Chapter 73: Diesel and Background Noise

At 6:40, the bus on the old national highway pulled into the station right on time. The smell of diesel hit with the morning mist, and the white exhaust from the tailpipe spread quickly in the cold air. Lin Chen carried his canvas bag on his back, testing the first step with his left foot, the cardboard insole bracing the ball of his foot, while his right foot followed quickly behind. There were not many people in the bus, so he chose a single seat by the window. The canvas bag lay across his knees, zipper turned inward. As the wheels rolled over gravel, the vibration traveled through the seat frame into his tailbone. He closed his eyes and adjusted his breathing to sixteen breaths a minute. The gauze over the weeping wound on his left foot was already faintly damp, and the smell of iodine had been smothered by the mildew of the canvas bag. He did not touch it. He only kept his muscles relaxed.

At 7:15, the bus crossed Qinghe Bridge. The paddies outside the window receded into blocks of gray-green. He took the stiff paper ticket from the pocket against his body, its edges already worn fuzzy. Eighteen yuan. The ledger was at zero. He slipped the ticket back into the flyleaf of his notebook, beside the unopened letter from home. The radio on the bus was broadcasting the morning news, the sound breaking in and out, mixed with the hiss of static. He listened, but in his mind he was running through formulas for compensating voltage fluctuation. A ±5% fluctuation was not a linear value. It was a peak value. The frequency drift of a diesel generator would stack onto the background noise. If the proctor refused to power up the equipment in advance, the capacitor charging curve in the instant of a cold start would eat up the first thirty seconds of usable readings. In his head, he had to convert that blind spot of three seconds into the step error of manual attenuation.

At 8:30, the gate of the Provincial Electronics Institute came into view. Red brick walls. Iron railings. On the glass of the guard booth was a faded sign that read, "Competition Access Only." Lin Chen got off the bus, shortening his stride to twenty-five centimeters. The concrete road was flat, but water had pooled in the joints. He detoured around them, walking an extra forty meters. Two minutes of time margin were gone. He checked his watch. 8:35. The theory exam room was a tiered classroom on the second floor. The stairs were concrete, their edges worn shiny. He supported himself against the wall on the right and went up one step at a time. When his left foot touched down, a dull pain came from his ankle, like a thin needle advancing slowly beneath the skin. He did not frown. He only shifted his center of gravity, making his right foot take seventy percent of the load.

The classroom door was open, and the invigilator was handing out scratch paper. Lin Chen found a seat near the back and sat down. He placed the canvas bag by his feet, zipper half-open, one corner of the insulated case exposed. In the front row, several candidates in brand-new tracksuits were flipping through books, their pages rustling. Someone was quietly comparing answers in an easy tone. Someone else took a brand-new Casio calculator from a leather pencil case, the buttons clicking crisply. Lin Chen did not listen. He drew out a black gel pen and wrote the first line of a formula on the scratch paper. The tip scraped over the page, the faint rasp covering the surrounding noise. At nine o'clock sharp, the proctor rapped the blackboard. The written theory exam began.

The papers were handed down. The questions fell into three types: fundamentals, circuit analysis, and error calculation. Lin Chen first scanned the whole paper. There were fifteen percent more questions than in the training outline, but the core testing points had not changed. He skipped the multiple-choice section and turned straight to the calculation problems. The first problem was the frequency shift of a series-resonant RLC circuit. He wrote out the equations and plugged in the parameters without stopping. The second was the equivalent impedance of an oscilloscope probe compensation network. He drew the topology and marked the node voltages. The third was a conversion between the dynamic range of a spectrum analyzer and phase noise. He paused for two seconds and worked through a fast derivation on the scratch paper. The trap here was that the local oscillator frequency given in the problem was the nominal value; the actual instrument had thermal drift. He had to multiply in the thermal drift coefficient, or the result would come out 0.3 dB too high. He wrote down the corrected value and went on.

Time passed second by second. The air in the classroom gradually turned still. There was only the sound of pages turning and pens writing. Lin Chen's left foot began to go numb, the gauze over the seepage sticking to the skin, the friction bringing a slight stinging pain. He adjusted his sitting posture and shifted his weight half an inch to the right. His pen kept moving. He did not do unnecessary checks; he only verified the critical nodes. Every step of the derivation corresponded to measured data from the lab. The red notes in his error notebook rose automatically in his mind: parasitic capacitance, ground loops, impedance mismatch. He eliminated those variables one by one, leaving the shortest path. At 11:20, he wrote down the answer to the final major problem. He checked it once. No corrections. He handed it in.

