right. I haven't used this system yet, so I don't know if it would be an improvement.

## X.0 10-gear (2012)

The pinnacle of SRAM engineering! The newest model, judging from the outside – it looks the same as the top of the line models sold today.

This model is pretty similar to the 9-gear X.O, except it introduces a bearing between the parts for whatever reason. I don't really see the point. It doesn't move any smoother from that one. Yes, it takes less force to overcome the ratchet, but that's not due to the bearings. Puzzling.

The bearing has been secured with some extra metal. I didn't even bother trying to remove it.

This is the one that I recommend NEVER to take apart. While replacing the spring on the X.O 9-gear was merely annoying, here it raises to a form of art. I've read forum posts where people bought a new one before succeeding. Me, having barely paid anything for this, I have completely different incentives. Naturally, I would not rest until the spring was back in.

I don't know if the pipe was part of the original package.

If you're ready for hell, the gates to it lead through two latches on the *inside* of the grip. Unlatch them and pull out the small plastic ring to enter.

### The nasty spring

Like before, the spring attaches between the moving white part and the moving grip half. Unlike the X.O 9-gear, though, there are no guiding pins where you can stick the ends. And if you try to brute force it, you will find that just as you're ready to press both pieces together, the spring jumps out onto a different track and you're hosed.

So let's go back to the beginning and stop before the spring runs for it. I'm sure you already got this:

Sorry, no picture of the spring in place. By that point, it was already so bent it would no longer stay there.

Take the spring by the ends, put it inside the grip. Use a cross head screwdriver to push the end outside and into the groove.

Now stop.

Make a semicircle out of a rigid, flat material. I cut some plastic with scissors. The inner part should fit onto the inner circle. Also don't make it too wide – read on to see why.

Take the moving part where you just stuck the spring. Cover the spring with the semicircle, and then finish the sandwich with the white part. The spring is not fitting because of the nib in the white piece. Thankfully, if you align the pieces right, there's a gap that opens up to the nib. Rotate the semicircle towards the back, letting the spring touch the white piece. The catch? The semicircle must be short enough to fit between the white end stop on the far side and the open part near the gap. Manipulate the spring so it goes on the white nib (I used a flat head screw driver).

There is an opening where you can stick a screw driver if you align the parts just right. Keep the left part of the spring in place using the semicircle. Use a screwdriver to move the other end onto the white nub (following the arrow).

Once you're ready, squeeze the pieces even firmer (I'm sure you've been doing a good job of it, but now you need to focus). The gap should be minimal now that the spring is more or less in place. Keep an eye on the spring falling out, and quickly remove the semicircle all at once. If you did it right, the gap is completely gone. Quickly redo the latches on the inside of the grip to make sure everything stays in place.

Sadly, I don't have a video of it. I haven't recorded my first attempt and I'm not ready to face this challenge again at the moment.

### Grub screw

In this design, the ring affixing the shifter to the handlebars is pushed completely outside of the enclosure. The ring encircles a tube-like thing made out of plastic which connects to the actual fixed half.

### Notes

This one doesn't have any rubber flap to open to replace the cable. In this design, you need to remove the flimsy-looking fairing to reveal the cable holder. I'm not sure what to think about it. It's certainly going to be annoying that the brake needs to be repositioned, but there's less rubber to get lost.

It also comes together with a full-length rubber handle that snaps onto the shifter from one side. Is it to stabilize the design? The holding ring is on the extreme other side, so maybe it doesn't offer enough support against twisting alone.

In addition, this model is a "1×10" setup, where there is simply no front shifter.

The extra-fixed bearing means that the third piece cannot be removed without much effort. Together with the nasty spring, this shifter is effectlvely useless for hacking.

Come on, SRAM. Replacing the spring on the X.O 9-gears is a breeze. Don't tell me you couldn't do that here.

## Summary

The biggest flaw of the grip shift is, to me, the rubbery grip. Or more like, the manufacturer's approach to it. All my grips are over 10 years old, and all are working. The most beaten part of each of them (with the notable exception of X.O 9-gear) is the rubber, and you can't buy a replacement anywhere. It's some uncanny valley of reliability. If the rubbers failed every 3 years, I would have learned pretty quickly to stock up for decades. But no, it took me over a decade to need a new one. The front one is still workable, after two decades! And now I have to resort to gluing bike tubes together to keep an otherwise perfectly useable device working. That's depressing.

This works a lot better than it looks.

Is anyone up for 3d-printing some?

One thing I was worried about when switching from the SRT to the Plasma was the size. On paper, there's a big difference, but on the bike I don't notice. Maybe because my bike is already all kinds of bent. Anyway, here's a big table with all the numbers, relevant or not.

### Comparison table

Build quality expresses how long I think the shifter will work before significantly degrading. I have only tested the SRT-300, and I have some data about damage on the X.O 9-gear. The other ones are educated guesses based on how similar the design is. The X.0 gues is not even educated.

Rain resistance is split into two numbers: the first takes into account the shape of the rubber, while the other ignores it, simulating using the shifter for over 10 years.

Hand to brake distance describes how wide the shifter is, effectively: the distance from where the brake lever is mounted to where you can hope to keep your hand. For the brake, it's the shifter's edge. For the hand, it's a bit more complicated. All the handles rapidly expand at some point to give place for the mechanism. I chose a point that's about 3mm wider than the base diameter.

Entries marked with "?" are guesses.

Scores are higher = better.

### Table of pull distances

The pull distances were measured in the following way: the shifter was set to its lowest gear (cable most pulled). The cable was marked with a marker pen or tape near where the cable goes into the mechanism. The shifter was positioned over a sheet of millimeter paper so that there's a thick line where the cable goes out. Then the cable was manually tugged to straighten it and pull it out. The position of the current gear was marked on the millimeter paper (except Plasma, where it was just eyed). Here's a picture of the setup:

The measurement method is not very reliable, so I measured some shifters twice. So it's not clear from this picture that Shimano 8-gears (MRX) is the same pull distance as 7-gears (SRT) over a larger range. It's also not clear that on a Shimano derailer and cassette, every range should be the same except for 7-gears. That's partially due to poor measurements and partially because shifters often have a longer 1st and last gear. Just because they can spare the edge space.

What *can* be seen is that the X.O 9-gear has a very different pull ratio. That range won't work with Shimano! 1.7/36.5*23 ≈ 1.1 so the math for 1:1 kinda sorta checks out.

What about X.0 10-gear? As far as I know, 10-gear systems use a different cassette. Possibly wider. The pull distance is now the more familiar 3mm, over 28.5mm total. What pull ratio is it tuned for? I have no clue. Back of the napkin calculations say something around 1.3 to 1.4.

### Table of angles

This is just a rough measurement involving a protractor, a piece of bent wire and an empty stomach. Don't trust those numbers much. The Plasma was not measured at all because it's on my bike at the time of writing. All angles values are in degrees.

Two superior ranges stand out: the SRT-300 (due to its differential mechanism) and X.O 9-gear (due to 1:1 pull ratio). The SRT has an unbeatable step.

## My own design

This blog post already took more effort than I could reasonably spare while I'm working on Jazda. As someone with a thousand projects going on, I'm not going to make any promises, but I can disclose I have some sketches in my notebooks. Once I turn them into something real, I can guarantee I will be excited enough to make a follow-up.

Whew. 5 shifters, nearly 7000 words, 4 animations, 4 videos, 6 diagrams, 28 photos. Now no one can say I collect junk and then don't do anything with it :)

## Moar?

There are designs I haven't covered. Want it tested? Send me your shifter and I'll do it justice.