MIL-STD-1553 — military avionics bus: Question bank

5.5.28 · D5 · Coding › Embedded Systems & Real-Time Software › MIL-STD-1553 — military avionics bus

Shuru karne se pehle, neeche use hone waale vocabulary ka ek do-line refresher (kabhi bhi aisa term mat guess karo jise tum clearly define nahi kar sakte):

Bus Controller (BC) ::: ek single "dictator" node — ek maatra node jo transaction shuru kar sakta hai; baaki sab wait karte hain jab tak unhe address na kiya jaaye.
Remote Terminal (RT) ::: ek slave device (radar, nav, weapons) jo sirf BC ke command ke reply mein bolta hai.
Bus Monitor (BM) ::: ek passive listener jo traffic log karta hai aur kabhi transmit nahi karta.
Word ::: bus par ek 20-bit unit = 3-bit sync + 16 data bits + 1 parity bit.

True or false — justify karo

TF1. "MIL-STD-1553 bas RS-485 Protocol ka ek slow version hai."

False. Dono differential twisted pair use karte hain, lekin RS-485 multi-master hai bina kisi arbiter ke, isliye collisions possible hain; 1553 strictly command/response hai ek BC ke saath, isliye collisions structurally impossible hain — ye ek completely alag guarantee hai, sirf slow version nahi.

TF2. "Kyunki Manchester II har bit par ek edge guarantee karta hai, ek 1553 receiver 32 identical data words par bhi kabhi clock sync nahi khota."

True. Har bit period mein ek mandatory mid-bit transition hoti hai, isliye bina edge ke sabse lamba gap

T /2 = 0.5 μ s

hai chahe data pattern kuch bhi ho — PLL ke paas hamesha kuch lock karne ko hota hai.

TF3. "Ek teesra redundant bus add karne se dual-bus design ki reliability roughly teen guna ho jaayegi."

False. Reliability

p^{n}

follow karta hai: do buses pehle se

p^{2}

dete hain (e.g.

1 0^{- 6}

), isliye teesra bus sirf loss probability ko ek aur

p

factor se giraata hai jo diminishing returns mein hai, jabki real weight aur cost add hoti hai — isliye avionics do par hi rukta hai.

TF4. "Ek RT apne aap transaction shuru kar sakta hai agar uske paas urgent data ho (e.g. koi threat warning)."

False. Ek RT sirf apne Status Word mein Service Request bit set karke request kar sakta hai; BC ko phir decide karna hota hai ki usse poll kare. RT kabhi bus khud nahi seize karta — ye determinism tod deta.

TF5. "3-bit sync field word ke front par sirf teen ordinary Manchester bits hain."

False. Sync ek deliberately invalid Manchester pattern hai (3

μ s

steady LOW ya HIGH bina kisi mid-bit transition ke), isliye ise kabhi data samjha nahi ja sakta — yahi cheez word boundaries ko unambiguous banati hai.

TF6. "Transformer coupling mainly signal voltage boost karne ke liye exist karta hai."

False. Ye galvanic isolation ke liye exist karta hai (ground loops todne ke liye) aur ek benign failure mode ke liye — ek dead stub bus ko short karne ki bajaye open circuit ki tarah dikhta hai. Voltage driver set karta hai, transformer nahi.

TF7. "1 Mbit/s 1973 mein bhi technically embarrassing tha; woh bas faster build nahi kar sake."

False (mostly). Ye ek deliberate trade-off tha: ~1 MHz par wavelength (~300 m) 100+ ft cables tolerate karta hai aur zyaadatar engine-ignition EMI (>10 MHz) ke neeche rehta hai, aur 1973 ke pulse transformers wahan clean edges maintain kar sakte the. Slow rehna sahi engineering choice tha.

TF8. "Bus Monitor ek teesri tarah ka active node hai jo BC ke marne par takeover kar sakta hai."

False. BM purely passive hota hai — ye sirf sunta aur log karta hai. BC failover ek backup BC ko promote karke handle hota hai, jo ek bilkul alag mechanism hai, monitor se nahi.

Error dhundho

SE1. "Word Count = 0 matlab RT kuch nahi bhejna."

Error. Word Count field ki value 0 32 encode karti hai (maximum), kyunki 5-bit field otherwise 32 tak nahi pahunch sakti. Zero data words ka request simply kabhi valid nahi hota.

