Two line items are competing for the same money in a scrap-handling upgrade: a separator sized to pull ferrous out of a mixed chip and offcut stream, or six weeks spent auditing what that stream actually contains. The equipment quote arrives first and is easy to defend in a capital meeting. The audit looks like delay. That fork is usually resolved by one observation about how a shop buys metal rather than by separator theory. On a mainstream online metal storefront, carbon steel and alloy steel sit in the same material selector as aluminum, brass, bronze, beryllium copper and copper. A shop buying from that selector every week is not running a magnetic stream with a little contamination in it; it is running a mixed stream that happens to contain steel.
Is your metal stream actually ferrous?
The asymmetry is what makes the fork feel unfair. Belt width, working gap, suspension height and throughput can all be quoted, tabulated and defended. The ferrous mass fraction of a stream cannot be quoted until someone pulls months of material records and works through a sample. Because of that, the separator decision tends to be made on the shape of the catalog rather than the shape of the stream. The same storefront that lists carbon steel and alloy steel also stocks bearings, shafts and gears, machine-ready stock, and brass and bronze bar to ASTM B505, because those grades are bought for machinability, wear behavior and pressure tightness. The separator decision is therefore a material-selection decision that was already made, months earlier, by whoever specified the part.
Ask the blunt question first: is the material in front of the magnet ferrous at all? The selection guides that govern shop buying answer it by omission. Copper C11000 is chosen where electrical or thermal conductivity is the requirement, at roughly 101% IACS. Brass C36000 is chosen where the job is high-speed machining at low cost per part. Bronze C93200 is chosen where the part must resist friction and wear. The ASTM B505 bronze and brass family pushes the same logic further: leaded gunmetals such as C83600 and C84400 sit near the top of the machinability scale, manganese bronzes C86300 and C86500 trade machinability for strength and wear resistance, and phosphor bronze C90700 serves heavy-duty gears and bearings. None of those criteria involve magnetic response, because none of them ever needed to.
The consequence is arithmetical rather than mechanical. A separator recovers the ferrous share of a stream and nothing else; every aluminum, brass, bronze, copper and beryllium copper particle that reaches the magnet slips past and keeps contaminating whatever sits downstream. Aluminum sheet catalogued from roughly 0.2mm to 6.0mm thick, in alloys from 1050 and 1060 through 3003, 5052 and 6061, arrives as trimmings and punch scrap light enough to travel with the air and coolant flow. The buying question therefore shifts from what the separator can capture to what fraction of the stream is even available to be captured. The next input that matters is not the alloy name but the size and shape of the ferrous pieces themselves.
Does part size decide whether a magnet can lift it?
Size and shape decide more about capture than alloy grade does, and the layout of a normal metal order explains why. Sheet stock is catalogued in a narrow thickness band, from roughly 0.2mm up to 6.0mm, while the same order can include plate, bar and machine-ready stock. Turnings from a lathe are small and curled, saw swarf is finer, punch slugs and offcuts from sheet are flat and light, and a short bar end is heavy and compact. A separator sized for one of those forms will underperform on the others, because holding force scales with ferromagnetic mass and the surface presented to the field. Sizing therefore starts with two measured numbers from the shop's own stream: the smallest particle that must be caught and the heaviest piece that must be held.
Condition state complicates the same calculation. Aluminum temper guides make the point in the language of forming: alloys such as 5052 and 6061 are selected by strength, weldability and corrosion resistance, but H14, H32 and T6 tempers decide whether the sheet bends cleanly, springs back, cracks, or finishes poorly. Temper is often the half of the specification that actually governs shop behavior, and it governs the stream the same way, because hardness and ductility decide whether a machining operation produces a fine powdery swarf or a curled turning, and whether a sheared edge leaves a burr that breaks off later. A separator specification written only in alloy names, 6061 or C36000 or C93200, names the material family and says nothing about the form the separator must handle.
Restated as a purchase requirement, the sizing line reads: catch everything down to the smallest ferrous particle the downstream equipment cannot tolerate, and hold the heaviest ferrous piece the flow can carry. Both numbers come from the stream, not from the separator catalog. That is why a specification beginning with magnet strength tends to end in rework; the field has to be strong enough at working distance for the smallest particles, and the belt or drum has to clear the largest pieces without blinding. Once those two figures are fixed, a second and less comfortable question appears. If only part of the stream is ferrous to begin with, what does the equipment actually achieve on the mixed material a shop genuinely sends through it?
Does a mixed stream defeat a stronger magnet?
A stronger magnet does not answer that question, and the reason sits in how the non-ferrous half was chosen. Machining guides frame brass and bronze selection as a production decision about what happens once the cutter meets the material: tool wear, cycle time, surface finish, scrap risk and cost per part. Brass earns its place in high-volume precision work with thin walls and fine cosmetic finishes because it cuts easily, runs faster cycles and wears tools less. Bronze earns its place in bearings, bushings and load-bearing or corrosion-prone components because it resists wear. Neither reason involves magnetic permeability. Those grades are in the stream precisely because they are good at their jobs, and being good at machining, wear resistance or conductivity is what keeps them out of magnetic recovery.
