MULTIPLEX WIRING SYSTEMS
`
`26.9
`
`ENGINE COMPARTMENT
`Washer Fluid
`
`DRIVER DOOR
`Door Clased
`
`Wiper Switch
`Key-in Switch
`Seat Belts
`
`Body
`Computer
`
`PASSENGER DOOR
`Deor Closed
`
`Door Ajar
`
`
`|
`Bus
`Ground
`
`INSTRUMENT PANEL
`Headlamps on
`Low Fuel
`
`Power
`
`
`
`Instrument
`Cluster
`
`
`Message
`
`Center
`
`FIGURE 26.9 Base vehicle with Class B data link, body computer, instrument
`cluster, message center, and Class A network for sensors.
`
`Class C Real-Time Control. The Class C networkis the least mature network, and consen-
`sus of opinion on requirements does not yet exist. Experts cannot agree on many facets and
`many of the requirements are controversial. Many automotive engineers believe that an ade-
`quate statistical latency achievable with a bit-by-bit arbitration-based media access protocol
`is sufficient for real-time distributed Class C multiplexing. Others believe a token-passing
`media access protocol is required because a maximum latency is guaranteed, because with an
`arbitration-based media access onlya statistical maximum latency is achievable. Therearestill
`others who argue that both the arbitration-based and token-passing media access is not good
`enoughfortightly looped distributed processing because both methods have too great a vari-
`ation in latency time. They argue that a time-triggered media access method is required
`because the network variations in latency should not affect tightly looped processing times.
`Other factors such as required data rates and the physical media type also remain open. It is
`clear that more research and developmentis required to resolve these questions.
`
`26.2 ENCODING TECHNIQUES
`
`The data encoding technique‘hasa significant effect on the radiated EMI. In orderto achieve
`the highest possible data rate,it is important to choose a data-encoding techniquethat has the
`fewest transitions per bit with the maximum amountof time betweentransitions and bit-syn-
`chronized so that invalid bit testing can be effective. PWM,for example, has two transitions
`per bit with % bit times between transitions. NRZ has a maximum ofonetransition per bit but
`is increased to provide for synchronization. Some of the disk drive encoding techniques such
`as modified frequency modulation (MFM) are synchronous with fewer than onetransition
`perbit. (See Table 26.1 for a comparison chart of a selection of encoding techniques used in
`vehicle multiplexing.)
`The variable column in the table describes an attribute whereby the transmission time for
`data byte is a variable quantity depending on the data value. VPWM and Bit-Stuf NRZ both
`have variable byte repetition rate (data variability).
`Some of the bit-encoding techniques synchronize on transitions that fall on or within the
`bit boundaries. 10-bit NRZ, Bit-Stuf NRZ, and E-MANall employ addedtransitions for syn-
`chronization (clock synchronization).
`
`541
`541
`
`

