Why a CCS session never gets past SLAC

SLAC is a measurement, not a negotiation. Where the sequence stops tells you whose turn it was, and the attenuation tells you whether to reach for a screwdriver or a configuration file.

A measurement, not a handshake

When a CCS session dies before the first IP packet, the reflex is to look for a software bug. Most of the time that is the wrong place to start. SLAC, the Signal Level Attenuation Characterization of ISO 15118-3, does not negotiate anything. The car sends a burst of sounding messages, every charger modem that hears them measures how loud they arrived, and the car picks the charger that heard it best. Only then does the charger hand the car the key of its HomePlug network.

A failed SLAC is therefore one of two things: a step that never happened, or a measurement the car did not like. That distinction decides where the afternoon goes. HomePlug Green PHY: powerline on the Control Pilot explains the modem layer underneath; this note is about reading the failure.

Find the message nobody sent

SLAC runs in a fixed order, so the last message in the capture tells you whose turn it was. Read it like this:

  • No CM_SLAC_PARM.REQ at all: the car never saw the 5 percent duty cycle that asks for digital communication, or it saw it and did not start. Without a Control Pilot log you cannot tell these apart, which is the best argument for always recording one.
  • CM_SLAC_PARM.REQ repeated, no CM_SLAC_PARM.CNF: SLAC is not running on the charger side, its modem is not in the charger role, or its host listens on the wrong interface.
  • Sounding, then nothing from the charger: its modem produced no attenuation profile, or its host never turned it into CM_ATTEN_CHAR.IND.
  • CM_ATTEN_CHAR.IND, the car's CM_ATTEN_CHAR.RSP, then silence: the car received the result and did not want the charger. This is the attenuation case.
  • CM_SLAC_MATCH.REQ without CM_SLAC_MATCH.CNF: the charger refused or ignored the match, often because the RunID or the addresses in it do not fit what it expects.

The analyzer does the same bookkeeping. SL-01 names the messages that are missing, and the verdict of a SLAC timeout names the last message seen and the side that never sent the next one.

Read the attenuation where it was measured

The number that matters is the average attenuation in CM_ATTEN_CHAR.IND. It is what the charger modem measured on the car's sounds, and it travels inside the message, so it does not depend on where your sniffer sits. A badly placed sniffer can miss frames, but it cannot make the link look worse than the charger measured it.

  • High in every carrier group: loss along the whole path. Think coupling circuit, a pilot contact, a long or damaged cable.
  • High in a few groups only: narrowband interference. The average may still pass.
  • Several responders with similar values: crosstalk from neighbouring charge points, and a car that may pick the wrong one (SL-09).

The slac-fail demo is the textbook case. Three attempts, one charger answering each time, 56.1, 55.9 and 55.8 dB, and each time the car acknowledges the result and starts over with a new RunID six seconds later. The analyzer splits the file into three sessions, one per attempt. The same number three times over is a steady cause, not a burst of noise. Worked example: SLAC that never matches walks through the capture.

What to check, in order

  • Record the Control Pilot with the capture. It answers the first questions, whether the 5 percent came and whether the car stayed in state B, before you open a single SLAC message.
  • Read the stage strip and the verdict, then the SLAC tab: attempts, RunIDs, responders and the attenuation per carrier group.
  • Swap one thing at a time: the same car at another charger, another car at this one, a short known good cable. The part that moves the number is the part at fault.
  • Check the modem configuration on both sides: the charger modem in the charger role, SLAC enabled, and a host that starts SLAC only once it sees state B.
  • Unplug to state A between attempts. Some cars try SLAC only a limited number of times per plug in.

What not to do

  • Do not raise slac.attenuationMaxDb until SL-06 stops warning. The 40 dB in the shipped profile is lab practice, not a value from the standard, and the car applies its own limit whatever the analyzer is set to. Set the threshold to what a healthy coupling of yours shows, so the warning means something.
  • Do not blame the charger because it is the bigger box. When the car walks away after CM_ATTEN_CHAR.RSP, the loss can sit in the car's inlet as easily as in the charger, which is why the verdict says Either.
  • Do not judge SLAC from a capture on the host interface of one side. The modem may handle some SLAC messages itself, and those never reach the host. A sniffer modem on the cable sees both sides.
  • Do not stop at SLAC when the match does complete. CM_SLAC_MATCH.CNF followed by silence usually means a network key that one host never wrote into its modem; SLAC matched, but no SDP answer or TCP connection covers that case.

The SLAC timers and the attenuation threshold of the limit profile are flagged for verification against ISO 15118-3.