Shaker Motor Randomness: Can Be Biased?
A shaker motor is not a random number generator, and treating the two as interchangeable is the single dumbest assumption in cabinet hardware auditing. The motor spins an offset weight, the weight shakes the frame, and the frame rattles a ball across a perceived random path — but that path is governed by torque curves, bearing wear, and voltage sag, not by entropy. When someone claims a shaker is "rigged," what they usually mean is the mechanical output is deterministic in ways the operator never measured. 🎯
The core mechanism is brutally simple: a DC motor drives an eccentric mass, and the resulting vibration amplitude scales with rotational speed squared. Any drift in supply voltage, any carbon buildup on the commutator, any grease migration in the bearing race shifts that speed. That is not randomness — that is a slow, repeatable bias hiding inside a component everyone assumes is chaotic. Shaker motor randomness is a myth peddled by people who never put a tachometer on the shaft.
Where the Bias Actually Hides
Run a shaker motor through ten thousand cycles and log the vibration signature. You will find amplitude clustering that correlates with temperature, not with any seeding event. Cold grease means higher current draw, lower RPM, and a longer dwell at the low-frequency end of the shake envelope. Warm grease inverts the profile. The result is a hardware bias that skews outcome distribution toward whatever physical state the machine happens to be in — and players read that skew as "the shaker is cheating me." ⚙️
Bearing tolerance is the second culprit. A race with 0.02mm of play produces a lateral wobble that adds a secondary frequency component, and that component can resonate with the cabinet mount if the mounting bolts are torqued unevenly. Resonance amplifies one axis of motion and dampens the other. Now your "random" shake has a directional preference. That is a measurable, exploitable bias, and it lives entirely in the hardware layer.
The Voltage Sag Problem
Shared power rails are where shaker bias gets nasty. Every time the shaker motor kicks on, it pulls a current spike that drags the rail down for a few milliseconds. If the control board shares that rail, the logic-level thresholds shift during the shake window. A marginal comparator can flip state under sag conditions and not under clean conditions. That is not a software bug — it is an electrical bias baked into the shaker motor randomness profile, and it is reproducible on a scope. 🧠
Anyone auditing this hardware needs to scope the rail during motor engagement, not just at idle. Capture the transient, correlate it with the outcome log, and the bias stops being mysterious. It becomes a number — a sag depth in millivolts and a duration in milliseconds that maps directly to a shift in outcome frequency.
Debunking the "Seeded Shake" Claim
There is no seed. There is no firmware coin-flip feeding the motor. The shaker is an open-loop actuator with a known transfer function, and its output is dominated by mechanical and electrical state variables. The real question is not whether the shaker is random — it is whether the auditor has instrumented the transfer function tightly enough to prove the bias is within tolerance. Most have not. Most never will. 🔧
Torque ripple from a worn brush card alone can inject a 3 to 5 percent amplitude modulation at twice line frequency, and that modulation is invisible unless you are sampling the accelerometer output fast enough to catch it. That is the whole game: Shaker motor randomness bias is real, it is physical, and it is sitting in the frequency domain waiting for someone to pull the FFT and call it what it is.