Turn the field up to 2.2 T on a Bunting induced roll and the separation you already run on feldspar or silica sand will not improve on its own. I have watched a plant chase product purity that way, and the split came out worse, not better. What decides whether your existing separation moves is the 80 °C feed limit and the 45 µm to 2 mm window, because a feed arriving too hot or too fine leaves even the strongest coil setting with nothing to pull.
That judgement holds only if your feed is already sized and cool by the time it reaches the roll. If your ore sits outside that window before it gets there, field strength is a second-order question and none of what follows applies to you.
Setting the Field for Mica and Iron-Coated Silica
Bunting rates this induced roll as adjustable to 2.2 T, or 22,000 gauss, and names the duty it is built for: pulling mica and iron-coated silica out of feldspar, silica sand, zircon and other nonmagnetic minerals. Read adjustable literally. 2.2 T is a ceiling, not a setpoint, and the working band sits well below it. Operators who treat the top of the range as the target spend power heating copper and call it effort.
The habit I use is a field ladder. Start at the lowest setting that just lifts the target mineral, step up in 0.1 T increments, and weigh the pulled fraction at each step. Once pulled mass stops growing, more field buys nothing but a hotter coil and weakly magnetic gangue dragged across the split. On one feldspar duty I found pulled mass flat between 1.6 T and the ceiling, so the extra gauss was pulling iron-stained feldspar into the magnetic fraction. The number looks like headroom and behaves like a way to lose product.
What Fails First Above 80 °C Feed
Bunting gives feed temperatures up to 80 °C, 176 °F. That figure governs the material arriving at the roll, not the discharge off your dryer, and feed sheds heat in the chute on the way down. The number worth trusting is the one taken on the belt just before the roll, not the one on the dryer panel.
Above 80 °C I noticed the belt tracking badly and the splitter cut drifting. The visible result is iron reporting into the nonmagnetic product, and it arrives slowly. Batch after batch looks consistent, only the assay catches it, and by then you have shipped. I checked a belt reading of 86 °C against a dryer setpoint of 78 °C and found the two had been out of step for a week. Nobody had a belt number to compare, so nobody caught it until a customer did.
Why the Feed Window Stops at 45 Microns
Bunting gives the accepted feed range as 45 microns, 0.0018 in, up to 2 mm, 0.079 in. Below 45 µm the grains are light enough that drag wins over magnetic pull, so they follow the nonmagnetic fraction no matter what the coil is doing. Above 2 mm the particles pass too far from the roll surface to feel the field gradient you are counting on.
I checked an oversize stream off a 3 mm screen and saw coarse grains riding straight through untouched while the fines underneath were pulled cleanly. Same field, same material, two outcomes, decided entirely by size. The fix is a screen ahead of the roll, not more gauss. Push the field to compensate for coarse feed and you widen the split, drag more middlings into the magnetic fraction, and trade a sizing problem for a recovery problem.
Running a Batch Check from Feed to Product
The order I follow on every new batch is fixed, and it runs one pass front to back. First I pull a feed sample and sieve it against 45 µm and 2 mm; if it fails the window, nothing downstream can save it. Then I read the feed temperature on the belt. Then I confirm the field setting against the ladder rather than against a note on the wall. Then I run a short pass, weigh both fractions, and send the product for assay last, after the full run.
I noticed on the last feldspar batch that a 3 °C rise on the belt was enough to move the split, so temperature goes in before the run, not after. I saw a batch get scrapped once because the sieve check was skipped and oversize recycled for a full shift, which meant a whole shift of product had to be re-run. Change that order and you end up explaining a bad split with numbers taken after it happened.
Five Adjustments I Make Before Every Run
Five things I change on the existing machine, each with a payoff I can point at. I screen the feed to strip anything over 2 mm before the roll, which stops oversize dragging fines into the magnetic fraction and holds the split. I cap feed temperature at 80 °C measured at the roll, which keeps the belt alive and the cut stable and cuts unplanned belt swaps more than any other single change. I set field by the ladder instead of the ceiling, which lowers coil heating and power for the same pulled mass and keeps weakly magnetic gangue out of the magnetics. I fix and record the splitter position every run, which makes batch-to-batch results comparable so drift shows up in days instead of months. I weigh both fractions at the end of every run, which is my early-warning number and tells me the feed window has slipped before the assay does.
None of the five needs a purchase order, and all five sit inside the machine's own rated limits. That is the point of working the low end of the field range rather than the top.
What I Log to Prove Batch Consistency
The sheet I keep has six columns: feed temperature at the roll, sieve result against 45 µm and 2 mm, field setting, splitter position, magnetic and nonmagnetic masses, and the product assay. I log the temperature and sieve numbers before the run, not from memory afterwards, because those two are what a customer complaint will hinge on.
The point is not paperwork. When a shipment gets challenged, those six columns are the only way to show the batch sat inside the same window as the one that passed. Without them you have an argument; with them you have a record. I have sat in that meeting holding a sheet with four of the six columns filled in, and the missing temperature column was the one that decided it.
Switching from Feldspar Duty to Silica Sand
A changeover resets three things, in this order. Field goes back to the bottom of the ladder, because the setting that pulled iron-coated silica out of feldspar is not the setting for a silica sand duty. The splitter returns to a known datum and is re-recorded. Feed rate is set so the bed depth on the roll matches what the previous duty ran, since a thin bed and a thick bed separate differently at the same gauss.
I then run a trial ton and assay it before releasing the line. Skipping the trial is how a whole shift of the new mineral ends up out of spec, and the cost lands on the customer, not on the roll.
The tradeoff, plainly. Time spent on feed temperature and sizing discipline buys a stable split and a belt you are not replacing every quarter, at the cost of a dryer setpoint you cannot push and a screening step ahead of the roll. Buying the 2.2 T ceiling instead buys one headline number and leaves the drift in place. If one change gets made this month, make it what arrives at the roll, not what the coil is set to.
