Headsets, Forks and Stems

Headset standards, and why the head tube decides

There are three questions to answer about any front end, in this order.

How do the bearings sit in the head tube? Directly in a machined seat, in cups pressed into the tube, or in cups that sit mostly outside it. That is the headset standard, and it is a property of the frame.

Is the steerer threaded or smooth? That is the fork, and it decides the stem type.

What diameter is the steerer? That is what the headset and stem have to match.

Those three are independent of each other in principle and heavily correlated in practice, because bikes came as sets. A frame with a threaded headset almost always came with a threaded fork and a quill stem. A modern integrated frame almost always came with a threadless fork and a clamp stem. But on used parts the assumptions break, and you should check all three rather than assume the third from the first.

The steerer diameters are exactly what the names say, which is worth stating because it is not always true. 1 inch is 25.40mm and 1 1/8 inch is 28.575mm, no rounding surprises hiding in either. Compare that to the 24mm crank spindle, which is really 15/16 inch and measures 23.81mm.

Modern BMX appears to have settled on one integrated design. Every source checked gives 41.8mm at 45 by 45 degrees, the Campagnolo specification. The other integrated geometry in the wider bicycle world, a 41.0mm bearing at 36 by 45 degrees, did not turn up on any BMX product. That is an absence rather than a proof, so read it as "nobody found selling it for BMX" rather than "it does not exist", but if you are buying an integrated bearing for a BMX frame, 41.8mm at 45/45 is what to expect.

The zero stack and external cup figures come from the general bicycle headset identification system rather than from BMX brands, so they describe the standard the frames are built to rather than anything a BMX manufacturer published themselves. Zero stack is uncommon on modern freestyle completes and more usual on race frames.

Integrated headsets

The design is the logical end point of the same thinking that produced the Mid bottom bracket. If the frame can be machined accurately enough, you do not need a cup to hold the bearing, and removing the cup removes weight, removes a part that can creak, and lets the whole assembly sit lower.

What you see when you look into the head tube of an integrated frame is a smooth angled seat, and a cartridge bearing sitting in it with nothing between the bearing and the frame. Above the bearing sits a compression ring and a dust cover, and that is the entire top of the headset.

Two things follow.

You cannot convert a frame. The seat is machined into the tube at manufacture. A frame that is cut for integrated bearings is cut for integrated bearings. There is no adapter that turns it into a pressed cup frame, and there is no reason to want one.

The frame is the wear part. On a pressed cup headset, a bearing that has been run dry and beaten up chews the cup, and you press a new cup in. On an integrated headset there is no cup, so a bearing that has been hammering away in a dry seat is beating on the frame itself. That is why an integrated head tube is worth inspecting closely on a used frame: look for a bright polished ring or a stepped groove where the bearing has been sitting.

Zero stack and pressed cup headsets

Both are pressed cup systems and both are found on used BMX frames, so the distinction matters.

On a traditional pressed headset, the cup is a top-hat shape: a short section presses into the head tube and a flange sits proudly on the face of the tube, with the bearing sitting in the part outside. Look at the frame from the side and you see a visible ring of headset above and below the head tube. This is what most older frames use.

On a zero stack, the cup presses further in and the bearing lives inside the head tube bore. From the side, very little shows. The head tube bore has to be larger to swallow the cup, which is why a zero stack frame and a traditional frame are not interchangeable even though both are "pressed cup".

The practical consequences on a used bike are height and identification. A traditional pressed headset adds more height above the head tube than a zero stack does, which raises the bars and changes how the bike feels before you have touched a single component. And when buying a headset for a used frame, "it presses in" is not enough information: you need to know which pressed standard the frame is bored for.

The good news for a buyer is that a pressed cup frame is repairable in a way an integrated frame is not. Cups take the wear, and cups can be pressed out and replaced.

Threaded headsets and quill stems

The threaded system was how every bike worked for a long time and it is what you will find on old school BMX. It is worth understanding properly because a lot of used old school parts are sold by people who have only ever seen threadless.

How the headset works. The lower race sits on the fork crown. The bearings run in cups in the head tube. The top race threads onto the steerer and sets the bearing preload, and a locknut is tightened down against it to stop the adjustment moving. Between the two there is usually a keyed washer with a tab that locates in a groove down the steerer, which is what stops the top race turning when you tighten the locknut.