The property comparisons make the dilution effect concrete. Copper C11000 is selected for conductivity at roughly 101% IACS, brass C36000 for high-speed machining and cost efficiency, bronze C93200 for anti-friction behavior and wear resistance. Within the ASTM B505 family the same logic holds at different points on the scale: leaded gunmetals C83600 and C84400 rate at the top for machinability, manganese bronzes C86300 and C86500 lower, and phosphor bronze C90700 in between. Every one of those ratings describes how the material behaves under a cutting tool, not under a magnetic field. These grades are also hard to tell apart by eye in a scrap bin, which is why mixed containers survive longer than expected.
The practical dilution source in a mixed shop is not exotic contamination; it is the standard stock the shop buys on purpose. A bearing change-out generates C93200 bronze bushings. A batch of turned fittings produces C36000 brass chips by the kilogram, and cable or busbar work leaves C11000 copper offcuts. All of it lands in the same container as the ferrous turnings, and all of it passes a magnetic separator untouched, which inflates the false-negative rate the shop later blames on the equipment. That is where the argument stops being about machinery. If recovery is bounded by a fraction nobody has measured, the next useful step is not a stronger magnet but a defensible number, and numbers come from documents.
Which documents prove what is in the stream?
Start with the paperwork the supply side already treats as a feature. Metal distributors publish an ISO-backed quality page, mill test reports for the material they ship, and reference tables for hardness conversion, thickness conversion, fraction conversion and melting points, because buyers already rely on them as procurement evidence. A mill test report ties a heat or lot to a chemistry and a standard, which lets a shop reconstruct what was bought in a given period, and a thickness or hardness table converts a purchase record into the physical sizes a separator has to handle. Documentation is unglamorous and slow to assemble, but it is the cheapest instrument available for measuring a stream, and it is already being collected.
The audit is a reconciliation rather than a research project. Pull purchase records for the last several months, group them by material family, mark which families are ferrous and which are not, and compare that non-ferrous list against what the shop actually buys. Aluminum sheet in 1050 to 6061 grades, brass C36000, bronzes to ASTM B505, copper C11000 and any beryllium copper tooling stock belong on the non-ferrous side whatever volume they occupy. The audit has to output three things: the ferrous mass fraction of the stream by weight, the size distribution of ferrous particles at the capture point, and a short list of the non-ferrous families that will pass through. Without those three numbers, a separator quote is an opinion with a price attached.
One condition makes the audit trustworthy before it becomes expensive. Where purchase records are incomplete, the same logic that governs material buying applies: test a small sample before committing to a production lot. Suppliers already sell sample packs and prototype boxes so a shop can try a material without buying a pallet of it, and a will-call counter makes trials cheap enough to repeat. The separator equivalent is a weighed sample trial on a few hundred kilograms of real stream material, run before and after separation, rather than a full-line specification written from a catalog description. A trial that produces one defensible number on ferrous recovery beats a quotation carrying three plausible capacity figures.
Should the audit or the equipment quote come first?
The rule is ordered, and the order is the whole point. Audit the stream first: ferrous fraction of the stream by weight, the smallest ferrous particle the capture stage cannot miss, and the non-ferrous families that will pass through regardless. Size the separator second, against those figures, using field strength at working distance for the smallest particle, belt or drum capacity and working gap for the heaviest one, and a documented expectation for recovery once the non-ferrous share is subtracted. The material evidence drives the order. An ASTM B505 grade list and a sheet thickness band from roughly 0.2mm to 6.0mm describe a stream far better than any separator datasheet describes a shop. Nothing in a magnet catalogue changes that sequence.
There is one place where the rule stops applying. If the separator is being bought to recover non-ferrous value rather than to protect downstream equipment from ferrous contamination, the argument changes completely, because the materials that dominate such a stream were selected for properties a magnet cannot exploit: conductivity in C11000 copper, anti-friction behavior in C93200 bronze, machinability in C36000 brass. No increase in field strength recovers those grades, and the same holds for aluminum alloys from 1050 through 6061, whatever the temper. The one exception worth flagging to a supplier is austenitic stainless behavior, which depends on the specific grade and condition and should be settled with trial data rather than assumed. Outside those cases, the audit-then-size order holds.
That leaves the purchase with a one-line test. A separator should not be specified until the shop can state, in a single sentence, what fraction of its stream is ferrous by weight and what the smallest ferrous particle at the capture point measures. If either figure is missing, the separator is the wrong first purchase and the audit is the right one. If both figures exist and the smallest particle is finer than the chosen model can hold at working distance, the audit has already answered the sizing question and the quote can be rewritten around it. The first move this week is not a magnet enquiry; it is reading last quarter's material purchases and marking which families on that list a magnet has never touched.
The two line items are still competing for the same money, and the way to choose between them has not changed: the audit is the engineering deliverable, and the equipment quote is its output rather than its starting point. A shop that buys carbon and alloy steel from a selector also selling aluminum, brass, bronze and beryllium copper is running a mixed stream, and the ferrous fraction of that stream sets what any separator can return. Where the goal is protecting downstream equipment, paperwork comes first and sizing second. Where the goal is recovering non-ferrous value, the stream is not a separator problem at all. The number that decides the purchase is a property of the material bought months ago, not of the magnet being shopped for now.
