I started with a ship full of max-level veterans, and I'm just a complete noob.

Jiang Ming hung up the phone, pressed the guidance report on the corner of the table, and then laid out five analog adaptive circuit diagrams according to the signal direction, asking Da Liu to check the DC operational amplifier tubes and high-stability resistors in the warehouse.

The 502 workbench was quickly filled with wooden boxes in the early morning. Da Liu measured the transconductance and zero drift of each double triode, while Fang Xudong grouped the resistors according to their tolerances, removing all components with a tolerance greater than 0.5% from the main circuit.

Jiang Ming selected two pairs of amplifier tubes with similar temperature drift and made them share the same shield and constant temperature base. The integrating capacitor also had to undergo charge and discharge tests. Only the limiting branch was assigned to the tubes with excessive leakage current.

Da Liu turned to the page about the disturbance generator. The soldering iron was stopped above the terminal block. In the diagram, the cam contact would periodically inject a small pulse into the power setpoint.

"The furnace temperature has finally stabilized, and we're deliberately pushing it a little harder. What's the difference between that and intentionally turning the steering wheel while driving?"

"Once the furnace ages, the original steering wheel will be off track. If you don't test it, it will keep sending false signals back."

Jiang Ming picked up the temperature measuring bridge and showed him the bridge output before and after the disturbance. The step amplitude only accounted for a small part of the allowable power change, but the temperature response was enough to expose the current thermal inertia.

"Follow the diagram, keep the disturbance amplitude, integration time, and limit value in place, put aside common experience for now, and wait until the modules are calibrated before making any changes."

Da Liu swallowed the rest of his question, reattached the soldering iron to the solder pads, and Fang Xudong recited the numbers one by one according to the wiring diagram. The first set of integral networks quickly lit up with a stable indicator light.

The disturbance generator uses a low-speed motor to drive a cam. After the contacts close, a narrow pulse is sent out. The response voltage enters the fast and slow branches, and then the gain and inertia estimates are formed by the diode multiplication network.

In the morning, the five modules completed their static calibrations. Jiang Ming then had Da Liu connect the furnace to simulate a load, using resistors and capacitors to generate three types of thermal inertia, and observed whether the parameter update voltage could automatically migrate.

The first two loads passed smoothly, but the third load had a long delay after it was switched on, and a continuous whistling sound came from the amplifier chassis, with the voltage meter needle on the panel hitting the limit post repeatedly.

Old Sun was installing a bracket on the control cabinet when he heard the sound and reached for the main switch. Jiang Ming had already pressed the test circuit breaker to cut off the parameter update loop separately.

A burnt smell emanated from near the relay base. After checking the records, Fang Xudong discovered that each burr corresponded to the release of a Soviet-made polarized relay, and the contact bounce sent a narrow pulse into the differential channel.

This sharp pulse is amplified by high gain and then treated by the multiplier as a change in the furnace body response. The parameter update voltage then surges to the upper limit, and the safety loop is also dragged into continuous correction.

Old Sun disassembled the relay casing and examined the contacts with a magnifying glass. The contact surface was still intact, and the mechanical bounce was due to a defect in the device's structure.

"Even if I replace it, it will still behave the same way. The contacts always bounce back a couple of times unless I grind the signal circuit smooth."

Jiang Ming extracted the differential channel diagram, added a stage of RC low-pass network to the relay output, and removed two mica capacitors from the old circuit board to form a filter branch with a variable time constant together with an adjustable resistor.

If the filter is too heavy, it will swallow up the real fast response; if it is too light, it will not suppress the contact glitches. He asked Da Liu to connect the signal to the oscilloscope, continuously toggle the relay, and adjust the potentiometer along the waveform himself.

As the spike gradually shrinks into the error band, the leading edge of the normal step remains clear. Jiang Ming then adds an inverse limiting diode at the amplifier input to clamp the remaining spikes within the safe voltage range.

Before powering on again, he reduced the parameter update speed by one level, so that the integrating capacitor only received continuous responses, and any transitions with too short a duration could not drive the variable gain network.

When Da Liu turned on the test power supply, the cam contact sent out the first disturbance pulse. After the pointer crossed the reference scale, it returned, leaving a new bias voltage at the output of the multiplier.

After the second set of simulated loads is connected, the control loop first observes the response slope, then reduces the proportional gain and extends the integral time, so that the servo motor retracts its drive before overshoot occurs.

The previous state of chasing the deviation back and forth disappeared. Each correction by the execution end was earlier than the temperature peak. Da Liu stared at the voltmeter and slowly put the soldering iron back on the table.

"It only needs to be tested once to know how much to collect next time. Can this thing really remember the furnace's temperament?"

"The capacitor stores the current parameters and continues to correct them for the next disturbance. The furnace changes slowly, so it can keep up."

Jiang Ming switched between three analog loads in succession, and the variable gain network completed the migration within two disturbance cycles. The limiter also kept the parameters firmly within the permissible range.

Standing on the side of the control cabinet, Lao Sun saw the servo motor's running light change from flashing frequently to lighting up at long intervals, and reached out to pat the welded chassis.

"Copper wire, light bulb, a few pipes—after you finished connecting them, you really became good at reading people. Furnace No. 3 has gained a brain this time."

In the afternoon, the entire control cabinet was transported to the No. 3 furnace room of the Institute of Physics. Da Liu connected the temperature measuring bridge, the lookup table feedforward, and the execution end, while Fang Xudong re-unified the time reference of the recorder, cam disturbance, and propulsion mechanism.

The no-load low-temperature test started at 800 degrees Celsius. The parameter identification circuit identified two power states in sequence. The newly added filter network continued to work, and the relay glitch no longer exceeded the error band.

Old Sun stood by the clamp, inspecting the newly replaced spring and copper foil. Jiang Ming handed over the high-temperature test procedures to Fang Xudong and rushed back to 502 to prepare to deal with another line.

At the same time, Wu Hanzhang returned from the factory and placed a thick Russian dossier on the table in the confidential compartment. The gray cover was sealed with a lead seal, and the top-secret mark was stamped with two red seals.

"The Soviet representative said that this set of parameters completes the low-latitude correction and lateral stability data, and suggested that it be used directly as the standard for the third demonstration meeting. The preparatory group has already started arguing about it."

Jiang Ming glanced at the clock and dialed the furnace room, asking Lao Sun to continue preheating under no-load to 1300 degrees Celsius, while Da Liu kept the disturbance response and parameter update voltage throughout the process.

After arranging the No. 3 furnace, he cut open the lead seal of the dossier, turned to the aerodynamic data section of the table of contents, and the first set of deflection coefficient tables quickly came into view.

The table is arranged in layers according to height, speed and angle of attack, with each column having the same number of decimal places, and the changes between adjacent intervals also maintaining a neat increasing relationship.

Jiang Ming pulled aside the original summary previously provided by the Su side. The old data had measurement fluctuations in several transonic speed ranges, but the corresponding positions in the new file had been smoothed into curves.

It's normal for the complete parameters to be revised, and the smoothing process may also come from new wind tunnel results. However, these two pages of matrices are too neat, even cleaning up the discrepancies that are common in measurement data.

His finger moved down along the aerodynamic deflection coefficient, finally stopping between two adjacent altitude layers, where the change remained at the same value for six consecutive times.