How the stem works. The steerer sticks up through the headset and is hollow. The quill (the long shaft of the stem) drops down inside it. An expander bolt runs the length of the quill and threads into a wedge at the bottom. Tightening the bolt pulls the wedge up into the tapered end of the quill, spreading it against the inside of the steerer. That is the entire mechanism, and it is why a quill stem can be set at any height within its range simply by loosening one bolt.

The system has real advantages: height adjustment is trivial, and there is only one bolt to deal with. It also has failure modes that a threadless setup does not.

The expander bolt seizes. This is the classic old school problem. The bolt is steel, sitting inside a steel steerer, often for decades, frequently with water in it. On a used bike, an expander bolt that will not turn is common and a stem that will not come out is common.

The wedge drops. Undo the expander bolt too far and the wedge falls off the bottom of the bolt and drops into the fork steerer, where it rattles and is a real nuisance to retrieve.

Overtightening splits the quill. The tapered end is thin. Riders who have had a stem slip tend to respond by tightening harder, which spreads the quill further than it was meant to go.

Minimum insertion is not optional. A quill stem has a maximum height line marked on it. Above that line there is not enough quill inside the steerer to clamp safely, and the failure mode is the stem coming out of the fork. A quill stem with the maximum line above the top of the headset does not get ridden and does not get sold like that.

Threadless headsets and compression

The threadless idea, generically known as an Aheadset after the trade name that popularised it, changes what the parts do.

On a threaded setup, the top race does the preload and the stem is an independent part hanging inside the steerer. On a threadless setup, the stem is a structural part of the headset assembly. The stack from bottom to top is: lower bearing, head tube, upper bearing, compression ring, dust cover, any spacers, then the stem clamped around the steerer. The stem is what holds the stack together once it is set.

Preload is set separately from clamping, and the order matters enormously.

The compression system pulls the stack down. A top cap sits on the stem and a bolt runs down through it into something gripping the inside of the steerer, either a star nut hammered into the tube or, more usually on BMX, an expanding compression plug or wedge. Tightening the cap bolt pulls the top of the stack downward against the fork crown, which takes all the slack out of the bearings.

The stem pinch bolts then clamp the stem to the steerer and lock that setting in.

If you tighten the pinch bolts first and then the cap bolt, nothing happens, because the stem is already locked and cannot move down. This is the single most common front end setup mistake, and the symptom is a headset that will not stop knocking no matter how hard you tighten the cap. Loosen the pinch bolts, set preload with the cap, then tighten the pinch bolts.

Two other points that matter on a used bike.

The cap bolt is not a structural fastener. It sets preload and that is all. Once the pinch bolts are tight, the cap bolt could fall out and the stem would still be clamped to the steerer. It is not carrying your weight.

BMX often uses a compression plug rather than a star nut, because BMX steerers are frequently thicker walled and because a plug can be removed and reused where a star nut is a one-way trip. If a used fork has a star nut hammered down inside it and you want to run a plug, the star nut has to come out, which usually means driving it down and out of the bottom of the steerer.

Top load and front load stems

This is the choice riders actually make on a build, and it is worth being clear about what it does.

Bar height. The clamp on a top load stem sits above the steerer clamp, so the bars end up higher. On a front load stem the bar clamp sits ahead of the steerer clamp at roughly the same height, so the bars end up lower. That is a real difference in how the bike feels, and it stacks with the bar rise you have chosen and with how many headset spacers are under the stem. Riders who want their bars high often run a top load with a taller bar. Riders who want a lower front end run a front load.

Ease of bar swapping. A front load faceplate comes off with its bolts and the bars lift straight out forward. Grips, levers and the gyro upper cables can all stay where they are. A top load usually means sliding the bars out sideways through the clamp, which means stripping one end of the cockpit. If you change bars often, front load saves you time every single time.

Rigidity and looks. Riders argue about which is stiffer. What is genuinely true is that a front load has a joint (the faceplate) that a top load may not, and that faceplate has to be tightened evenly. Tighten a faceplate in a criss-cross pattern with an even gap top and bottom, not one bolt at a time until it stops. An uneven faceplate gap is a stress riser on the bars.

There is a third arrangement worth naming: some stems clamp the bars in a way that is neither cleanly top nor front load, and old school stems in particular vary a lot. Judge those by looking rather than by the category name.

