Purpose

Enables the ability to switch endpoints on/off, provides redundant power from AUX/DCDC to all loads, and protects the LV bus from endpoint short circuits by disconnected damaged endpoints.

Pin Mappings

Open

Load Switches

Diagnostic mode is always pulled high for current sensing. Toggling the sense select pin high reads output 1, and low reads output 0.

Expander 1

Address: 0x21

Expander 0

Address: 0x20

Power Input

Transient Suppression

Each power input is coming from a long wire originating from the battery box and going to the front of the car. These wires will have a lot of inductance, causing voltage spikes which may damage components. A TVS diode is placed at each input to clamp the rail voltage in the event of any transients.

Current Sensing

There is current sensing for monitoring/logging purposes. An INA180 will be used due to it having good performance being the cheapest current sense amplifier available. The sense resistor will be placed on the low side due to the INA180 have much lower input offset with no common mode DC bias.

Resistor and Gain

Using a gain of 100V/V, the input measurement range is 33mV for a 3.3V output. Using a 6mR sense resistor, the current measurement range is 5.5A at 600mV/A. This is above the maximum current in the power budget of 3.5A with some margin.

Power Supply

The input PSRR is typically 8uV/V. Assuming 3V3 from controller board has a ripple of 50mV, the error caused by that would be 0.4uV. This would amount to an additional error of 0.07mA, which is negligible. Thus an LDO is not required to power the IC.

Input Capacitance

Input capacitors are placed before the current shunt to reduce the common mode swings across the sense resistor caused by the ESL of the source during changing loads.

Power Path

LTC4418 - Dual Channel Prioritized PowerPath Controller (ciiva.com)

Enables dual powering of loads, where if the primary supply (V1) is no longer available (voltage not within specified OV/UV range for a certain amount of time), it will switch automatically to a secondary supply.

Shutdown

Shutting down the power path ICs provides the ability to fully de-power the vehicle. The shutdown pin of both ICs are tied together and connected to the switch on the dash.

Inrush

LTC4418 minimized startup inrush by soft starting the gate drive of the first input to connect the output. However, during supply switchover excessive inrush due to the voltage difference between the 2 supplies can lead to undesirable UV fault.

Open

LTC4418 Datasheet pg. 16

When DCDC switches to AUX:

• Vout Min = 9V (DCDC supply 10V - 1V droop)

• Vin Max = 15V (Fully charged AUX)

• Rsrc = 400mR (200mR cells, 165mR 6m round trip 2x 22AWG, 35mR misc. fuse, conn, shunt, etc.)

• ESRcout = 10mR

• Rdson = 5mR

Inrush = 14.3A

When AUX switches to DCDC:

• Vout Min = 9V (Depleted AUX 10V - 1V droop)

• Vin Max = 13V (DCDC supply)

• Rsrc = 230mR (30mR DCDC output cap, 165mR 6m round trip 2x 22AWG, 35mR misc.)

• ESRcout = 10mR

• Rdson = 5mR

Inrush = 16A

Given that the two inputs are similar in voltage and the ESR of the source is high due to long interconnects, expected inrush is low and thus slew rate limiting the gate drive as described in Figure 4 of the datasheet is not required. In reality, inrush is likely to be even lower due to ESL of long wires.

Output Capacitance

Sufficient capacitance must be present to minimize the output voltage droop during supply switchover. The highest minimum voltage I could find for a priority load is 9V for the WS22 motor controller. Given the UV threshold is 10V, allowable voltage droop is 1V to keep the LV bus above 9V.

Using the following values:

• Imax = 3.5A (LV Power Budget [MSXV])

• tg = 4us (datsheet worst case)

• Vdroop = 1V

• ESR = 10mR

Cout = 14.6uF

Use of AU caps was avoided since they are prone to damage while reworking nearby components with hot air. A few different MLCC caps were considered below.

As expected, a larger package size and more stable dielectric contributes significantly to the voltage derating of the part, leaving 1210 X7R parts as the clear option. The capacitance at 12V between 22uF and 10uF parts is very similar, with the 10uF 35V part having surprisingly low voltage derating so it is chosen. Two of them will be used to satisfy the capacitance requirement.

Input Capacitance

Input capacitance is added to decouple the long ESL of the power wires, and reduce the voltage droop during input switchover. 20uF of input capacitance, the same as the output, is added which should theoretically cut the voltage droop in half.

Operating Thresholds

The operating threshold is determined by the voltage range of the aux battery which is from 10-15V. This allows the aux battery to be disconnected automatically when it becomes depleted. Some margin will be added so that, accounting for resistor divider and comparator error, resulting in the following thresholds:

• UV = 10.5V

• OV = 14.0V

• Hysteresis = 500mV

An input will be considered valid given that it is between 11V and 15V with 500mV of hysteresis to allow for voltage droop due to inrush or transients due to ESL of input.

Hysteresis will be set with a external current of 500nA, giving Rhys to be 126k. We will use 127k as the closest common value, yielding Iext = 496nA.

Using Iext from above and 500mV desired hysteresis, R3 = 1.008M, 1M is used as the closest value.

The comparator threshold is 1V, so R1 = 78.9k, with 78.7k used.

R2 = 26.32k, with 26.7k used.

