Get a bench. Build a device.

00Read this part first

Most lab handouts open by telling you what you are going to discover. This one opens by telling you what is out of reach, because that is the information you can actually act on, and because a document that overstates what it can deliver is the exact failure this whole research programme spends its time correcting.

So this page does two things instead. It gives you the real access ladder, rung by rung, with the emails and forms that actually move it — because learning to ask for an instrument is a more portable skill than any single measurement. And it gives you a build that carries an exact prediction, costs about $100, fits on a table, and is yours to keep.

01The access ladder, honestly tiered

Tier	What it gets you	Who can open it	Realistic timeline
1 · Bench Your own build	Chladni / DML plate. Standing-wave modes, an exact symbolic prediction, a speaker you keep.	You. Today. ~$100 and an afternoon.	Same day
2 · Teaching lab Campus bench	Function generator, oscilloscope, real accelerometer, laser vibrometer if the department has one. Quantitative mode frequencies.	A faculty member who signs for you, plus EHS training.	2–6 weeks
3 · Core facility Instrument time	Microscopy, mass spec, DRIFTS/FTIR, materials characterisation, at published fee-for-service rates.	A PI's account, or a company's voucher — see §03.	1–3 months
4 · Funded collaboration The WP-62 test	fs direct-write waveguide arrays, IR camera, sub-micron coupling.	A photonics PI with an existing grant. Not a student route.	6–24 months

Read that table as a strategy, not a ranking. Tier 1 is the one that produces a result this semester, and a clean tier-1 result with a proper log is the single strongest thing you can put in front of the person who controls tier 2 and 3. The ladder is climbed from the bottom, with evidence in hand.

02How to actually get a bench at NJIT

NJIT runs the relevant machinery through three offices, and they are all reachable by email. What follows is public information, verified 2026-08-13; the numbers and requirements change, so treat every figure below as a thing to confirm on the call, not a fact to repeat.

The three doors

• Undergraduate Research & Innovation (URI) — the front door for a student who wants a research bench. It runs a Provost summer fellowship (10 weeks, a $5,000 stipend, faculty advisor required, ending in the summer research symposium) and a seed-grant track during the year. research.njit.edu/uri · uri@njit.edu
• Environmental Health & Safety (EHS) — the gate. No training, no bench, no exceptions, and this is correct. Lab Safety Training is the baseline; Laser Safety Training runs about an hour and is required before you go near a class 3B/4 source. Your advisor also has to file an Undergraduates in Laboratories and Shops form. Training schedule · healthandsafety@njit.edu
• Otto H. York Center — the core facility: microscopy, mass spectrometry, environmental measurement, materials characterisation, available to the university and for contract research. Core facilities. It also runs an instrument-usage seed grant — read that page before you assume you need money.

The email that works

The reason most student requests fail is that they ask for access instead of offering a specific, small, already-half-done piece of work. Invert it. Send this, adapted, to a faculty member whose recent papers you have actually read:

Subject: 20 minutes on the [instrument] — undergraduate, protocol attached

Dr [Name],

I'm a [year] in [major]. I've built a driven-plate rig (32 mm exciter, 4 W,
aluminium plate) and measured nodal patterns at [n] drive frequencies. Logs
and photographs attached.

I have a prediction I can't test at my bench. The mode functions are
antisymmetric under x <-> y, so the diagonal x = y must be nodal in EVERY
mode. My plate shows it, but I can only see it in salt, not measure it. I'd
like [45 minutes] on [the scanning vibrometer / the shaker table] to get
amplitude along the diagonal against amplitude 5 mm off it.

I have completed EHS Lab Safety training [date] and can complete Laser
Safety before any session. I'm asking you because of [specific paper].

I can work whenever the instrument is otherwise idle, including evenings.

[Name] · [ID] · [phone]

Four things make that email work, and they generalise well past this project: you already did something; you name one measurement, not a programme; you have removed their liability problem (training, already done or scheduled); and you have made yourself cheap (idle time). A "no" to that email usually arrives with a referral attached, which is worth more than the bench.

03The $40,000 — what it is, and what it is not

New Jersey runs an R&D Voucher through the Commission on Science, Innovation and Technology (CSIT) with the NJEDA. The Catalyst and Clean Tech rounds each let an eligible applicant claim up to $40,000 in aggregate over any twelve-month period, to buy time at an in-state core facility: a NJ two- or four-year college or university, or a NJ federal or non-profit lab, that offers fee-for-service access and names a contact person. NJIT's York Center is exactly the kind of facility the programme is built around.

If you take that route, get these in writing before you start

1. Your name on the facility work order. Not "the company's student." Named. It is what makes the hours provable later.
2. Authorship position, agreed in advance, and what specifically would change it. Retroactive authorship conversations go badly for the person with less power.
3. Whether you are paid, and how. Voucher funds flow to the facility for services — they are not salary. If your time is unpaid, that is a decision you should be making with the number in front of you, not discovering afterward.
4. Data rights: may you use the raw data in a thesis, a poster, a course report? Get the yes before the run, not after.
5. IP. If the work touches anything patentable, NJIT and the company both have policies. Ask which one governs your contribution. "We'll sort it out later" is not an answer; it is a deferral in someone else's favour.

That list is the point of this section. The measurement is a semester; knowing to ask those five questions is a career.

04The lab-partner protocol

Work in pairs, and structure the pair so that the two of you are a check on each other rather than two people agreeing. The corpus this handout belongs to has spent months repairing claims that were wrong precisely because one fluent account went unchecked.

