Diagnosing an AR-15 From the Evidence It Leaves — Brass, Primers, Bolt and Mount
Ejection at 1 to 2 o'clock means over-gassed; 7 to 9 means short-stroking. Brass smeared on the case mouth is bolt-over-base from a weak spring, and a zero that walks 2 inches after 200 rounds is almost always a mount at the wrong torque rather than an optic fault. Read the evidence, run one confirming test, then change one part.
A malfunctioning AR-15 leaves a written record. Ejection pattern tells you the gas system's energy balance. Case-head marks tell you timing. Primer condition tells you pressure and firing pin fit. Group-shift signature tells you which fastener is moving. Read those four and you will usually have the cause before you have opened a parts catalogue.
The alternative — swapping parts until it stops — works eventually and costs a lot. It also destroys the evidence, because once you have changed the buffer, the spring and the gas block in one session you have no idea which one it was and you will do it again next time.
Everything below is arranged the same way: symptom, the evidence that distinguishes it, ranked causes, the one test that confirms, and the fix with a rough cost band. Work the tests in order and change one thing at a time.
Ejection clock position: what it does and does not tell you
Stand behind the rifle, imagine a clock face on the ejection port, and note where brass consistently lands. Consistency matters more than the exact angle — a rifle throwing brass to 4 o'clock every time is healthier than one alternating between 2 and 7.
Clock position Reading Typical cause Distance brass travels Action 12 to 1 o'clock Severely over-gassed Oversized gas port, adjustable block wide open, or a suppressor on a carbine system 15 to 25 ft, often forward Reduce gas or add buffer weight — this beats the rifle up 1 to 2 o'clock Over-gassed Carbine gas on a 16in barrel with a light buffer, or a suppressed unregulated system 12 to 20 ft Heavier buffer, or adjustable gas block 2 to 3 o'clock Slightly over-gassed Common on factory carbines with H buffers 10 to 15 ft Acceptable; tune only if brass is being damaged 3 to 5 o'clock Correct Balanced system 8 to 12 ft Nothing 5 to 6 o'clock Slightly under-gassed Mid-length gas with a heavy buffer, or a dirty rifle 5 to 8 ft Watch it; clean and re-check before changing parts 6 to 8 o'clock Under-gassed Undersized port, gas leak at the block, weak ammunition, buffer too heavy 2 to 5 ft, sometimes onto the bench Diagnose properly — see the gas branch table 8 to 11 o'clock Short stroking Bolt not travelling far enough to clear the magazine or lock back Brass hits the deflector or falls in the port Stop shooting and diagnose before firing more Use this to place the rifle on the gas spectrum, not to make a diagnosis. It narrows the search; the tests in the next table close it. Two caveats that people get wrong constantly. First, ejection pattern is ammunition-dependent — the same rifle will throw steel-case 55gr and brass-case 77gr to different clock positions, so evaluate with the ammunition you actually use. Second, a brand new rifle with a stiff extractor spring and a dry bolt will read under-gassed for the first 200 rounds and then settle. Lubricate properly and shoot a box before you conclude anything.
The rifle also has to be clean enough to be honest. A carbon-choked bolt carrier and gummed-up gas rings will imitate an under-gassed rifle perfectly. Clean first. It is free and it fixes more of these than any part does.
Gas branch: symptom to cause to test to fix
Ranked by likelihood, not alphabetically. Work down each row's cause list in order — the first item is right more often than the rest combined.