SE2. "T/R bit RT ko batata hai ki Bus A ya B par jawab de."

Error. T/R (Transmit/Receive) RT ko data ki direction batata hai — 1 = RT transmit karta hai, 0 = RT receive karta hai. Bus A vs B ka faisla is baat se hota hai ki BC ne command kis physical channel par drive ki thi, word mein kisi bit se nahi.

SE3. "8-data-word transmit command ke liye total time

20 + 20 + 160 = 200 μ s

hai."

Error. Tum response-time gap bhool gaye. Ye hai

20

(command)

+ 8

(RT response gap)

+ 20

(status)

+ 160

(8 data words)

= 208 μ s

. Inter-message gaps deterministic timing ka hissa hain, ignore karne wala overhead nahi.

SE4. "Kyunki Manchester ko har bit par ek edge chahiye, ek 20-bit word mein exactly 20 transitions hoti hain."

Error. Mid-bit transition hamesha hoti hai (unke 20), lekin consecutive bits ke beech boundary transition ho bhi sakti hai ya nahi bhi — yeh depend karta hai ki levels match karte hain ya nahi — plus sync deliberately koi mid-bit edge nahi rakhti. Isliye "exactly 20" dono siron se galat hai.

SE5. "BC Bus A par bhejta hai, wait karta hai, aur Bus B tabhi use karta hai jab A timeout ho jaaye."

Error. Normal dual-redundant operation mein BC same message dono buses par bhejta hai (thodi si offset ke saath), aur RT jo bhi valid copy pehle aaye use use karta hai. Bus B sirf ek cold spare nahi hai jo timeout ke baad jaagta hai.

SE6. "Differential signaling line se saari noise hata deta hai."

Error. Ye sirf common-mode noise cancel karta hai — woh interference jo dono wires par equally padti hai. Noise jo dono wires par alag-alag couple hoti hai (differential-mode) survive karti hai; pair ko twist karna hi zyaadatar real noise ko common-mode rakhta hai.

SE7. "8 mein se data word 5 par parity failure matlab BC words 1–4 rakh le aur sirf 5–8 re-request kare."

Error. 1553 mein koi partial-message recovery nahi hai — ek parity failure poore transaction ko invalidate karta hai aur BC poora message retry karta hai (typically 3 attempts tak).

Why questions

WY1. Ethernet, 1000× faster hone ke bawajood, missile-release control loop ke liye kyun unsuitable hai?

Kyunki iska latency non-deterministic hai — CSMA/CD backoff aur switch buffering matlab hai ki tum worst-case arrival time guarantee nahi kar sakte, aur ek control loop ko hard ceiling chahiye, fast average nahi. Contrast dekho Time-Triggered Architectures se.

WY2. Command/response exactly zero collisions kyun deta hai?

Kyunki BC bus time ka har microsecond own karta hai — koi bhi tab tak transmit nahi karta jab tak BC ne abhi command na di ho — isliye do nodes kabhi medium ke liye contend nahi kar sakte. Contrast karo CAN Bus se, jo contention allow karta hai aur priority arbitration se resolve karta hai.

WY3. 1553 simple ACK bit ki jagah Status Word kyun use karta hai?

Ek bit sirf "received" kehta hai; Status Word mein message-error, busy, service-request aur doosre health flags pack hote hain, isliye ek hi reply acknowledge bhi karta hai aur RT ki condition report bhi — zaroori hai jab tum flight mein ek box physically inspect nahi kar sakte.

WY4. ARINC 429 (airliners mein use hota hai) fighter ke tightly-coupled avionics ke liye kyun prefer nahi kiya jaata?

ARINC 429 one-transmitter-per-wire aur unidirectional hai, isliye bahut saare aapas mein baat karne wale boxes ko point-to-point links ki ek gunje hui jaali chahiye; 1553 ka single shared, bidirectional, BC-scheduled bus ek pair par 31 RTs tak scale karta hai guaranteed timing ke saath.

WY5. Higher signaling speed aircraft cable ke 100+ ft par actually hurt kyun karta hai?

Faster edges matlab shorter wavelengths, isliye reflections, attenuation aur impedance mismatches long runs par waveform ko badly distort karte hain — slow 1 MHz signaling us range mein rehti hai jo cable cleanly carry kar sakta hai.

