CobraNet Software
Design Resources
CobraNet Firmware Version 2.09.16
Changes since 2.9.14 release
All platforms
- Fixed a timing problem in which the MOP_PACKET_RECEIVE HMI operation does not clear the 0x1 bit in hackSatus promptly. In this circumstance, the host can be lead to believe there is a receive packet available when there is not.
- Added SNMP physical memory inspector capability to test variants of the firmware.
- Source code can now be built with default latency and sample rates other than the standard 5-1/3ms at 48kHz.
CM-1
- In external master clock synchronization scenarios it is important to enter external master clock sync mode only when the master clock input is known to be stable and within frequency specification. This release fixes an audio interface re-initialization problem which could occasionally cause the audio channel mapping to be scrambled when the synchronization mode was changed via syncConductorClock and/or syncPerformerClock variables.
Errata
All platforms
- Acquisition time, stability and accuracy issues may still exist in External word clock synchronization mode.
- Audio transmission may be inhibited of bundle transmitters and receivers are not turned off (by setting bundle assignments to 0) before modeRateControl is modified.
- Address Translation MIB-II variables atIfIndex, atPhysAddress and atNetAddress are not available.
- Bundle priority does not correctly control transmission ordering. Transmission ordering is unimportant on repeater networks and only of marginal importance on a switched network.
- Possible transmission lockup triggered by broadcast storm on repeater network. Hot plugging the Ethernet connection resetting the interface will clear this condition. This is not a new problem.
- TFTP server will not retransmit a lost up-going data block if the client requests it. Read operation will timeout. Read operation can be restarted to recover. The TFTP server has always mishandled this condition.
High capacity reference design
- On large networks with many active bundles there is a potential for audio dropouts when interface is maximized to transmit 32 channels while receiving 32 channels. This is due to inadequate processing cycles to perform both management and audio functions. Dropouts may occur on transmit and receive. If the overloaded interface is serving as the conductor, there a potential for network-wide dropouts due to the overloading. This problem can be worked around by backing down the number of audio channels sourced and/or sunk by the interface and/or partitioning a large network into multiple smaller networks. This problem was first recognized in the 2.8.6 release but is not believed to be a new problem. Releases beginning with 2.8.7 contain efficiency improvements that address but have not entirely solved the problem.
Low capacity reference design
- Potential for ERROR_SECTIONLENGTH_UNEXPECTED and ERROR_RXBUFFER_OVERFLOW if all 4 receivers are activated and receiving 8 channel bundles and SNMP queries are being fielded by the device.
CM-1
- Attempting Receiving heavy UDP broadcast traffic via the packet bridge has been shown to crash the CM-1. A very high performance host processor is required to produce this condition.
- FBOOT script will not properly program SCI debug monitor into the boot sector of the flash memory when executed via SCI debug monitor. FBOOT operates properly when operated via the JTAG/OnCE interface.