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

Jiang Minggang had just finished writing the words "Soviet cybernetics system" at the bottom of the page when Lao Sun threw the old clip onto the workbench. The single-sided shrapnel had been baked out of shape by the high temperature, and the copper foil was now just a narrow strip of contact mark.

"If you keep burning it with this thing, no matter how many times you adjust the parameters, it's all in vain. Just clamp the two ends of the rod firmly first."

Old Sun took out a section of high-temperature resistant nickel-chromium alloy wire from the toolbox, wound it around the steel core to form two springs, and then disassembled the pressure plate that was originally under pressure on one side, changing it into a clamping structure that is symmetrical on both sides.

Two springs press against the outer edge of the pressure plate. When thermal expansion occurs, the clamps can still apply pressure on both sides. After the preload is checked by the spring scale, it is increased to about three times that of the old structure.

The treatment of the contact surface is more laborious. Lao Sun cut three layers of copper foil of different thicknesses. The inner layer is responsible for filling the undulations at the end of the rod, the middle layer disperses the clamping pressure, and the outer layer is repeatedly polished with fine sandpaper until it can be connected as one piece under the light.

Jiang Ming picked up the polished copper foil and examined it. The edges were smooth and there were no horizontal scratches on the surface. After pressing it into the clamp, it should be able to increase the actual conductive area.

"The multi-layered copper foil softens when heated, and the spring continues to press in, making the contact surface more and more solid as it heats up. As long as the nickel-chromium wire doesn't lose its elasticity first, this structure can hold up."

After Lao Sun finished speaking, he put the silicon carbide rod back into the furnace. Da Liu took a multimeter and measured from the center of the rod all the way to the two clamps at both ends. Fang Xudong numbered each item according to the position recorded last time.

The room temperature measurement results quickly filled half a page. The equivalent resistance of the two connection terminals was reduced by about 70% compared to before the modification, and the difference in values ​​after multiple disassemblies and reassemblies was also reduced to one-third of the original value.

Old Sun was still not reassured, so he loosened the tablet and reassembled it, deliberately turning one of the copper foil layers in a different direction, and had Big Liu retest it. The readings were still within the acceptable range.

"This time the contact surface isn't picky about the hand; it'll be about the same regardless of who's wearing it."

Fang Xudong plotted the repeated measurement results next to the old structure curve, and the data points that were originally scattered up and down were narrowed into a narrow band. The contact resistance drift was estimated to be reduced by at least 60% based on high temperature.

The second no-load heating began immediately. Da Liu turned on the main switch and gradually increased the input power according to the newly determined slope. Jiang Ming watched over the paper tape recorder and aligned the current changes with the furnace temperature response segment by segment.

After the temperature exceeded 800 degrees Celsius, the current curve of the double-spring structure remained smooth, and the 40-second periodic oscillation that had previously appeared near 1100 degrees Celsius left only with very small fluctuations.

Fang Xudong pressed the old and new paper strips side by side under the glass plate. The old line swung back and forth on both sides of the set value, while the new line remained close to the center line even at 1200 degrees.

As the furnace temperature continued to rise, the silicon carbide rod gradually entered the high-temperature resistance change zone, the current curve began to deviate from the trend of the medium temperature range, and the temperature paper tape also showed low-frequency fluctuations again.

When the temperature approached 1400 degrees, the fluctuation range stabilized between ±0.5 and ±0.6 degrees, which was half of the previous ±1.2 degrees.

"The modified clamp worked, and the remaining section matches the change in the resistance of the rod."

Fang Xudong drew a line downwards along the two peaks, and the corresponding point fell at the bend of the silicon carbide rod current curve, thus separating the two types of disturbances that were originally mixed together.

Jiang Ming stared at the paper strip, his heart tightening again as he had just relaxed. The ±0.6 degree was enough to surpass the old furnace of the Institute of Physics, but it was still more than seven times worse than the ±0.08 degree of the germanium era.

Silicon regional purification involves a hotter melting zone, and the material is more sensitive to impurities and interface fluctuations. This level of precision is only enough to push the furnace to the doorstep.

He instructed Liu to maintain the temperature, and then divided the 800 to 1420 degree Celsius range into three temperature zones, preparing to remeasure the step response of each zone.

In the first segment, from 800 to 1000 degrees Celsius, the silicon carbide rod exhibits a negative temperature coefficient. As the temperature rises, the resistance decreases, and the current increases accordingly under constant voltage power supply. The system's equivalent gain is higher than that of the old molybdenum wire heating element.

In the second segment, from 1,000 to 1,200 degrees Celsius, the resistance curve enters a turning point. For the same power step, different rates of temperature rise will occur at the two ends of the interval.

The third segment spans from 1200 to 1420 degrees Celsius. The rod gradually enters the positive temperature coefficient region, and the thermal inertia of the furnace reaches its maximum. By the time the output change is transmitted to the thermocouple, there is already a significant delay.