Symptom Evidence it leaves Ranked causes Confirming test Fix and cost band Short stroke — bolt does not lock back on empty Brass at 8 to 11 o'clock or in the port; carrier stops short of the buffer 1. Gas leak at block or tube joint 2. Undersized gas port 3. Buffer too heavy 4. Broken or worn gas rings 5. Carbon-locked carrier Remove handguard, fire two rounds and look for carbon streaking at the block-to-barrel joint and around the roll pin Re-index and re-torque block, $0; new gas tube $15-30; drill port, gunsmith $60-120 Failure to extract — case stays in chamber, next round jams into it Extractor mark torn through the rim; case in chamber with a fresh round behind it 1. Weak or missing extractor spring insert 2. Rough or dirty chamber 3. Over-pressure ammunition 4. Broken extractor claw Pull the bolt, push the extractor with a punch — it should have firm, springy resistance and snap back Extractor spring, insert and O-ring kit $10-20; extractor $25-45; chamber polish, gunsmith $50-80 Failure to feed — round noses into the feed ramp Bullet tip deformed or scuffed; round jammed at an angle 1. Magazine feed lips or follower 2. Under-gassed, carrier moving too slowly 3. Feed ramp mismatch between barrel extension and upper Swap magazines. Genuinely — try three known-good ones before touching the rifle Magazine $12-20; ramp cut, gunsmith $80-150 Bolt over base — bolt rides over the case rim of the next round Case head scored or the round is pinned under the bolt 1. Weak or short buffer spring 2. Buffer too light for the gas 3. Magazine spring weak Measure buffer spring free length: carbine spec is roughly 10.5in to 11.25in new; rifle spec roughly 12.75in to 13.5in Buffer spring $10-25; heavier buffer $25-60 Brass smeared or scuffed on the case mouth Bright drag marks on the case mouth or shoulder; brass hitting the deflector hard 1. Over-gassed, ejecting before the case has cleared 2. Sharp edge on the ejection port or deflector 3. Suppressor with no gas regulation Fire five rounds with an adjustable block backed down two clicks; if smearing stops, that was it Adjustable gas block $60-140; deburr port, $0 Over-gassed — violent cycling, brass to 1 o'clock, harsh recoil impulse Case heads with swipe marks from the bolt opening early; primers flattened or cratered 1. Oversized gas port 2. Carbine gas on a long barrel 3. Suppressor fitted 4. Buffer too light Fit an H2 buffer and re-check ejection. If the pattern moves back to 4 o'clock, the diagnosis was buffer mass H2 or H3 buffer $30-60; adjustable gas block $60-140 Double feed Two rounds jammed in the receiver 1. Magazine 2. Bolt not fully returning to battery 3. Weak extractor releasing a case mid-cycle Load a single round in a known-good magazine and cycle by hand ten times Magazine $12-20; extractor kit $10-20 Fails only when suppressed Everything shifts one or two clock positions clockwise; heavy blowback at the charging handle 1. No gas regulation for the increased back pressure 2. Buffer too light for suppressed operation Shoot 20 rounds unsuppressed and note the clock position, then 20 suppressed and compare Adjustable gas block $60-140; suppressor-rated bolt carrier or heavier buffer $30-200 Do the confirming test before you buy anything in the fix column. Roughly half of these resolve at zero cost once the test tells you what is actually happening. The gas leak at the block is the single most under-diagnosed fault on this list, and it is nearly always the same thing: a set-screw block that has drifted off the port, or a clamp block that was never torqued. Pull the handguard and look. Carbon streaking on the barrel around the block is unmistakable once you have seen it — a grey-black fan spreading back from the joint.
The other one worth naming is the extractor spring insert. The little rubber or polymer insert inside the extractor spring is a consumable and it hardens and shrinks with heat. It costs a couple of dollars, it takes five minutes, and it fixes an enormous share of intermittent extraction faults on rifles past about 3,000 rounds. Replace it before you consider anything more expensive.
Reading case heads and primers
Pick up ten cases, lay them out primer-side up under a bright light, and compare against this. It takes two minutes and it tells you about pressure and timing that nothing else will.
What you see What it means Severity What to do Clean round firing pin dent, primer slightly rounded at edges Normal pressure, normal timing None Nothing Primer flattened flush with the case head, edges square High pressure, or a case sized to minimum headspace Watch Check the ammunition lot and chronograph if you can Cratered primer — raised ring around the firing pin dent Excess pressure, or an oversized firing pin hole in the bolt Watch Compare against a different ammunition lot before blaming the rifle Pierced primer, hole through the centre Excess pressure, or a firing pin with a damaged tip Stop Stop shooting that lot. Inspect the firing pin tip and bolt face Bright swipe or smear across the case head Case moving while still under pressure — bolt opening too soon, usually over-gassed Act Reduce gas or increase buffer mass Extractor mark torn through the rim Extraction under too much residual pressure, or a weak extractor letting go and re-grabbing Act Extractor spring kit, then re-check Ejector mark — small crescent gouge at the case head edge Over-pressure or over-gassed; the ejector is embossing brass into its own recess Act Reduce gas; if it persists with mild ammunition, check headspace Circumferential bright ring near the case head Incipient case head separation, usually reloaded brass sized too far back Stop Discard the lot. Do not fire more of it Soot down the case body, halfway or further Case not sealing the chamber — low pressure ammunition or an oversized chamber Watch Try a different lot; if it persists, have the chamber measured Anything in the Stop row means put that ammunition away and inspect the rifle before firing another round. Everything else is diagnostic information, not an emergency. One warning about cratered primers specifically, because they generate more false alarms than anything else in this table. Many AR-15 bolts have a firing pin hole large enough that mild cratering appears at completely normal pressures, particularly with soft-cupped commercial primers. Cratering on its own, with no other pressure sign, is usually the bolt rather than the load. Cratering plus a flattened primer plus a swipe mark is a real pressure story.
When the zero walks instead
Same diagnostic logic, different branch. A zero that moves has a signature, and the signature tells you which interface is failing.
