Echelon LonTal Stack - Manual (Page 47)

LonTalk Stack Developer’s Guide 35
o
sizeof(network_variable)
o
sizeof(message_length)
•
Interface
BPT represents data transfer overhead for the serial interface, and
has one of the following values:
o
1 bit per transfer for SPI
o
10 bits per transfer for SCI
•
exp
PPS is the expected packet-per-second throughput value
Example: For an average network variable size of 3 bytes, no explicit messaging
support, and a TP/FT-10 channel that delivers up to 180 packets per second, the
minimum bit rate for an SCI interface is 19 200 bps. To allow for larger NVs,
channel noise, and other systemic latency, you should consider setting the device
bit rate at the next greater value above the minimum calculated from the
formula. Thus, for this example, a bit rate of 38 400 or 76 800 bps is
recommended.
To calculate the expected packet-per-second throughput value for a channel, you
can use the Echelon Perf utility, available from www.echelon.com/downloads .
However, the bit rate is not the only factor that determines the link-layer transit
time. Some portion of the link-layer transit time is spent negotiating handshake
lines between the host and the Echelon Smart Transceiver. For faster bit rates,
the handshaking overhead can increase, thus your application might require a
faster clock speed for the Echelon Smart Transceiver to handle the extra
processing.
Example: For a Series 3100 Echelon Smart Transceiver running at 10 MHz and
an ARM7 host running at 20 MHz, the link-layer transit for a 4-byte network
variable fetch, the handshaking overhead can be as much as 22% of the total
link-layer transit time at 19 200 bps, and as much as 40% at 38 400 bps.
Even though a Series 3100 Echelon Smart Transceiver running at 5 MHz can be
sufficient for the demands of a power-line channel, a typical Echelon Smart
Transceiver operates at 10 MHz even when used exclusively with a power line
channel. The maximum clock rate for an Echelon Smart Transceiver based on a
PL 3120, PL 3150, or PL 3170 Echelon Smart Transceiver is 10 MHz.
For a performance test application that attempts to maximize the number of
propagated packets, the application is likely to show approximately 3% increased
throughput when operating with a 40 MHz Series 3100 Echelon Smart
Transceiver compared to a 10 MHz Series 3100 Echelon Smart Transceiver (for
FT 5000 Echelon Smart Transceivers, the comparison is between the 20 MHz
system clock setting and the 5 MHz system clock setting). However, for a
production application, which only occasionally transmits to the network and has
unused output buffers available on the Echelon Smart Transceiver, a faster
Echelon Smart Transceiver reduces the time required for the handshake
overhead (by up to a factor of 4 for Series 3100 devices – or up to a factor of 16 for
Series 5000 devices, compared to Series 3100 devices) so that a downlink packet
can be delivered to the Echelon Smart Transceiver more quickly, which can
improve overall application latency. Thus, depending on the needs of your
application, you can use a slower or faster Echelon Smart Transceiver.