Jiang Ming first had Da Liu increase the fixed power in the first temperature zone, and Fang Xudong recorded the time required for the temperature to reach 63% of the final change. Then Jiang Ming calculated the time constant and the equivalent gain.

After measuring the three temperature zones in sequence, three sets of significantly different parameters were obtained on the paper, and the original set of unified PID tuning values ​​were completely crossed out.

"As the silicon carbide rod moves from the medium-temperature zone to the high-temperature zone, the controlled object changes accordingly. We continue to use the same set of parameters, which is equivalent to measuring three different things with the same ruler."

Following the segmented tuning method used in the germanium era of Furnace No. 3, Jiang Ming calculated the proportional, integral, and differential parameters for the three temperature zones respectively, while Da Liu disassembled the control cabinet and added three sets of parameter networks next to the original circuit.

The switching relay is driven by a thermocouple signal. Once the temperature crosses the preset range, the corresponding resistor and capacitor network is connected to the controller, avoiding delays caused by manual switching.

When the first segment was put into operation, a parameter jump occurred near 1200 degrees Celsius, and the output power spiked along the paper tape. Da Liu immediately adjusted the relay contacts and added a narrow overlap band between the two temperature zones.

The switching after the rework was much smoother, and the furnace temperature fluctuations within the range of 800 to 1200 degrees Celsius were gradually reduced to within ±0.3 degrees Celsius.

The truly challenging part remains in the highest temperature zone. When the temperature deviates from the set value, the resistance of the silicon carbide rod changes with the temperature, and the current under constant voltage power supply moves in the opposite direction. However, the heat accumulated in the furnace continues to push the temperature upward.

The integral stage continuously accumulates corrections based on the old deviations. By the time the corrections are actually transmitted to the furnace, the temperature has already begun to drop, so the output once again exceeds the appropriate position.

Jiang Ming plotted the power, rod resistance, and temperature curves of the third segment on the same coordinate system, and marked the delays point by point along the time axis, gradually revealing the direction of change of the positive feedback loop.

He weakened the integral effect and strengthened the differential prediction. The overshoot in the first round of experiments was narrowed, but the small noise in the thermocouple signal was amplified synchronously, and the actuator began to make frequent adjustments.

"If we add differentials, the relays will break first, and the furnace temperature will fluctuate with the high-frequency power."

Da Liu closed the control cabinet side panel, his hand still on the switch, clearly unwilling to put the equipment on this road.

Jiang Ming also knew that conventional PID could only make corrections based on the deviation and its changes, but the loop gain of the silicon carbide rod would change with temperature, which was also affected by the hysteresis caused by the grain boundary resistance and the thermal inertia of the furnace.

Relying solely on the differential element to predict in advance can at most suppress a small temperature range. Once the operating point is changed, the parameters that were just adjusted will become ineffective again.

He wrote down the relationship between the heating element resistance, drive current and furnace temperature in a kraft paper notebook, placed the linear PID in the right corner, and left a blank space in front of it.

This gap requires a dedicated nonlinear compensation circuit to determine the change in the rod's resistance based on the real-time temperature, and to adjust the drive power in advance so that the PID only handles the remaining small deviations.

How to construct the compensable function, and how to incorporate the hysteresis during the heating and cooling of the rod into the model, are questions that the previous linear control experience of Furnace No. 3 could not answer.

Before leaving get off work, Fang Xudong re-filled out the progress sheet, including the results of the chuck drift compression of 60%, the no-load fluctuation of ±0.6 degrees, and the segmented PID test results for the day.

In the "Expected Completion Time" column, he wrote that the nonlinear compensation problem would take three to five days to solve, and the silicon material loading test would be arranged at the earliest in two weeks.

After reading the time on the paper, Wu Hanzhang picked up a red pencil but did not urge Jiang Ming to speed up the process. Instead, he wrote a line in the blank space.

From 500 to 1400 degrees Celsius, the temperature control team went through the process again.

Jiang Ming folded the schedule back into the folder, and after returning to his dormitory at night, he locked the doors and windows and laid out the kraft paper notebook, Qian Xuesen's lecture notes, and old hardcover memos in turn.

He also dug out the segmented data left by the demagnetization compensation of Yuanlei No. 1 permanent magnet from the old pages. That method could correct the output according to the state, but it was still missing a key step in dealing with high temperature hysteresis.

He turned to the lineage of the Kluge node in the Tongtianlu, extending eastward along the records of precision instrument cooperation from the Berlin period, and an old line leading to the Moscow Power Institute gradually lit up.

The line was marked with the records of a Soviet expert who died in 1951, as well as records of more than twenty sets of automatic control systems for Ural high-temperature furnaces.

Jiang Ming placed his hand on the connection, and the old laboratory in his mind immediately transformed into a furnace control room with rough walls, where two resistance curves in opposite directions were laid out on the table.