Phase 4 Progress: 0%

Interactive Execution Guide

Phase 4: Leak-Down & Flywheel

The test that decides whether the flywheel or the crankcase ever has to come apart.

Do not pull the flywheel or split the case yet. The leak-down test decides whether either is necessary, and the saw is currently in exactly the right state to run it — cylinder on, both crank seals in place, everything else already off. The flywheel comes off for only three reasons: a leaking flywheel-side seal, replacing the ignition module, or splitting the case. None of those is established yet.

The decision tree

STEP 0  Piston check through the exhaust port   ← free, muffler already off
          ├─ scored / alu transfer ──► top end comes off; parts list changes entirely
          └─ clean ──► continue

STEPS 1-4  Leak-down gate                       ← cylinder is on; do these now
          ├─ HOLDS ──► no flywheel, no case. Clean, rebuild carb, reassemble. DONE.
          ├─ flywheel-side seal ──► STEP 5 (flywheel off) — case stays together
          ├─ clutch-side seal ──►   Phase 3 page (clutch)  — case stays together
          └─ case joint / rough bearing ──► STEP 6 (split the case)
Step 0 is already written up. It is Phase 1, Step 4 — flashlight through the exhaust port, looking for piston-skirt scoring, a ring melted into its groove, and aluminium transfer on the cylinder wall. If you have not done it, do that first. It costs nothing and it can change everything below.
A second heat path, found in the teardown video (2026-08-11). Two videos shot during the first teardown had never been reviewed. Keyframes pulled from them show the cylinder cooling fins packed solid with compacted sawdust — not dusty, blocked. On an air-cooled two-stroke the fins are the cooling system, and a blocked fin stack overheats the engine regardless of how the carburettor is jetted. So there are now two independent overheating mechanisms on this saw: unmetered air from the intake side, and obstructed cooling from the outside. Clean the fin stack completely before any tuning — otherwise a correctly-jetted engine will still run hot and you will chase the mixture forever.

UNVERIFIED The video does not answer the piston question. It was shot 2026-08-11, before the muffler came off, so no frame shows the exhaust port or the piston. Step 0 still has to be done by hand.

Phase 4A: The Leak-Down Gate

1. Crankcase Pressure & Vacuum Test

TOOL Husqvarna set for this saw: cover plate 502 71 38-01, spacer 502 71 40-01, pressure gauge 502 50 38-01, vacuum gauge 502 50 37-01. A hand vacuum/pressure pump with a 0–15 psi gauge plus shop-made blanking plates does the same job — record the exact rig you used. An improvised test is valid; an undocumented one is not.

  1. ACTION Blank the intake and exhaust ports with plates and gaskets. Fit an adapter at the spark plug hole.
  2. CRITICAL Plug the crankshaft through-hole first. The 3120 crank is shared with the 3120K cut-off saw, where that hole feeds lubricant to the sprocket bearing. Leave it open and you will chase a leak that is there by design. Electrical tape plus a hose clamp is the field method.
  3. ACTION Pressure to 7–8 psi (~0.5 bar). Watch the needle.
  4. ACTION Vacuum to 5–7 inHg. It must hold too. A case can pass pressure and fail vacuum — seal lips seat one way and lift the other. Skipping the vacuum half is the classic way to miss a bad seal.
  5. INSPECT Leaking? Soapy water around the crank seals, the case joint, and the cylinder base. Bubbles locate it.
TestApplyPass
Pressure7–8 psi / 0.5 barNeedle steady over 10–20 s. Common tolerance: no lower than ~5 psi after 20 s.
Vacuumsee note ↓Must hold with minimal decay. Take the figure and its tolerance from the same source — see the correction note below.

UNVERIFIED The pressure figures above are cross-manufacturer norms, not Husqvarna's own. Take the authoritative numbers from workshop manual 101 88 55-26, p.114 and treat the table only as a sanity band.

Correction — vacuum figure withdrawn (2026-08-14). An earlier version of this page said "apply 5–7 inHg" and "tolerance ≤0.3 bar loss in 20 s". Those came from two different sources and contradict each other: 5–7 inHg is only 0.17–0.24 bar, so a 0.3 bar loss is arithmetically impossible and every leaking seal would have passed. The figure is withdrawn rather than patched with a guess. Pull the vacuum value and its decay tolerance from p.114 of the workshop manual, from the same table, so the two are dimensionally consistent. Note that common two-stroke practice pulls a considerably deeper vacuum than 7 inHg — another reason not to trust the withdrawn pair. Caught by an independent review lane, not by me.

๐Ÿค– Consult F100 Copilot

2. Carburettor Pressure Test

  1. ACTION With the carburettor assembled, pressurise the fuel inlet to about 7 psi.
  2. INSPECT Falling pressure points at the inlet needle, the metering diaphragm, or the pump diaphragm.
  3. INSPECT With the carburettor wet, spray around both shafts. Bubbles there mean worn bushings.

TOOL Carburettor pressure test procedure: workshop manual p.100. Rebuild kit is 503 59 74-02 (Walbro K12-WG).