Symptom Group-shift signature Ranked causes Confirming test Fix and cost band Zero shifts progressively over a range session Group creeps in one direction, 0.5in to 2in per 50 rounds, always the same way 1. Mount cross-bolt under-torqued 2. Optic screws under-torqued 3. Thread locker never cured Mark screw heads and mount with a paint pen, shoot 50, look for line offset Re-torque with an inch-pound driver, $0; torque driver $30-70 Zero shifts after removing and refitting the optic Consistent 1in to 4in offset at 100 yd, repeatable 1. Mount is not a genuine return-to-zero design 2. Rail slot indexing changed 3. Lever tension not set Remove and refit three times, shooting three rounds each. Measure the spread of the three group centres Set lever tension per maker, $0; quality QD mount $90-250 Zero shifts between range trips but holds during a session No creep within a session; a step change between them 1. Handguard or rail movement against the barrel nut 2. Optic stored under pressure in a case 3. Temperature-driven shift in a cheap mount Grip the handguard and try to twist it against the receiver. Any perceptible movement is the answer Barrel nut re-torque, gunsmith $60-100; handguard hardware $10-30 Sudden large shift, 6in or more One-off jump, then stable at the new position 1. Optic screw sheared or backed out 2. Footprint or recoil lug mismatch letting the optic slip 3. Impact to the optic Remove the optic and inspect the screws and the recoil lug seats for bright wear or deformation Screws $2-10; correct-footprint plate $30-80 Dot dim, flickering or dead No group shift; the aiming point itself misbehaves 1. Battery contact fouled 2. Battery at end of life 3. Emitter failure Clean the contacts with isopropyl, fit a known-fresh battery, and leave it on for 24 hours Battery $3-8; warranty claim if the emitter is dead Group opens up with no directional shift Group size doubles; centre stays put 1. Barrel nut or muzzle device loose 2. Ammunition lot change 3. Barrel heat with a thin profile Fire five-round groups with two-minute cool periods and compare against rapid strings Re-torque muzzle device, $0; barrel nut, gunsmith $60-100 Point of impact differs between magazines or supported positions Consistent shift depending on how the rifle is held 1. Handguard contacting the barrel 2. Bipod loading against a non-free-float handguard 3. Sling tension on the barrel Slip a strip of paper between barrel and handguard and run it the full length Free-float handguard $80-250; remove barrel contact, $0 Start every one of these by checking torque with an actual inch-pound driver before you suspect the optic. The mount is wrong far more often than the optic is. The progressive-creep row is worth dwelling on because it is so commonly misread as an optic defect. An under-torqued mount does not slip all at once — it creeps a fraction of a millimetre per recoil impulse, which prints as a group centre marching steadily in one direction. People shoot a group, adjust, shoot another, adjust again, and conclude the turrets are not tracking. The turrets are fine. The mount is walking.
The paint-pen test settles it in one range trip and costs nothing. A single line drawn across the screw head onto the optic body, and another across the mount onto the rail. If the lines are still aligned after 100 rounds and the group still moved, then and only then start suspecting the optic.
Fastener torque, the whole rifle
Half the faults above are a torque value nobody applied. These are the standard figures; where the component maker publishes a different one, theirs wins.
Fastener Torque Thread locker Staked? Notes Barrel nut 30 to 80 ft-lb None — use grease on the threads No Torque to 30 minimum, then continue only as far as needed to align the gas tube hole. Never exceed 80 Castle nut 38 to 42 ft-lb None Yes, into the end plate Stake in two places. An unstaked castle nut backs off and takes your buffer retainer with it Muzzle device 15 to 20 ft-lb None with a crush washer No Time it with a crush or peel washer, never by over-torquing Gas key screws 35 to 40 in-lb High-strength Yes, mandatory Unstaked gas key screws are the most common cause of intermittent short-stroking in home builds Gas block set screws 25 to 35 in-lb Medium (blue) Optional dimple Dimple the barrel if the block is set-screw type. Set screws on a round barrel drift Gas block clamp screws 25 to 35 in-lb Medium (blue) No Alternate sides, two passes Handguard mounting hardware 25 to 35 in-lb Medium (blue) No Follow the maker's pattern and sequence if they publish one Pistol grip screw 30 to 40 in-lb Medium (blue) No Over-torquing cracks polymer grips and strips the lower's threads Optic mount to Picatinny (cross-bolt) 45 to 65 in-lb None on QD levers No Quick-detach levers are set by tension adjustment, not torque — follow the maker Scope ring caps 15 to 18 in-lb None No Alternate diagonally in three passes; even gaps both sides Red dot to mount or plate 10 to 15 in-lb Medium (blue) No Use the optic maker's published figure where they give one, e.g. 15 in-lb for Trijicon RMR, 12 in-lb for Leupold DeltaPoint Pro Buffer retainer and detent Hand tight under castle nut None No Check it is captured before staking the castle nut Buy an inch-pound driver and a foot-pound wrench and use them. Feel is not a torque spec, and the two most expensive failures on this list — stripped receiver threads and a sheared optic screw — both come from doing it by hand. One sequencing point that catches builders out. Torque the barrel nut and align the gas tube before you fit the handguard, and check that the gas tube passes through the barrel nut and into the carrier key without any rubbing. A gas tube that contacts the barrel nut hole will wear through and give you an intermittent gas leak six months later that will read exactly like a bad gas block.
