Precharge

Precharge

Watch the walkthrough video

A device to precharge the Tractive System. This prototype features voltage feedback to protect AIRs. In open-loop systems, if a wiring fault develops (eg. precharge resistor is disconnected or discharge is stuck on) then the precharge is ineffective and AIRs may become damaged.

Figure 1: The prototype Precharge module

Important Directories / Files

• code/precharge - The source code
• docs/schematic-v1.1.pdf - The schematic, as built.
• docs/pinout-v1.1.pdf - Pinout diagram with connection descriptions.
• docs - Data collected during commissioning, image assets for this document.
• Precharge - The PCB design files

Contents

• Specifications and Features
• Operation
• Commissioning Steps - Precharge
  • Characterise Voltage-Frequency Relationship
  • Simulate Precharge
  • Results
• Commissioning Steps - PDOC
• How to modify this design
• Recommendations

Specifications and Features

• Wide input voltage 0 - 600V
• Powered by shutdown circuit: 12V
• Integrated PDOC
• Isolated HV measurement
• CAN interface (provisional) for status output
• Serial interface for status and commissioning data
• FSAE one-size-fits-all. This module is capable of precharging any voltage accumulator (within FSAE rules) without significant hardware/software rework. To modify for a new accumulator/tractive-system, select appropriate precharge resistor and update relevant software parameters, discussed in How to modify this design

Operation

A brief description of the Precharge logic follows. Refer to Figures 2 & 3 for state-flow information.

• Initialise in State: Standby and monitor for a stable Shutdown Circuit.
• If Shutdown Circuit is stable, enter State: Precharge:
  • Close the precharge relay
  • Monitor Accumulator voltage and Tractive System voltage
  • Once TSV is close enough to AV (eg >95%), precharge is complete
  • If precharge is too fast or too slow, a fault is likely present eg. Wiring fault, discharge stuck-on, stuck relay.
• If precharge completed with no errors, proceed to State: Online:
  • Enable the AIR, open precharge relay

If at any point the Shutdown Circuit voltage becomes too low, enter State: Standby

Figure 2: A top-level look at the state-flow behaviour

Figure 3: precharge sequence timing. When the TS voltage reaches the target voltage, precharge is complete and the AIR closes. The target voltage should be 90-95% of the Accumulator voltage

Commissioning Steps - Precharge

Refer to the docs/schematic-v1.1 for component and test-point references.

Refer to docs/Experiments.xlsx for sample data collected during prototype construction.

This section describes the steps taken to construct and characterise the Precharge Module. Repeat these steps as necessary when re-commissioning later versions of the circuit.

Characterise Voltage-Frequency Relationship

• Assemble power supply components and V-F converter circuitry U5 - U9.

• Characterise voltage-frequency performance: perform V-F experiment code/01-v-f-experiment Refer to Figure 4 for sample data collected during prototype construction

  • Apply 0.15-10V at V-F inputs: TP5, TP6.
  • Record output frequencies: TP1, TP2 vs input voltages.
  • Create linear fit and record the gain and offset parameters.
  • Update code/precharge/measurements.cpp with the linear-fit parameters: V2F_slope_accu, V2F_ofs_accu, V2F_slope_ts, V2F_ofs_ts

• If resistors R31,R32,R36-39 or R33,R34,R40-R3 are different to schematic:

  • Find the gain of these voltage dividers: eg Gain = R39/(R31+R32+R36+R37+R38+R39)
  • Update code/precharge/measurements.cpp: gainVoltageDivider with the new gain value.
  

  Figure 4: Results from the V-F experiment - data collected from the prototype during construction. The V-F response is extremely linear. Expect f(0V) = 0Hz, which is supported by very small y-intercept constants. While the V-F converter circuits are nominally identical, they should be characterised separately to account for component tolerance.

Simulate Precharge

Here, we simulate a precharge sequence using low voltages that bypasses the prescaling voltage dividers.

Figure 5: Simulate a precharge sequence using a low voltage. The prescaling voltage dividers are bypassed by connecting Testpoint:TP5 to the switched accumulator voltage; and Testpoint:TP6 to the HV OUT connector.

Voltages listed are referenced to the GNDS net, connector J3 labelled TS-

• Connect a known capacitive load eg 1000uF to HV OUT. Include a large parallel resistance (eg 20k) for capacitor discharge.
  • The voltage divider created by this resistance and the precharge resistor R46 will limit the maximum voltage seen at capacitor. Large resistances will reduce this effect, but the capacitor will discharge more slowly during this test.
• Bypass the prescaling voltage dividers:
  • Connect TP5 to precharge resistor (accumulator side)
  • Connect TP6 to HV OUT
• Apply 4-10V to HV IN. Exact voltage will depend on accumulator voltage to be simulated: Accumulator Voltage multiplied by gainVoltageDivider (0.015 from prescribed components) gives this voltage.
• Monitor serial data from the microcontroller via USB
• Power the circuit at J1
• Monitor precharge behaviour in serial console and note precharge percentage and duration.
• Modify and upload code/precharge/precharge.ino as necessary. Relevant parameters are:
  • MIN_EXPECTED[ms] The minimum allowable precharge time. Times faster than this are likely due to wiring fault.
  • MAX_EXPECTED[ms] The maximum allowable precharge time. Times slower than this are likely due to wiring fault, stuck discharge circuit.
  • TARGET_PERCENT[%] Precharge to this percentage of accumulator voltage.
    • The discharge resistance used during this simulation affects the maximum TS voltage attainable. I found a maximum voltage of only 90% when using Precharge:390Ohm and discharge 4k7.