The oversize bar clamp question is separate and it is a bar question rather than a stem question. Stems and bars have to agree on clamp diameter exactly, and more than one diameter exists. Do not assume, and do not force a bar into a clamp that seems close.

22.2mm is the number to know, and like the spindle it is a rounded name. The exact figure is 7/8 inch, 22.225mm. Nothing turns on the extra 0.025mm in practice, but if you put calipers on a bar and read 22.22, that is a standard bar and not a worn one.

Oversize at 25.4mm is real but niche. S&M and Eclat both offer it, so it is a live option rather than a dead one, but 22.2mm remains the default across the market and an oversize bar needs an oversize stem to match. Treat 25.4mm as a deliberate choice rather than something you might have by accident.

On stem stack height, the finding is the inconsistency rather than the number. Figures in circulation run roughly 30 to 33mm, but brands do not measure the same thing: some publish the steerer clamp height, some publish overall body thickness, and many omit it entirely. That means two stems quoting the same stack height may not stack the same, and comparing published figures between brands is not reliable. If stack height matters to your setup, measure the stem rather than trusting the listing.

One brand does state its method, which gives the rest a reference point. Eclat defines stack height explicitly as steerer clamping height, the vertical length of steerer the stem bore actually grips, and publishes it at 32mm. That is the useful definition, because it is the number that tells you whether you have enough steerer left.

So when a stem quotes stack height without saying what it measures, Eclat's definition is the one to compare against, and a figure noticeably larger than about 32mm is probably overall body thickness rather than clamping height. Confirmed

Steerer tube length and stack

A fork that is too short for a frame simply will not work: the stem clamp will hang off the top of the steerer with not enough tube inside it. A fork that is too long is normally fine, because the excess goes into spacers under the stem, and on a threadless setup a steerer can be cut down.

The measurement that actually matters is not the total length of the steerer but how much of it is left above the head tube once the headset is in. That is why the sum is worth doing rather than guessed:

Head tube length plus headset stack height (the amount the headset adds above and below the head tube, which differs a lot between integrated, zero stack and traditional pressed) gives you the height at which the steerer emerges. Everything above that is available for spacers and the stem clamp.

On a used fork, the steerer has usually already been cut for whatever frame it was on. A cut steerer cannot be uncut. So when buying a used fork, measure the steerer, work out what it will do in your frame, and do not rely on a listing that says "fits any BMX frame".

On a threaded fork, there is a second constraint: the threaded section has a finite length, and the top race and locknut have to land inside it. A threaded steerer that is long enough physically but has run out of thread is no more usable than one that is too short.

Fork offset

This is the one front end spec that most people ignore and then feel immediately when they change it.

Imagine the line of the steerer extended down to the ground. That line hits the ground somewhere in front of the tire contact patch, and the distance between the two points is trail. Trail is what makes a bike want to steer straight: turn the bars, and the contact patch drags behind the steering axis, generating a force that pulls the wheel back into line. More trail, more self-centring, more stability. Less trail, less self-centring, quicker steering.

Increasing fork offset moves the axle forward, which reduces trail. So a fork with more offset gives quicker, lighter steering. A fork with less offset gives more stable, heavier steering. That relationship is the whole reason offset is on a spec sheet.

The catch is that offset only means something in combination with head tube angle, and both are frame and fork properties that only interact once the parts are on the same bike. A fork that gave one bike lively steering can make another feel vague, because the head angles differ.

What you can do practically, when replacing a used fork, is measure the offset on the fork that is coming off and match it. The bike you already know is a better reference than a number off a chart.

Head tube angle is the other half of the trail calculation, and that figure is now sourced. Published geometry charts on named frames put street and park at 75.25 to 75.75 degrees, dirt at 74.5, and race at 74.0 to 74.5. Those come from BSD, Cult, Sunday, S&M and Chase, and the full table with the frames listed sits on the Frame ID and Geometry guide.

So the angle side of trail is settled. The offset side is not.

That is better than a two-to-one spread across the range, which makes offset one of the few BMX numbers where the difference is large enough to feel immediately. Short offset puts the axle closer to the steering axis, which increases trail and makes the front end slower to turn but easier to hold on a nose manual. Long offset does the reverse.

On how it is measured, the news is good. The brands that document their method measure offset the same way: the perpendicular distance from the steering axis to the axle centre. No brand was found using a different convention, which means published offset figures can be compared across brands. That is not something you can assume elsewhere in this guide, so it is worth saying where it holds.