WY6. Standard 4–12

μ s

RT response window mandate kyun karta hai, "turant reply do" ki jagah?

Ek window slow aur fast hardware ko coexist karne deti hai jabki worst case bounded rehta hai, isliye BC ek fixed schedule compute kar sakta hai. "Immediately" immeasurable aur unenforceable hai alag-alag RT designs mein — dekho Real-Time Scheduling Theory.

WY7. DO-178C certification ek active switched network ki jagah passive shared bus kyun prefer karta hai?

Nodes ke beech kam active components matlab analyze aur prove karne ke liye kam failure modes; DO-178C Certification Level A mein exhaustive failure-case coverage chahiye, aur ek transformer-coupled passive bus ko safe argue karna switch firmware se bahut simpler hai.

WY8. Is bus ke liye plain NRZ ki jagah Manchester Encoding kyun choose kiya jaata hai?

NRZ identical bits ke run par bina edge ke 100

μ s

ja sakta hai, receiver clock drift kar sakta hai; Manchester ki guaranteed per-bit edge PLL ko locked rakhti hai aur clock ko self-carry bhi karti hai, alag clock wire ki zaroorat nahi.

Edge cases

EC1. Agar ek RT ka stub cable shrapnel se physically cut ho jaaye to bus par kya hota hai?

Iska transformer-coupled stub open circuit ban jaata hai — ek benign failure jo main bus ko short ya load nahi karta — isliye har doosra node normally communicate karta rehta hai.

EC2. Do RTs accidentally same address par set ho jaate hain. Kya toot ta hai aur kab?

Kuch bhi collide nahi hota jab tak BC us address ko transmit karne ka command nahi deta — tab dono RTs bus drive karte hain ek saath, reply corrupt ho jaata hai. Bus aaram se theek hai; fault sirf duplicate address ka transmit poll hone par dikhta hai.

EC3. BC (bus master) khud mid-flight fail ho jaaye. Kya bus chalti rehti hai?

Nahi — bina BC ke, koi transactions initiate nahi karta aur bus silent ho jaati hai. Survivability ek pre-designated backup BC ke takeover karne se aati hai, RTs ke self-organize karne se nahi.

EC4. Ek RT ko transmit karne ka command diya jaata hai lekin wo abhi bhi data compute kar raha hai. Wo kya karta hai?

Wo apna Status Word reply karta hai Busy bit set karke, data defer karte hue; BC Busy dekh kar baad mein re-poll kar sakta hai. Ise bus ko wait mein nahi rokna chahiye — ye response window blow kar dega.

EC5. Bus A aur Bus B dono ek RT ko same message ki valid copies deliver karte hain. RT kya rakhta hai?

Jo pehle aaye; doosri valid copy simply ignore ho jaati hai. Redundancy "first good wins" hai, vote ya merge nahi.

EC6. Ek Word Count field receive command par value 0 rakhti hai — RT kitne words expect karta hai?

32 — maximum. Zero 32 ka encoding trick hai kyunki 5 bits 31 tak max out hoti hain; genuine "no words" transaction exist hi nahi karta.

EC7. Bus A command carry karta hai lekin noise ek data word mein ek bit flip kar deti hai (parity fail), jabki Bus B ki copy clean hai. Kya hota hai?

RT/BC clean Bus B copy use karta hai — yahi dual redundancy ka poora point hai: ek independent physical channel usually usi transient se hit nahi hoti, isliye ek valid copy survive karti hai.

EC8. Normal (non-sync) 1553 word mein kahin bhi koi bhi voltage transition na hone ka sabse lamba interval kya hai, aur ye kyun matter karta hai?

0.5 μ s

— half a bit period — kyunki Manchester har bit mein mid-bit edge force karta hai; ye bounded gap hi exactly woh cheez hai jo receiver PLL ko kabhi lock lose nahi karne deti.

Recall Har trap family ki ek-line summary

True/False traps identity confusions attack karte hain (1553 vs RS-485/Ethernet); Spot-the-error traps encoding aur timing slips attack karte hain (Count=0, 208 $μ s$ budget, whole-message retry); Why-questions design-goal reasoning force karte hain (determinism aur certification raw speed se zyaada); Edge cases degenerate-input thinking force karte hain (severed stubs, dead BC, busy RTs, duplicate addresses).