Download Industrial communications first test

Industrial communications first test
DDC = direct digital control
DCS = distributed control systems
Why digital: more data, multidrop (many
devices on same com. media), robustness
(accurate values transferred), variety of
protocols.
Stanards: - cost, + quality; interoperability
(primary concern!); TAG (Technical Advisory
Group – national standards body), IEC
(International Electrotechnical Commission),
ISO, ISA (Instrumentation Society of America),
CEN, IEEE, EIA, TIA
Factory automation: ideal network types for
simple I/O focus on low overhead and small
data packets. Examples: Seriplex, CANbus,
AS-I (actuator-sensor interface) – Sensor
buses/bit level buses. Advanced: DeviceNet,
SDS, CANopen – device buses, byte-level
buses. Process automation: contin. regulatory
control; nets: Foundation Fieldbus, Profibus PA,
HART.
SCADA (supervisory control and data
acquisition) – defines a comp. system used for
gathering and analyzing real time data coming
from industrial processes → operator interface
is important – HMI.
Business/Enterprise level – big network,
LAN/WAN, several protocols used at the same
time. Low cost per node. Control level – LAN,
broadcast/point2point messaging between
automation nodes; protocols → backbone for
PLCs, SCADA, HMI. Ethernet → low cost.
Device/Field level: field-level means com. nets
that link indust. field devices (sensors,
actuators, controllers), ‘networking of I/O’. Uses
HART (FSK – Frequency Shift Keying). HART
uses a superimposed digital signal (at a low
level) on top of the 4-20 mA (built upon 4-20mA
Current Loop (CL)). Compatible with existing
analog devices, but enables usage of digital
signals. Device and field buses are almost on
the same level. Standards issues due to control
system supplier competition. Bit/sensor level –
simple buses. Small overhead, large cycle
times.
Criteria for choosing a best-fit network:
Redundancy, efficiency (how hard is to send a
message, how many msg’s for RW op-s, how
much the host computer has to do?), speed
(bandwidth – raw speed of data within a
channel), determinism (e.g. when using
Ethernet due to CSMA/CD performance cannot
be guaranteed), distance, length of messages,
cabling (tw.p or fiber-opt?), vendor support,
maintainability; how to choose: focus on
application, consider the costs, access the net
connectivity, understand hidden changes and
impacts; evaluate interoperability.
at lowest levels, used client-server model, VMD
– virtual manufact. device. Protocols of MAP:
ISO TP4 transp. – reliable conn.-oriented
service, 3way handsh.
Peculiarities: environment? EMI? Type of
data (time critical?), dependability (fail-safe?
process error situations? avoid downtime?
decentralization? error recovery? origin of
data? priorities of messages? fast changing
technology)
Integration of manufacturing enterprise
Requirments: dependability (handle errors and
emergencies, fault tolerance), autonomous
operation (decentralization – contradiction with
master/slave), time-tagged data, distributed
information of global time (via say local real
time clock), guaranteed delivery time, real-time
traffic (traffic of packets in com. channel should
be independent of presence of errors;
broadcast messages are important), datagrams
(~ connectionless), manageability (the com.
system needs to be able to reconfigure itself
according to various situations), scaleability.
MRP – Material requirements planning, MRPII
– manufacturing resource planning, ERP –
enterprise resource planning (on-line response
times).
Reqs @ field level: very short resp. time,
tolerance for harsh environments, long
distance, power distribution; Reqs @ control
level: short response times, tolerance for harsh
environments, very high availability (MTBF –
many years), security, power backup, net
management. Automation integration can be
achieved through extensive use of standards,
e.g. MAP.
MES – manufacturing execution systems;
developed to provide infrastructure for info
(real-time/on-line plant floor and logistics info)
on Manufacturing Enterprise Collaboration. →
carry out the plan. MES:
MAP (for interoperability) – manufact.
automation protocol – to overcome com.
problems between multi-vendor automation
devices (developed by GM and Boeing).
MAP3.0 specs published in 1984 → FullMAP
(flexibility for com. stations, not good for realtime), MiniMAP (reduced OSI stack, suitable for
time critical), EPA (merge MiniMAP/FullMAP).
MIS – manufacturing intelligence system; the
goal is to gather info and prepare it for
presentation and analysis.
MAP at higher levels: more complex info, long
distance, async timing; lower levels: simpler
info, short distances, sync timing. MAP
generated the token bus protocol and a new
message exch. protocol MMS – Manufacturing
Message Spesification ISO9506, most suitable
Transmission
impairments
–
noise,
attenuation distortion, delay distortion. BER –
bit error ratio, affected by bandwidth, SNR,
transmission media and distance environment.
S/N(db)=10log10(S/N). Noise types: Thermal
noise (white noise), atmospheric noise,
intermodulation (different frequencies on same
medium), crosstalk, power noise, transients
(impulse noise), noise coupling: EMI, inductive
(current), capacitive (voltage), RFI, common
impenance (different circuits share common
wires), conducted noise (via transm. noise by
wires) – normal mode between signal line and
circuit reference (diff. voltage, cannot be
distinguished from the transducer signal),
common mode noise between signaling circuit
and ground (picked up on both leads from
ground).