Actual configuration, not accounting for error is as follows:

• UV = 10.49V

• UVhys = 496mV

• OV = 14.05V

• OVhys = 513mV

Pmos Selection

Given the large amount of inductance that the long power cables would have, it is conceivable that during input switching, there can be large inductive voltage spikes on the power rail which might exceed 20V. Thus, the Vds of the Pmos was chosen to be at least 30V, and the lowest cost option with a Rdson of < 10mV was chosen.

Load Switch

Current Sensing

The IS pin outputs a current proportional to the load current during nominal operation. Sensing voltage to load current calculation can be found in output table above.

Clamping Diodes

During a fault or open load scenario, the ISense voltage can exceed the 4V maximum rating on the MCU ADC pins. No discrete clamp diode is added since the mux clamps inputs to Vcc + 0.5V, where Vcc is 3.3V.

Digital Inputs

EN

Enable pins should have have pull down resistors to ensure that the load is off during when not being actively driven. In Rev 2.0, the pins are left floating if the controller board is not flashed, resulting in all the switches being enabled. The datasheet suggests that if a pulldown exists, a load might stay on during loss of ground, however, the risk of this is low compared to a fault that causes the controller board to fail, leave the pins floating and all the loads on.

Layout

Given the max steady state current is anticipated to be 3.5A, minimum conductor width will be sized for 4A with a temp rise of 25C. Conductors will be enlarged beyond the minimum where possible to reduce power loss.

Inner Layer 0.5oz: 6mm

Outer Layer 1oz: 3mm

0.3mm/0.5mm via: 1.1A

Thermal Reliefs

Where pads connect directly to very large planes (typically ground) a thermal relief is added. Due to the high density of parts many smaller polygons do not feature thermal reliefs to pads, since they would take up a significant portion of the polygon area.

Mounting Holes

Mounting holes and any attached standoffs are grounded. Test equipment may be clipped onto these points to help with debug.

Component Designators

Component designators have largely been omitted from the layout due to space constraints. The usefulness of designators decreases with increasing board size, as having “C57” among hundreds of 0603 parts does not make it easier to find. Eventually the “Top Designator” layer will be updated with all designators so that they are included in the PDF export.

Issues

1 - VIN Short

Symptom: Both 12V inputs were shorted to ground. When injecting current to try and find the short, the impedance of the short was so low that no significant heat was generated.

Cause: NC drill files got messed up, which resulted in holes where there weren’t supposed to be as well as via pads without holes. The erroneous plated holes shorted through all the layers.

Resolution: Holes were drilled out, and jumper wires were soldered to make up for the missing vias. NC drill file was regenerated which resolved the issue.

2 - Startup Inrush

Symptom: When 12V power is applied to either input while a controller board is connected to a programmer, the board makes a buzzing noise and 12V fails to reach the output of the power path ICs

Cause: On startup, the LTC4418 limits the slew rate of the output to reduce the inrush current. However, slew rate limiting on startup is only active if the output is below 2.3V. When a programmer is plugged in, it provides 3V power to the 12V rail, bypassing the slew rate limit on startup. The inrush current results in the input voltage dropping below UVO, causing the IC to shut the output fets and continually retry.

Resolution: Since there will be no programmer connected to the controller board in the vehicle, no hardware changes will be made.

3 - Source Switching

Symptom: Switching sources at > 3A load current is unsuccessful. As soon as the higher priority source is connected (purple), the output (yellow) starts to fall. Overshoot caused by inrush is also very high, peaking at 18V.

Cause: It seems that the excessive inrush from connecting the new source messes something up inside the IC, causing it to think that the original source is out of the valid window and shutting it down. CH2 shows that the original source does not go outside the valid window. This is backed up by the fact that when the bench supply is made to be the original source and the aux battery the new source, switchover happens successfully and the issue only exists when the sources are swapped (bench is new, and IC switches from battery to bench). Presumably this is because the internal resistance of the battery is much higher than the ESR of the output caps on the bench supply, so when the aux is connected the inrush is much lower.

I could have verified this by connecting the power supply while it is off, and then turning it on, rather than having the power supply on before hand and then connecting it to PD. This would have greatly limited the slew rate of the source, and thus the inrush as well. Due to dumb reasons this test was not performed.

Resolution: In the vehicle, this shouldn’t be an issue since PCS would always be connected first and then turned on, so there wouldn’t be any significant inrush. Nevertheless input capacitance can be reduced by depopulating a 10uF input cap.

Change Log

Rev 2.0

• Rev 1.0 was a copy of the MSXIV, and did not fit into the MSXV system. Rev 2.0 should really be “1.0”.

Rev 2.1

• It was found that the 5V buck ICs were damaged (shorted) when testing the OVP. As such, the OVP was lowered from 15 to 14V. The aux battery when fully charged settles over time to ~13.75V so it should still be okay.

• Wack vias were fixed by regenerating NC drill files

• Pulldowns for all the EN pins added, since if the controller board is not flashed and the output to the IO expanders are floating, all the outputs are on in their default state.

• Slew rate limiting added for power path fets

  • To accommodate for potentially reducing input capacitance (issue 3)

• Decoupling cap on load switch from 100nF to 10uF to add additional bus capacitance

  • To accommodate for lower slew rate limiting above. Adds ~1.5uF @ 12V DC bias per cap, +4x caps for normally aux and +6x caps for normally pcs buses.

• Changed input TVS from 14V to 13V rated diodes

  • Min reverse breakdown is 15V which is a little lower than the previous 14V diodes but still higher than operating voltages. Inrush from the connector had the voltage spiking up to 18V, this should help clamp that down a little better

• Added indicator LEDs to power input