๐Ÿค– Consult F100 Copilot

3. Throttle & Choke Shaft Play

This is the real test for the scratched butterflies. Air leaks around the shaft, not past the plate face. A visibly scratched plate on a tight shaft is fine. A pristine plate on a sloppy shaft will run the engine lean and cook it. Plate scoring is largely cosmetic; shaft play is not.

  1. ACTION Grip the carburettor body. Push the throttle shaft perpendicular to its axis, both directions.
  2. INSPECT Any perceptible rock or click means worn bushings (502 10 31-01, two off). You are feeling for radial play — rotational free-play is normal.
  3. ACTION Repeat on the choke shaft.
  4. INSPECT Cross-check running: at idle, spray a little carburettor cleaner at each shaft end. An RPM change confirms the leak.
FindingWhat it means
Shaft tightClean and reuse both plates despite the scratches. A rebuild kit alone is the fix.
Shaft has playThe carburettor is the fault. A rebuild kit alone will not fix it — shafts and bushings, or a replacement carburettor.

๐Ÿค– Consult F100 Copilot

4. Muffler Flange Flatness

The muffler on this saw was fitted with a shop-made bare-steel gasket — no fibre or composite layer. Something drove somebody to cut one. The mating face is the suspect: it carries a white deposit ring and visible erosion around the oval port.

  1. ACTION Clean the mating face back to bare metal.
  2. INSPECT Lay it on a surface plate or float glass with engineer's blue.
  3. INSPECT Full contact — a new gasket cures it. Dished or eroded — the leak returns whatever gasket you fit.

TOOL The IPL lists no separate muffler-to-cylinder gasket in the MUFFLER group. The sealing gaskets live in CYLINDER PISTON (503 13 13-01, 503 13 46-01) and in the master set 503 14 25-01. Discard the shop-made plate.


๐Ÿค– Consult F100 Copilot
GATE. Do not go past this point until steps 1–4 are recorded with results. If the crankcase test passes, you are finished with disassembly — clean, rebuild the carburettor, and reassemble. Steps 5 and 6 exist only for the case where it fails.

Phase 4B: Flywheel — only if the gate sent you here

5. Flywheel Removal

WARNING FLYWHEEL NUT IS RIGHT-HAND THREAD — COUNTERCLOCKWISE TO LOOSEN.
The clutch is the left-hand one. Do not confuse them.

If the nut will not move under firm, steady pressure, stop and re-check rather than escalating force. Getting this backwards destroys the crankshaft threads.

  1. TOOL Remove the spark plug and fit piston stop 502 50 33-01 in the plug hole.
  2. ACTION Remove the two starter pawls.
  3. INSPECT The ignition module can stay in place if you are only removing the flywheel. Do not disturb the air gap unnecessarily.
  4. CRITICAL Loosen the nut only until it is flush with the end of the shaft — do not remove it. It catches the flywheel when the taper lets go, protecting the threads and your hands.
  5. TOOL Fit the puller: 502 50 26-01 with holding tool 502 51 49-01 and bolt set 502 51 54-01. Thread the two M5×25 bolts in evenly, then tighten the centre bolt while holding against rotation until the taper releases.
    UNVERIFIED The "approx. 10 turns" figure often quoted for this step comes from the 362XP/365/372XP manual, not the 3120's own Electrical System section. Thread depth is not transferable between puller sets — run the bolts only as deep as they need to seat squarely, and check they cannot bottom out.
  6. ACTION Remove the puller, nut, washer and flywheel.
  7. INSPECT Check the flywheel for cracks and check the Woodruff key 735 88 06-00. A sheared key shifts the timing and is a known failure mode on this saw.

Refitting

  1. ACTION Clean the tapers on both the flywheel and the shaft. It is a taper fit; contamination there causes slip.
  2. ACTION Locate the flywheel, turning it gently until the key mates with the shaft recess. Fit washer and nut.
  3. TOOL Set the ignition air gap to 0.30 mm with feeler gauge 502 51 34-02, measured only at the two lower pegs of the module.

UNVERIFIED Flywheel nut torque: the 3120 has its own torque diagram in the workshop manual, with values keyed positionally to an illustration. That figure is deliberately not reproduced here because it cannot be tied to a specific fastener from the manual's text alone. Read it off the Saw 3120 torque page. For scale only: the 372XP calls for 35 Nm and the 346XP 25–30 Nm — the 3120 is a larger engine, so do not assume either. Absent beats estimated.


๐Ÿค– Consult F100 Copilot

Phase 4C: Clutch & Crankcase — gated

6. Crank Seals & Case Split

SCOPE This step is deliberately a stub. It will be written out properly once your Step 1 result exists and says it is needed — writing a full case-splitting procedure for a case that may never be opened would be guesswork dressed as guidance.