ASDC – Automatic Send Data Control. Needs
special circuitry that senses that data is being
transmitted. Preferred method because it
reduces software overhead and simplifies
programming.
Handling noise: differential signals + twisted
signal leads; current signals (vs. voltage
signals), proper GND and earthing, layout,
routing of cables, shielding, NSF (noise suppr.
filters), galvanic isolation.
Error detection and correction: EDC (error
detection codes): parity, checksums, CRC;
ECC (correction c.): hamming, reed-solomon.
Hamming distance – # of different bits in code
words. CRC is widely used in practice (ATM,
HDLC), performance: - can detect for r
bits/frame frame len. < 2r-1: all patterns of 1,2,3
err.; all burst errors of r or fewer bits; random
large # of err. with prob. of 1-2^-r. Correction:
AQR (Automatic Repeat reQuest) with seq.
numbers, acks, nacks, sacks & timers →
methods: stop&wait ARQ (1/2 duplex); slidingwindow: go-back-n; selective repeat.
Cable spacing 5cm-1.2m. Place cables over
AC lines only at correct angles. Shielding –
connect to GND only @ one end (normal
circumstances).
Wiring: plain pair, shielded pair, coaxial cable,
twisted pair (magnetic fields cancellation, most
common – grade 5 UTP cable, lightweight,
easy to pull & terminate, but susceptible to
EMI). STP – improve signaling rate being
heavier and more difficult to manufacture
(shield attenuates electrical fields). Fiber-optic –
best solution, expensive.
RS232 – intended primarily for DTE-DCE links,
adopted for char. oriented peripherals.
RS232C – volt. levels +/-15V
Differential transmitter: generates 2 signals of
equal, opposite polarity for each bit. Reciever:
sensitive only to difference between 2 signals
@ its inputs – noise is thus absorbed by both
wires and doesn’t affect receiver + good
common mode rejection.
232C vs. 232D
Transmit example: (DTE→DCE): DTR, RTS,
wait for DSR, wait for CTS, transmit the data.
Receive: (DCE→DTE) DTR, wait for DSR,
receive data.
Min RS232 signals: async: TD, RD, SG (signal
gnd); sync: TD, RD, SG, TC, RC, XTC (external
transmit clock).
RS232 problems: unbalanced transmission
(common-mode noise), top speed: 20Kbps,
max distance 15 m between DTE and DCE.
Hamming:
RS-422,-485:balanced → driver produces 1
signal that flows thru 2 lines (2...6 V across A
and B). ‚Enable’ connects the driver to its
output terminals.
‚disabled’ → third state (tristate).
RTS – DTE→DCE, CTS (clear t.s.)–
backwards. DSR (data set ready) – DCE to
DTE; DTR (data terminal ready – DTE to DCE,
means DTE is ready to accept data from DCE).
Error correction: FEC (e.g. hamming): extra
bits to detect & correct – large overhead,
cannot recover from huge errors, used in
simplex transm., & where transm. times are
long. BEC – only to detect, simple, effective,
least expensive.
485 vs. 232: 485 is balanced, uses dif.
signaling on pair of wires – small voltage
detection thresholds, common mode rejection –
improves performance over longer distances
(1200m, any rate below 9600 baud); uses
driver enable/disable – multidrop (up to 32
driver/receiver pairs can share a net).
A RS485 bus behaves like a transm. line,
therefore must be terminated to avoid
reflections (120Ohm res between A and B at
each end of the bus). Two Wire or Four Wire:
422 needs a dedicated pair of wires for each
signal, whereas 485 allow a single pair of wires
in half-duplex (reduced cable cost). Four-wire
with 485: one node a master node, others –
slaves. Tristate control may be achieved using
a RTS signal (0→tristate, 1→driver on).
Termination: to match impedance of a node to
the impedance of transm. line, otherwise
transm. signal is not completely absorbed by
the load and reflected back to the transmitter
(term. increases load on the drivers and
installation complexity). Biasing: in order to
maintain the proper idle voltage state in the line
(using pull-up (B) and pull-down (A) resistors) –
to maintain a minimum of 200mV between B
and A data lines. Underbiasing → decreased
noise immunity or even complete data failure.
485 driver control: 1. use a control line (e.g.
RTS handshake line) to enable/disable the
driver (may have timing related issues); 2.
Flow control: 1. point-to-point (softw. flow ctrl.
Xon/Xoff based on ASCII → not suitable for
binary/raw data transfers, h/w flow ctrl
RTS/CTS signal lines);
2.End-to-End flow ctrl – ACKs/NAKs
(stop&wait, sliding window : go-back-n,
selective repeat). Stop&wait works in half
duplex or when receiver buffer is limited in size
(1 frame). Efficiency=tframe/(2tprop+tframe). Sliding
window: transm. can send # blocks with seq.#
with no ACK (# determined by window size).
Go-back-N: retransmit all unACKed frames
from the last ACKed frame upon a receipt of an
out-of-sequence frame; Selective-repeat:
retransmit only corrupted frames, transmission
order is allowed to change. ‚Inflight’ data
amount ideally =bandwidth*delay [product ~
BDP]; Flowrate=RWin/RTT – if receiving app
can’t keep up; RWin<BDP then receiver limits
flow, otherwise chan. properties (RTT, b/width).