Combined with the head tube angles above, that makes trail an actual calculation rather than a vibe.

Fork dropouts and axle sizing

On BMX forks, the dropouts are at the bottom of the legs and the slot usually opens downward or slightly rearward. Three things to look at on a used fork.

Slot width. Measure it. If a front hub's axle is larger than the slot, the options are a different hub, a different fork, or filing the dropouts, and filing dropouts on a fork is not something to do lightly.

Slot depth and the axle seating. The axle should sit right at the closed end of the slot with the wheel centred in the fork. A wheel that will not pull up to the end of the slot is either the wrong axle or has hardware in the way.

Dropout thickness and peg clearance. Front pegs bolt onto the axle outside the dropout, and how much axle protrudes past the dropout face determines whether a peg gets enough thread engagement. Thick dropouts eat axle length.

A unicrown fork, where the legs bend inward and meet the steerer directly with no separate machined crown, is the common modern construction and is also an identification marker: a fork with a separate brazed or welded crown is generally an older design. That is a dating cue rather than a fitment one, but it is useful when working out what era a mixed-up bike is from.

Fork dropout slot width is not published by anyone. Odyssey, Sunday, Eclat, Shadow Conspiracy and S&M were all audited and every one of them states axle compatibility, 3/8 inch or 10mm, without ever stating the slot dimension. The same holds for rear dropouts on frames, across nine brands. Two separate research passes have now come up empty, so the figure is not being withheld here, it does not appear to exist publicly. Measure the fork.

Three front axle sizes have been in use and none of them interchange. 5/16 inch on old school forks, then 3/8 inch and 10mm on everything since. As covered on the wheels page, 3/8 inch is 9.525mm and 10mm is 10mm, and no source treats them as the same size. On an old fork, expect 5/16 and expect a modern hub not to go in.

On peg thread engagement, nobody publishes a figure and that is worth saying plainly. Odyssey, Sunday, Cinema, BSD and Shadow Conspiracy were all checked, manuals and safety sheets, and none of them state a minimum number of threads or turns for a peg on a front axle. General fastener practice suggests engagement of at least one axle diameter, but that is engineering rule of thumb rather than anything a peg or hub maker has put their name to, and this page is not going to dress it up as a specification.

What that means practically: if you are running pegs and the axle looks short in the threads, the person who can tell you whether it is enough is the manufacturer, and they have not written it down. Err long, and ask before you hang your weight on it.

Headset play, wear, and checking a used front end

The check itself takes ten seconds and everyone should know it.

Stand beside the bike, pull the front brake on hard, and rock the bike back and forward while resting a finger across the gap between the head tube and the top of the headset. If you feel a knock and a tiny movement at that joint, you have play. If you feel a knock but the movement is somewhere else, look at the stem clamp, the hub bearings, or the brake itself before blaming the headset.

Then lift the front wheel off the ground and let the bars fall to one side. A healthy headset lets the bars swing freely and smoothly. If the bars fall and then stop in one particular position, or if turning the bars feels notchy, the bearings have indexed: the balls have hammered small dents into the races, usually from being run loose or from repeated hard landings, and they now want to sit in those dents. That is not adjustable away. It is a new headset, or on an integrated frame it may be a new bearing and a very careful look at the frame seat.

What causes it, in order of how often it turns up on used bikes:

Run loose. A headset with play does not just knock, it destroys itself, because every knock is a bearing hitting a race with the whole weight of the bike behind it. A headset ridden loose for a month is done.

Run dry. No grease, water in, corrosion in the races.

Overtightened preload. Less common, but a headset done up until the bars will not turn freely is loading the bearings continuously and will wear.

Bad installation. Cups pressed in crooked, a crown race not seated fully down, or a compression ring left out entirely. All of these show up as a headset that never feels right no matter how it is adjusted.

Alongside the headset, three things to check on any used fork.

Straightness. Sight down the fork from the front with the wheel out. The legs should be symmetric and the dropouts level with each other. Then look from the side: a fork that has taken a hard front impact bends backwards near the crown, and it is easier to see from the side than from the front.

The crown race seat. The machined step at the bottom of the steerer where the crown race sits should be clean and square. A chewed crown race seat, usually from someone levering an old race off badly, means the new race will not sit square.