What is established so far:

  • Clutch is left-hand thread — clockwise to loosen. Full procedure is on Phase 3.
  • Crankcase dismantling tool: 502 51 61-01.
  • Crank seals / bearings: 503 25 00-02.
  • Master gasket set for reassembly: 503 14 25-01.
  • SAFETY The workshop manual warns that the crankcase halves are hot when replacing crankshaft bearings — use protective gloves.

A flywheel-side or clutch-side seal can usually be replaced without splitting the case. Only a leaking case joint or a rough main bearing forces a split.


๐Ÿค– Consult F100 Copilot

Machine data for this saw

SpecValueSource
Build year2001CONFIRMED operator
Serial rangeearly — up to 20115099999CONFIRMED build year + muffler casting
Muffler503 83 65-02CONFIRMED casting read from photo
CarburettorWalbro WGCONFIRMED cast marking; WM says WG 6, IPL says WG10 / WG 7A
Ignition systemSEM AM 37WM 101 88 55-26
Ignition air gap0.30 mmWM
Spark plugRCJ 7Y · 503 23 51-08IPL + WM, gap 0.50 mm
Displacement119 cm³WM
Bore × stroke60.0 × 42.0 mmWM
Idle speed2,500 rpmWM
Clutch engagement3,300 rpmWM
Chain pitch / gauge.404" / 1.6 mmWM
Fuel / oil tank1.25 L / 0.7 LWM
Fuel mix50:1 UNVERIFIEDNot in any source I hold. The workshop manual states only "lubricate using two-stroke oil" for assembly and gives no fuel ratio; the IPL has none either. This figure came from a local "operator's manual" PDF that turned out to be a 1-page synthetic stub, now removed. 50:1 is Husqvarna's usual ratio and is very likely right — but confirm it against a genuine operator's manual before mixing fuel. Corrected 2026-08-14.
Max RPM — a genuine conflict in the sources. The 1997 workshop manual gives 11,500–12,500 rpm for the 3120XP. The 2018 IPL lists ignition and flywheel variants that cut out at 11,000 and 9,800 rpm; Husqvarna revised the limiter down across the production run. Your 2001 saw shipped with module 503 13 85-01, but the current replacement 595 31 26-01 cuts out at 9,800. Tune to whichever module is actually fitted — never to the 1997 figure.

TVP claim ledger

Truth Verification Protocol. Every claim carries one of four states and a brutal reality check naming the specific missing artifact. A claim is promoted to CONFIRMED only by a receipt — never by assertion.

ClaimStateBrutal reality check
Muffler is 503 83 65-02, early serial rangeCONFIRMEDCasting legible in IMG_0843, corroborated by 2001 build year
Carburettor is a Walbro WGCONFIRMED"Walbro" cast into the body, IMG_0875/0876
Flywheel nut is right-hand threadCONFIRMEDFour Husqvarna workshop-manual procedures agree; clutch is separately flagged left-hand
Intake tract contaminated with sawdustCONFIRMEDFilter holder caked inside and out, IMG_0860–0867
One carburettor mounting screw missingCLAIMEDPhotos show one short screw where the IPL specifies two — needs a physical count against the IPL group
Throttle/choke shaft bushings wornUNVERIFIEDNot measurable from a photograph. Step 3 of this page settles it
Piston and cylinder conditionUNVERIFIEDNo top-end images exist in any batch. Step 0 settles it
Muffler flange is warpedUNVERIFIEDDeposits and erosion visible; flatness needs a surface plate. Step 4
Crankcase integrityUNVERIFIEDNever tested. Step 1 is the whole point of this page
Leak-down pass figures quoted aboveUNVERIFIEDCross-manufacturer norms, not Husqvarna's. WM p.114 is authoritative
Flywheel nut torqueUNVERIFIED3120 torque diagram is positional; the figure is deliberately not reproduced here
Doc spec "Tillotson HS-300A carburettor"CONTRADICTEDThe part itself is cast "Walbro"; both the IPL and the WM say Walbro WG
Site plug listed as "NGK BPMR7A" aloneCLAIMEDNot wrong — BPMR7A is a widely-used cross-reference — but the OEM spec is Champion RCJ 7Y, part 503 23 51-08, and that number was missing. Both now shown
Site top end "506 29 42-71"UNVERIFIEDThis number does not appear in the IPL text retrieved, which lists cylinder assy 576 27 00-03 and piston assy 501 89 41-03 separately. 506 29 42-71 may be a kit or a superseded number — not disproven, just unlocated. Both are now shown
Vacuum test figure, first published versionCONTRADICTED"5–7 inHg" with "≤0.3 bar loss" is self-contradictory — 0.3 bar cannot be lost from 0.24 bar. Withdrawn 2026-08-14, caught by an independent review lane
Extensionless page links used hereCONFIRMEDFlagged in review as possibly 404-prone; measured live — /runbook-phase4 returns 200 with 0 redirects, .html returns 308. Extensionless is canonical on Cloudflare Pages. Finding rejected on evidence
"WASM 11D Physics" readout on phases 2–3CONTRADICTEDGenerated heat-flux and shear-stress figures with Math.random() and presented them in W/m² and MPa. Removed — no physics engine exists behind it