Pre-register, on paper, before you power anything on

PRE-REGISTRATION — sign and date, both partners

Prediction : the diagonal x = y is nodal in every mode of the square plate
Because    : f(x,y) = cos(n pi x)cos(m pi y) - cos(m pi x)cos(n pi y)
             so f(y,x) = -f(x,y), hence f(x,x) = -f(x,x) = 0 identically
Measures   : (a) does powder clear the diagonal, y/n, per frequency
             (b) diagonal line width, mm, +/- 1 mm
Frequencies: list them NOW, e.g. 200 400 630 880 1150 1400 Hz
CONTROL    : re-bolt the exciter 10 mm off-centre, repeat frequency #3,
             then re-centre and repeat it again.  <-- do not skip this
Falsified if: powder ACCUMULATES on the diagonal at any listed frequency
              in the centred configuration
ALSO failed : the off-centre run looks IDENTICAL to the centred run
              (then the diagonal was never symmetry-driven and the
               derivation was not doing the work you thought)

Partner A ______________  Partner B ______________  Date __________

Division of labour

• Partner A drives — sets frequency, holds amplitude constant, never touches the log.
• Partner B logs — records to the sheet, photographs each pattern with a ruler in frame, never touches the generator.
• Swap at the halfway point. If the second half disagrees with the first, you have learned something about your procedure, and that is the most valuable thing you will learn all day.
• Run the control step blind if you can — one partner re-bolts off-centre while the other looks away, then the observer calls whether it was moved from the pattern alone.

05Bring these — you are building a thing you keep

The device is a distributed-mode loudspeaker. A voice-coil exciter bolted to a flat plate drives the whole plate into bending modes: at audio rates it plays music, and at a single held tone the same plate shows you the mode as a standing pattern in powder. It is one object doing both jobs — the speaker is the instrument — and that is the whole reason it is worth an afternoon.

Full sourcing, part numbers, prices and the safety section are in WP-63. Summary, per station:

Item	Spec	Cost	Who brings it
Exciter	32 mm, 40 W, 4 Ω, two-hole bolt mount (not adhesive)	$22.98	Station
Amplifier	2×15 W class-D with Bluetooth	$34.98	Station
Power supply	12 V, ≥2 A	~$14 est.	Station
Plate	Aluminium or brass, ~250 mm square, 0.8–1.0 mm, deburred	~$15 est.	Station
Wire + terminals	16 AWG, spade terminals	$18.96	Station
Hardware	M3 bolt, nut, washers, centre-drilled plate	~$2	Station
Bring yourself:
Medium	Fine table salt, semolina, or lycopodium — a cup is plenty	$0–4	You
A phone	Bluetooth + a tone-generator app + camera	—	You
Music	Something with strong low end. You will hear it come out of a metal sheet.	—	You
Notebook	Paper. Signed pre-registration from §04.	—	You
Eye protection	Powder does become airborne	—	You

≈ $95–105 per station · ≈ $627 for six · ≈ $489 on the economy tier. Prices verified at Parts Express 2026-08-12; items marked est. are estimates and are not verified.

06What you are actually testing

For an ideal square plate, the nodal patterns come from mode functions of the form

f(x,y) = cos(n pi x) cos(m pi y) - cos(m pi x) cos(n pi y)

Swap the arguments and the whole thing changes sign: f(y,x) = -f(x,y). Put x = y and you get f(x,x) = -f(x,x), so f(x,x) = 0. Not approximately, not for particular n and m, not at particular frequencies — the diagonal is nodal in every single mode, by symmetry alone. Verified symbolically (SymPy) in WP-63.

That is why this build is worth doing rather than merely pretty. It is not a demonstration of a fitted curve; it is a two-line argument that makes an unconditional claim about a physical object, and you can put salt on the object and look.

Step 4 is the whole experiment: move the bolt

Chladni plates are all over the internet and nearly every version stops at look at the pretty pattern. That is a demonstration. A demonstration cannot fail, which is exactly what is wrong with it — nothing you do to it teaches you anything, because there was no claim on the table.

So do the one step everyone skips. Unbolt the exciter and re-bolt it 10 mm off-centre. Nothing else changes: same plate, same salt, same frequency, same amplitude. But the argument in the box above rests on one hypothesis and one only — that the system is symmetric under x ↔ y. Drive it off the diagonal and that hypothesis is false. The prediction has no ground left to stand on.

We are unapologetic that this is also a beautiful object — the patterns are gorgeous, the thing plays music, you will show it to people. Good. Build it in brass if brass makes you happier. But the art and the physics are not in tension here and we refuse to trade one for the other: the control step costs one bolt and it is what turns a nice-looking workshop craft into an experiment that could have come out the other way. Keep the beauty. Keep the control. The DIY project earns the word lab at exactly the moment you are willing to break your own result on purpose.

07What you walk out with

1. A working loudspeaker that you built, that is yours, and that also happens to be a mode visualiser.
2. A signed pre-registration and a filled log — the artefact that makes your next email to a faculty member land differently from every other student's.
3. A completed EHS training record, if you booked it, which is the single slowest item on the path to tier 2 and the one you can start today.
4. A named contact at a facility, a lab, or a company — obtained by asking a specific question rather than a general one.
5. The habit of writing the prediction down before the measurement. Everything else here is replaceable. That one is not.