When he walked out of the classroom, the wind swept down the corridor. He leaned against the wall and slowly moved his left ankle. The swelling ache wandered through the seams of the bone, but it was still within a controllable range. He took out his canteen and drank a mouthful. The water was already cool. There was no cafeteria at noon, so he sat on the stair landing and took cold steamed buns and pickled mustard greens from his canvas bag. He chewed slowly, his throat dry when he swallowed. After finishing the food, he wiped his fingers clean with a paper napkin. At 1:30 in the afternoon, the practical exam room opened. He arrived at Laboratory No. 3 forty minutes early. The door was still closed, and several candidates stood in the corridor reciting parameters. Lin Chen did not join them. He walked to the window at the end of the hall and looked down. In the courtyard stood a yellow diesel generator, pale white smoke lifting from its exhaust pipe. The machine was idling, and the low-frequency vibration rose through the floor slabs, very slight, but unbroken. A ±5% voltage fluctuation was not a number on paper. It was a physical mechanical tremor. He stared at the white smoke from the exhaust pipe and memorized the frequency of its rise and fall.

At 1:45, the invigilator opened the laboratory door. The smell of rosin and dust in the air was even heavier than yesterday. The HP8591E spectrum analyzer stood on the workbench by the window, its power indicator dark. Cold machine. Lin Chen went in, found the desk with his number, set down the canvas bag, and took out the insulated case. He sat and laid both hands flat on the desktop. His palms were slightly cool. The invigilator began reading the rules: twenty minutes of preparation time for the practical exam. No powering on equipment in advance. No external power sources allowed. Voltage fluctuation would be borne by the backup generator, and candidates had to calibrate for it themselves. All data would be judged by the final output.

When the rules were finished, the exam room fell quiet. Only the low hum of the generator rose and ebbed in the background. Lin Chen stared at the spectrum analyzer. Fifteen minutes of warm-up. Four minutes of ripple calibration. Three minutes of background-noise scanning. Two minutes of wiring. The timeline meshed together again in his mind. He ran through the merged steps in his contingency plan once more. Power on. Do not connect the signal source. Use the first seven minutes to record the voltage fluctuation curve at the same time. Substitute software compensation for hardware calibration. The amount of manual calculation would increase, but the time could be won back. The risk was that the momentary voltage sag when the generator started might trigger the instrument's protection circuit. Before switching on, he had to factor in the discharge time of the power filter capacitor.

At two o'clock sharp, the invigilator pressed the stopwatch. Begin. Lin Chen reached out and pressed the power switch on the spectrum analyzer. The fan gave a low hum, and the screen lit up. Green self-test characters scrolled past line by line. He kept his eyes on the stopwatch. First minute. The background-noise curve began to climb. He picked up his pen and recorded the voltmeter readings on graph paper. The fluctuations jumped between ±4.2% and ±5.8%. Not uniform. Pulsed. His pen moved faster, connecting the data points into a broken line. Third minute. A distinct dip appeared in the curve. The generator load had switched. He immediately substituted the values into the compensation formula on the scratch paper. Thermal drift coefficient multiplied by voltage attenuation rate. The result came out 0.15 dB too high. He fixed that correction in his mind.

Fifth minute. The warm-up indicator had still not turned green. The background noise was still drifting. He could not wait. He connected the test lead, leaving the other end unloaded. The screen showed that the input impedance match was normal. He closed his eyes and ran the parameters in the correction formula through his head once more. There was no shortcut. He could only drive the tolerance for error down to the limit. He opened his eyes, his fingers hovering above the manual attenuation knob. Ten dB. Step. Compensation. Output. Not one move could be wrong. One wrong move, and the chain of data would break.

Seventh minute. The generator gave a muffled cough, and its speed suddenly dropped. The background-noise curve on the screen shook violently. The voltmeter needle snapped hard to the left. Lin Chen's fingers did not move. He stared at the curve and waited for it to settle. Three seconds later, the speed recovered. The curve began to climb again. He quickly wrote down the new fluctuation peak on the paper. The compensation value had to be adjusted upward by another 0.08. He finished the calculation and looked up. The warm-up indicator had turned from red to green. Thirteen minutes remained. He picked up the signal-source lead. His fingertips touched the cold threads of the BNC connector. He screwed it on and tightened it. No extra movement.

The gears of the practical exam began to turn. Lin Chen's breathing stayed steady. His left foot was swelling faintly under the desk, but it did not affect the precision of his hands. He watched the spectrum graph on the screen, waiting for the first peak to appear. A crack opened in the clouds outside, and sunlight slanted in across the metal chassis. The wind carried the smell of diesel and wet earth. He pressed the measurement key. The data began to jump. The reference-level indicator at the lower left of the screen flickered once and did not reset to zero. His pupils tightened slightly. The baseline had shifted by 0.2 dB. It was not the voltage fluctuation. It was the instrument's internal reference oscillator slipping in phase during the transition from cold machine to warm machine. That variable was not in the contingency plan. In the nine minutes that remained, he would have to use manual stepping to pull that offset baseline flat again.