Imported from GitHub: pollen-robotics/elec_relay_R2 · commit 2b0699f · license Apache-2.0

Description

Relay 24V/100A + fuses to power Reachy 2

README

"Relay" for R2

"Relay for R2" is dedicated to sourcing every parts of R2 (except the mobile base).

High inrush current

We have been facing issues with the power supply of R2. It concerns battery/BMS/relais. Took some sweat to fix that and still a bit of time to be 100% confident about the fix. Long story short, the sum of the capacitor of R2 are putting pressure on the supply equipment. An extra capacitor has been added over the relay, this soften the startup of R2.

Here are the debugging steps and explanations:

• the mobile base relay has turned obsolete and has been swapped by the manufacturer equivalent
• BMS was triggered from time to time of some batteries
• new reference of the relay has been tested and we discovered that is started faster (1ms on the older ref. and 10us on the newer)
• at the same moment, torso relay were regularly burning. As no burning prints were observed and no fuse melted, we thought at overvoltage.
• tries with an extra heavy-duty diode: the loop is then closed with the help of the freewheeling diodes of the motor side. No effect has been seen, the zener diodes on the BLDC controller are probably doing a good job.
• measurements of the battery current shows 500A+ inrush current. Those are pretty impressive and slightly over specs of the relay (500A for 1ms, 100A continuous). Starting a single BLDC driver is still quite few Amps, it induces a voltage decrease on the battery (even though BMS is given for 135A for 10s)
• total amont of capacitor of R2's torso is 7500uF
• U=RI, we give a try with thiner cable (sic!) and then a big resistor (0,44 Ohms - 100W) directly on the battery. Thermal dissipation is pretty big (~85°C) an charging is a disaster (~30% of the capacity is lost and 100% charge is just impossible).
• Finally a trick is found: adding a big capacitor on the terminals of the relay: this will charge when the robot is off, then discharge/supply the BLDC drivers when the relay is turned on.
• measurements show that this operate well: the capacitor is more reactive than the relay and it taking care of a good part of the inrush current.
• a large amont of tries is done, switching on/off while moving, charging. A simulution is made to verify the operation of the trick

• light blue: relay_in (that is to say: battery voltage), it is almost shorted 100us after startup, one can then see the charge of the capacitor of R2
• yellow: relay_out.
• blue: current (10mV/A), directly on the battery output, first bump is made by the torso relay, second is the mobile base. This screenshot is made with the extra capacitor. Thought it looks like a big amont of current, it is below 500A so it will be handle by the torso relay
• white: reference current (10mV/A), before the adding of the capacitor.

• blue: inrush current of the original wiring
• brown: inrush current after adding the extra capacitor over relay
• red: extra capacitor current

Basically

• Designed with KiCAD 7, updated in 9