Results

A precharge sequence was simulated following the above procedure with results shown in Figure 6. Simulated voltages were measured accurate to the nearest volt.

Parameters:

• V Accumulator: 5.05V (336V equivalent)
• Tractive system capacitor: 1000uF
• Tractive system discharge capacitor: 4k7 (connected in parallel with capacitor. Larger value as in Figure 5 is preferred)
• gainVoltageDivider= 0.015 (code/precharge/measurements.cpp)
• Target precharge percentage: 88.0 (discharge resistor is only one order of magnitude larger than 390Ohm precharge resistor)

Figure 6: Results from a simulated precharge. Here, the accumulator was simulated with 5.05V => 337Vequiv. The TS voltage at TP6 reached 4.561V => 304Vequiv. which is 90.2% as indicated. The precharge finished at 90% instead of 88% because some settling time is imposed by the program.

Of note are the seemingly high precharge percentages observed between 0-200ms. These are artifacts of moving average filters used to smooth voltage measurements and reject spurious measurements. Since the smoothed accumulator voltage does not rise instantaneously, the precharge voltage represents a significant percentage in early stages of the precharge cycle.

Commissioning Steps - PDOC

The Precharge module includes an integrated thermal overload protection (PDOC) (Figure 7). A thermistor TH1 monitors the precharge resistor temperature. As temperature increases, the inverting-input voltage of comparator U10A rises. Once the voltage at the inverting input rises above the reference voltage (noninverting input) the PDOC will trip. The reference voltage created by R26 and R27 therefore sets the temperature threshold. If R27 = R29 then select R26 to be equal to the value of TH1 at the desired trip-temperature. Referring to thermistor data in Figure 8, for a trip-temperature of eg. 80 degC, R26 should be about 12kOhm.

Figure 7: The Precharge Overload Circuit monitors the temperature of the precharge resistor and triggers a fault if the temperature becomes too high. R26,R27 set the threshold temperature, TH1 is the temperature-monitoring thermistor.

Figure 8: The thermistor response. R_25 is 10kOhm for the selected device. At 80 degrees Celcius the resistance will be 1.242 x 10kOhm = 12.4kOhm

How to modify this design

When moving to a new tractive system configuration and/or accumulator voltage, the only hardware component that may require respecification is the Precharge Resistor R46. A HS25 series resistor (datasheet) from TE Connectivity is specified for this component - available in a large range of resistances. As per the datasheet, these devices are capable of short-duration overloads many times in excess of their continuous-duty rating. It is likely that only the resistance will need to be respecified, ie. select a HS25 family resistor of the appropriate resistance.

Short term overloading of a power resistor is acceptable. This design relies on short term overloading to keep the power resistor small, light and cheap.
(Graphic Source: TE Connectivity Type HS datasheet)

In 2020, NU Racing specified a 400V accumulator and a TS capacitance of 1600uF. Using these as design constraints, a 390Ohm precharge resistor was selected to precharge the TS quickly without excessive overload. As seen below, the resistor is overloaded to 400W (16 times its rating) for a very brief period at the start of a precharge sequence. After approximately 1 second the resistor is no longer overloaded. The initial precharge current(~1A) is well within the capacity of the precharge contactor. The resistance is above the minimum required for 2020 motorcontrollers (47 Ohm).

A 390Ohm 25W resistor precharges the TS to 95% in about 2.5 seconds, experiencing a brief 16x overload. Overloads of this nature are acceptable for the selected family of power resistors and allow cheaper, smaller, low-power resistors to be specified.

Once the appropriate resistor has been selected, follow the Simulate Precharge steps to identify any software parameters that require modification.

Recommendations & ToDo

V-F Conversion: Either include offset in V-F converter circuit so that min frequency is eg ~100Hz instead of 0Hz (Datasheet: Figure 14) or replace V-F converters with a microcontroller programmed for the same task. The challenge is that low frequencies require a long timeout period. Applying a 100Hz offset means the longest timeout would be 10 milli-seconds. The current solution works well, using a combination of outlier rejection and Exponential Moving Average filters.

PCB: The footprint assigned to the 4N35 optocouplers is slightly off. Serviceable, but not perfect. Consider re-assigning the footprint.