reworks because of 'tx_mode 2' and protocol issues

° added markers (GIO12, GPIO14) for logic analyzer debugging

° change UART initialisation
  - set RxFIFO full to 1 char, so we can detect active telegram on EMS-Bus
  - set RxTOT to 1 char for the same reason

° in ems_parseTelegram: ignore telegram if  emsRxStatus != EMS_RX_STATUS_IDLE
  because we missed to respond in a timely manner

° ems_tx_buffer is not void anymore, instead we return the Tx status
° extended the emsTxStatus
  - EMS_TX_WTD_TIMEOUT: timeout while waiting for echo of last Tx byte
    out telegram was ignored by the bus master, so the byte wasn't echoed
  - EMS_TX_BRK_DETECT:  BRK detected while sending the telegram
    bus master send us a break because of protocol violation

° add burnStarts, burnWorkMin and heatWorkMin to MQTT boiler message

TODO:
  + take care of ems_tx_buffer return code
  + accept direct telegram in ems_parse_telegram and do a proper acknowledge
  + split the heartbeat in static (version, IP-address) and dynamic (RSS etc)
  + change the hearbeat to a JSON structure so we can easier pick the uptime
This commit is contained in:
Susis Strolch
2019-07-19 13:16:23 +02:00
parent 7743ff5d56
commit 5587b9d267
5 changed files with 151 additions and 90 deletions

View File

@@ -724,9 +724,23 @@ void ems_parseTelegram(uint8_t * telegram, uint8_t length) {
EMS_RxTelegram.timestamp = millis();
EMS_RxTelegram.length = length;
// check if we just received a single byte
// it could well be a Poll request from the boiler for us, which will have a value of 0x8B (0x0B | 0x80)
// or either a return code like 0x01 or 0x04 from the last Write command
/*
* check if we just received a single byte
* it could well be a Poll request from the boiler for us, which will have a value of 0x8B (0x0B | 0x80)
* or either a return code like 0x01 or 0x04 from the last Write command
* Roger Wilco: we have different types here:
* EMS_ID_ME && length == 1 && EMS_TX_STATUS_IDLE && EMS_RX_STATUS_IDLE: polling request
* EMS_ID_ME && length > 1 && EMS_TX_STATUS_IDLE && EMS_RX_STATUS_IDLE: direct telegram
* (EMS_TX_SUCCESS || EMS_TX_ERROR) && EMS_TX_STATUS_WAIT: response, free the EMS bus
*
* In addition, it may happen that we where interrupted (f.e. by WIFI activity) and the
* buffer isn't valid anymore, so we must not answer at all...
*/
if (EMS_Sys_Status.emsRxStatus != EMS_RX_STATUS_IDLE) {
myDebug_P(PSTR("** We missed the bus - Rx non-idle!"));
return;
}
if (length == 1) {
uint8_t value = telegram[0]; // 1st byte of data package
@@ -816,6 +830,7 @@ void ems_parseTelegram(uint8_t * telegram, uint8_t length) {
// Assume at this point we have something that vaguely resembles a telegram in the format [src] [dest] [type] [offset] [data] [crc]
// validate the CRC, if it's bad ignore it
if (telegram[length - 1] != _crcCalculator(telegram, length)) {
LA_PULSE(200);
EMS_Sys_Status.emxCrcErr++;
if (EMS_Sys_Status.emsLogging == EMS_SYS_LOGGING_VERBOSE) {
_debugPrintTelegram("Corrupt telegram: ", &EMS_RxTelegram, COLOR_RED, true);