Gas port diameters for reference
If you are chasing a gas problem to its source, the port is often the answer, and knowing the normal band tells you whether yours is out of family. Measure with pin gauges through the barrel, not with a drill bit.
Chambering and gas system Typical barrel Common port band Runs rich above Runs lean below 5.56 pistol-length 7.5in to 11.5in 0.0700in to 0.0860in 0.0900in 0.0680in 5.56 carbine-length 14.5in to 16in 0.0625in to 0.0700in 0.0760in 0.0600in 5.56 mid-length 16in to 18in 0.0700in to 0.0785in 0.0830in 0.0670in 5.56 rifle-length 20in 0.0890in to 0.0960in 0.1000in 0.0860in .300 BLK pistol-length 8in to 10.5in 0.0860in to 0.0930in 0.0980in 0.0820in .300 BLK carbine-length 16in 0.0700in to 0.0780in 0.0820in 0.0670in 7.62x39 carbine-length 16in 0.0930in to 0.1015in 0.1060in 0.0900in Use this to decide whether the port is the problem before you spend money on buffers. If the port measures inside the band and the rifle still short-strokes, the fault is downstream — leak, rings or spring. Opening a gas port is a one-way operation. You cannot put metal back, and a barrel drilled 0.010in too far is now permanently over-gassed and needs an adjustable block to be usable. If you are going to do it, go up in increments of 0.003in to 0.005in, shoot a magazine between each, and stop the moment the bolt locks back reliably with the weakest ammunition you intend to use.
The far better first move on an under-gassed rifle is an adjustable gas block, which costs about the same as one visit to a gunsmith, is reversible, and lets you tune for suppressed and unsuppressed without touching the barrel.
Maintenance intervals by round count
Most of the faults on this page are wear items announcing themselves. Replacing them on a schedule is far cheaper than diagnosing them mid-match.
Component Inspect at Replace at How you know it is due Cost Bolt gas rings 3,000 rds 5,000 rds Stand the bolt on its face on a bench — if the carrier collapses under its own weight, they are gone $5-12 Extractor spring, insert and O-ring 2,000 rds 3,000 to 5,000 rds Extractor feels soft under a punch; extraction becomes intermittent when hot $10-20 Ejector spring 5,000 rds 10,000 rds Ejection distance drops with no other change $5-10 Buffer spring 5,000 rds 10,000 rds or when short Free length below roughly 10.5in carbine or 12.75in rifle $10-25 Cam pin 5,000 rds 10,000 to 15,000 rds Visible galling or a flat worn on the bearing surface $10-20 Firing pin 5,000 rds On damage Chipped or mushroomed tip; light strikes $15-30 Complete bolt 5,000 rds 10,000 to 15,000 rds Cracks at the cam pin hole or between the locking lugs — inspect with magnification $70-160 Gas tube 10,000 rds On erosion or wear Elongated or worn where it enters the carrier key $15-30 Barrel (match accuracy) 5,000 rds 6,000 to 10,000 rds Group size grows with no other cause; throat erosion measurable $180-600 Barrel (service accuracy) 10,000 rds 15,000 to 20,000 rds Keyholing, or accuracy outside your acceptance standard $180-600 Optic mount torque check 500 rds after install Every 1,000 to 2,000 rds Paint-pen witness lines out of alignment $0 Log your round count somewhere. A notebook in the range bag is enough, and it converts every entry in this table from guesswork into a date. If you only adopt two habits from this table, make them the gas ring check and the extractor spring replacement. Between them they account for a very large share of the intermittent faults that get diagnosed as something expensive, and together they cost under $30 and about fifteen minutes.
The order to work in
Discipline beats knowledge here. A methodical owner with a mediocre understanding of gas systems will fix the rifle faster than an expert who changes three things at once.
Clean and lubricate first, properly, and shoot a magazine. Then swap magazines — three known-good ones. Then change ammunition to a different lot or brand. Only then start on the rifle, and when you do, change exactly one component and shoot enough rounds to be confident before you touch anything else. Twenty rounds is not enough to confirm an intermittent fault; sixty is closer.
Write it down as you go. Date, round count, what you changed, what the ejection pattern was before and after. Six months later, when it comes back, that notebook is worth more than any diagnostic chart including this one.