The casting comes back with metal fins across a flat face, and a few grit pockets next to them. The first reaction is to look at prime coat viscosity, and the viscosity is fine, because it is checked twice a shift and recorded. So the investigation moves to the slurry, then to the sand, then to the kiln.
Nobody walks over and looks at the pattern cluster sitting on the bench waiting to be dipped. That is where the answer usually is.
The chain nobody connects
Wax is not wettable by a water-based slurry. That is not a defect, it is a material property. The whole point of the cleaning step is to change the surface of the pattern so that silica sol will spread across it instead of pulling away.
When that step is skipped or degraded, the failure is not subtle and it is not local. It runs like this:
Pattern not properly degreased → prime coat fails to wet the surface → thin or absent prime coat → exposed refractory and metal penetration at pour.
The defects that show up on inspection are metal fins, grit inclusions and pinholes — three different names for one upstream event. And because they appear in a scatter across the face rather than in a pattern, they get diagnosed as three separate problems, each with its own corrective action, none of which addresses the bench.
The step itself
The whole operation takes about five seconds per cluster. That is the first reason it gets shortened to three.
| Parameter | Value |
|---|---|
| Cleaner | ZF-301 or equivalent |
| Bath temperature | 24 ± 3 °C (room temperature) |
| Minimum standing time after assembly | 45 minutes |
| Dip | Reciprocate 3 times, about 5 s total |
| After | Blow dry |
The 45-minute minimum is not about the cleaner. A cluster straight off the assembly bench still has warm wax and a warm joint at every gate. Dipping it early means the bath is doing its job on a moving surface, and the wetting agent has less time to act before the wax cools and the surface film sets. Clusters that queue at the dip tank by design — because the bench feeds faster than the tank drains — are the ones that fail the wettability check intermittently, and intermittent failures are the hardest to trace.
Three reciprocations is not a ritual. Wax surfaces hold air in fine detail, especially at gate roots and in lettering or part numbers. A single dip and pull leaves those pockets dry. The reciprocation is what displaces the air.
The check that makes the whole thing verifiable
Here is the part worth adopting even if none of the rest of this matches your shop. There is a pass/fail test, it takes under a minute, and it does not need a lab.
Take a cleaned cluster and dip it into silica sol containing 0.5% wetting agent. If the slurry wets the entire surface — no breaks, no beads, no dry patches — the batch passes. If it does not, the entire batch goes back through the tank. Not the cluster that failed. The batch.
This is the correct granularity. Cleaning failure is a bath condition, not a cluster condition. A cluster that fails the test is evidence about the bath, and every cluster that went through that bath since the last change is suspect.
Bath life, and what “contaminated” looks like
The working life is roughly 40 days or 6,000–8,000 clusters, whichever comes first. Wax carry-over, shell dust from the bench and general grit load the bath continuously, and there is no visual cue that tells you it has crossed the line.
Recovery when the bath is loaded:
- Stop using it. Let it stand 24 hours.
- Draw off the top two-thirds.
- Top up with fresh solution.
What you do not do is run it to failure and then double the concentration to compensate. Over-strength cleaner leaves its own film, and a film between wax and prime coat is the same outcome as no cleaning at all.
The same job, done twice
The prime slurry also contains wetting agent — 2.5 ml per kg of silica sol, added before the zircon flour and never after, because the addition order in the prime coat is silica sol, wetting agent, zircon flour, defoamer, and reversing it produces a mix that will not wet anything.
So the cluster gets a wetting agent in the tank and another one in the slurry. That is not redundancy for its own sake. The tank treatment removes the contamination; the slurry additive handles the surface energy of the fresh silica sol. Skip the tank and the slurry additive has to fight a contaminated surface with a fraction of the dose. It will not win, and the operator will conclude that the wetting agent is bad and order a different brand.
Where cleaning bleeds into the next layer
Two disciplines sit either side of this step and fail in the same way.
Never dip a cluster that has already been stuccoed into the prime slurry. This one is absolute and it is written that way in the work rules. A cluster that goes back into the prime tank after its first stucco coats the bath with sand, and the sand does not settle out fast enough to save the batch. Everything dipped after it inherits the contamination.
Blow the loose sand off before the next coat. Not brush it, blow it. Residual stucco between the prime and intermediate layers is one of the standard routes to primary-to-secondary layer separation, and that failure appears later as rat tails, scabs and buckled shell on flat faces. By the time those show up, the cleaning bench is three processes behind the investigation.
Why small shops get away with it and larger ones do not
In a shop running one shift with a handful of people, the person who dips the clusters is often the person who assembled them hours earlier, and the queue naturally builds up past 45 minutes. The bath gets changed because someone notices it looks tired.
Add a second shift and a target, and the queue compresses, the bath ages past its life, and three defects appear that were never in the scrap report before. The process did not degrade. The margin that was absorbing the missing discipline went away.
What to put on the wall next to the tank
- 45 minutes minimum standing after assembly. Write the assembly time on the board, not on the cluster.
- Three reciprocations, five seconds, timed once so people know what five seconds feels like.
- Bath change date and cluster count, both, on the same label.
- The 0.5% wetting agent test, run daily on the first cluster of the shift, result recorded.
- No cluster goes back into the prime tank after stucco. No exceptions, no “just this one”.
None of that is expensive. The test kit costs less than a rejected casting, and it is the only item on the list that turns a judgement call into a measurement.