Cracks at the welds. Look where the legs meet the crown or the steerer, and around the dropouts and any brake mounts. A crack in a fork is a hard no. If a fork is cracked, bent, or sketchy, it does not get sold.

Aheadset

Treat it as two words doing two jobs. When a current listing says Aheadset, read it as threadless, and everything under threadless headsets above applies unchanged.

When a mid school part says Dia-Compe Aheadset, the name is the point. Restorers building era correct bikes search for that name specifically, and a genuine period Dia-Compe unit is a different proposition to a modern threadless headset that happens to fit. That is worth knowing on both sides of a sale: it is what makes the part findable, and it is why the name is worth keeping in a listing rather than normalising it away.

This guide does not publish a date for when the design arrived, because no source consulted gave one that could be checked. The parts themselves date a bike better than a year range does.

Quill stems

The mechanism is worth understanding because it explains every quill stem problem. Everything is held by one bolt working through a wedge, so the bolt and the wedge are the parts that fail, and they fail where you cannot see them.

Four checks before buying or selling a used one. Does the expander bolt actually turn, or has it seized in the quill, which is common on anything that has sat outside. Are the wedge threads intact, because a stripped wedge means the stem cannot be tightened at all. Any hairline cracks around the bar clamp, which is the other high stress area. And look at the quill shaft above the minimum insertion mark for gouging or a bend, which tells you it has been run too high.

That minimum insertion line is not advisory. A quill run above it puts the wedge near the top of the steerer where there is far less material, and steerers do break there.

Unicrown forks

The alternative is a crowned fork, where the legs socket into a separate machined or cast crown that the steerer also fits into. Look at the shoulders where the legs meet the steerer. A visible flat block with defined edges is a crown. A continuous curve into a weld is unicrown.

For a buyer this is mostly an identification and era point rather than a fitment one, because it does not change what headset or stem you need. It does tell you roughly where a fork sits in the timeline, and it is one of the features worth noting when you are dating a bike from its parts rather than from a claimed year.

Questions

Steerer tube diameter

This is the first number to check on a used fork, because getting it wrong means the fork will not go into the frame or the stem will not clamp the fork. Measure across the outside of the steerer above the crown.

Steerer tube length

Measure from the crown race seat to the top of the tube. Then add up what it has to clear: head tube length, headset stack, your spacer stack, and the stem clamp height.

The steerer must sit slightly below the top of the stem, not flush with it. If it sits flush the compression cap bottoms out on the steerer and cannot pull the bearings together, so the headset stays loose no matter how far the bolt goes in.

Star nut

Its sprung arms bite into the inside wall of the steerer, so it only works in a steel tube. It must never go into a carbon steerer, and it is not needed on a fork with an internal thread cut into the steerer.

Compression cap

The order matters. Loosen the stem pinch bolts, tighten the cap bolt until the play disappears and the bars still turn freely, then tighten the pinch bolts. Do it the other way round and the cap is fighting a clamped stem.

Internal compression bolt

This is what most modern BMX forks use. It lets the steerer be welded and heat treated as one piece, and on some forks the bolt is hollow so a front brake cable can pass through it. If you buy a used fork with no bolt, you need the one that matches that fork, not a generic top cap.

Crown race

Split crown race ring

The split works because a cartridge bearing sits on it and the preload holds it centred. That also means it only works with cartridge bearings. A loose ball headset needs a continuous race with no gap in it. Many modern BMX forks skip the part entirely and have the race angle machined straight into the crown, so if a used fork appears to be missing one, look closely before ordering a replacement.

Headset bearing angle

Buy bearings by the frame's headset standard rather than by measuring the old ones, because a worn bearing can measure close to the wrong spec.

Headset stack height

Headset spacer

Whatever the stack, the stem has to end up fully clamped on the tube, and the top of the steerer has to finish just below the top of the stem so the compression cap has something to pull against.

Top load stem

Beyond height, the two feel different under load, and which one a used bike has tells you something about how it was set up. The clamp diameter has to match your bars, which is a separate question covered under oversize bar clamp.

Stem reach

It is the cheapest way to adjust how a frame fits you, and worth considering before deciding a used frame is the wrong size.

Torque specification

Undertightening is its own hazard: a bar that slips in the clamp or a stem that turns on the steerer will do it at the worst moment.

The compression cap figure is an order of magnitude below the others on purpose. It is setting bearing preload, not clamping anything.