`

`26.10
`
`SAFETY, CONVENIENCE, ENTERTAINMENT, AND OTHER SYSTEMS
`
`TABLE 26.1 Comparison of Multiplexing Bit-Encoding Techniques
`
`PWM VPWM=10-BitNRZ Bit-StufiNRZ L-MAN E-MAN MFM
`
`Variable
`No
`Yes
`No
`Yes
`No
`No
`No
`Synchronizing
`Yes
`Yes
`No
`No
`Yes
`No
`‘Yes
`Arbitrates
`Yes
`Yes
`‘Yes
`Yes
`Yes
`Yes
`Yes
`Transition/bit
`2
`1
`$1.25
`1.015
`<2
`$1.25
`<1
`Maxdata rate
`TAK
`112K
`13.5K
`16.6 K
`84K
`135K 168K
`dBV <PWM
`Base
`9
`11
`14
`5
`1
`15
`Oscillator tolerance
`429.2% +29.2%
`+5.1%
`+9.7%
`+29.2%
`+9.7% +10.7%
`Integrity
`Perfect Good
`Fair
`Fair
`Perfect
`Fair
`Superb
`
`All of the encoding techniques considered are capable of bit-by-bit arbitration. This is not
`commonly recognized with some of the encoding techniques, e.g. MFM, and will be addressed
`in Sec. 26.2.7 (arbitrates). Bit-by-bit arbitration is calculated on the numberoftransitions per
`bit of data.
`A suitable data-encoding technique should not generate excessive levels of EMI, andthis
`consideration is a dominating challenge in making an encoding choice. The CISPR Standard*
`is usually considered adequate.This factor determines the maximum allowable data rate of the
`encoding technique, in order to maintain a level of EMI below the CISPR standard break
`point, i-e.,-60 dBV at 500 kHz. The values predicted in Table 26.2 used the same technique
`described in the nextsection.
`Equation (26.1) in Fig. 26.10 can be used to predict’ the EMIlevels radiated by a single
`wire in a vehicle wiring harness. The technique was usedto calculate and plot a Fourierseries
`of a sample trapezoid wave to determine the values given in Table 26.2, The calculations
`assumed a 10.4 Kbps data rate at a 42 percent factor of the minimum feature size (minimum
`pulse width) to determine the rise time. Consider the trapezoidal wave shownin Fig.26.10.
`The shortestrise time (42 percentof shortest pulse width), shortest pulse time, and fastest rep-
`etition rate should yield the worst-case EMI in dBV. The actual measured EMIwill be a few
`dBYbetter than the calculated dBV because the output driver frequency bandwidth doesact
`as a low-passfilter [EMI below PWM (dBV < PWM)].
`
`* CISPR/D/WG2(Secretariat) 19 Sept 1989 Radiated Emissions Antenna & Probe Test Documenthas been gener-
`ally interpreted by most RF engineers to specify a break point at 500 kHz of -60 dBV.
`
`Oo, =
`
`bhoy
`T
`
`freq =
`
`nis
`T
`
`where 7 is an integer
`
`a=n- Oh
`
`b=n- Oh
`c=nhA' OG
`
`-1
`R=— +
`ty
`
` Cas(a) x Cos(b) — Cos(c)
`4
`B-h
`
`ss —Sin(a) A Sin(c) — Sin(b)
`B-t
`=
`3,
`
`MAG =20log V2M2-L VREGF (26:1)
`(2-n-ny
`
`
`ts
`ts
`| Time
` |.
`
`to
`
`ty
`
`Volts
`
`Vamp
`
`T
`
`FIGURE 26.10 Trapezoidal wave shaping.
`
`542
`542
`
`

`

`MULTIPLEX WIRING SYSTEMS
`
`26.11
`
`There are a numberof hardware constraints that affect network synchronization andoscil-
`lator tolerance. The values given in Table 26.2 are calculated without considering these con-
`straints because they are not generally considered a factor for evaluating encoding
`techniques. For all encoding techniques, the same nominalbit rates or average bit rate, as in
`the case of VPWM,was used. The small decrease in data rates for 10-Bit NRZ, Bit-Stuf NRZ,
`and E-MAN,due to the addedbits for synchronization, is normally neglected; Le., Baud rate
`was used.
`The technique used by the receiver to detect a synchronizing transition plays a role in
`determining oscillator tolerance. Many different sampling or integration techniques could be
`used for a comparison, but for the sake of obtaining a reasonable judgment, for the encoding
`techniques under consideration, a very simple pulse width counter technique was assumed. A
`12.5 percent of minimum pulse width (PW min) was assumed for variability in integration
`time (IT). The maximum time for synchronization waseither the maximum pulse width (nom-
`inal) or time (nominal) between synchronization transitions.
`Equation (26.2) in Fig. 26.11 yields the natural oscillator tolerance for the encoding tech-
`nique. Figure 26.11 illustrates the maximum fast clock and minimum slow clock that can
`determine the logic value, either a “1” or a “0”, for the symbol decoded by the symbol
`decoder. The IT is the time of uncertainty in determining the pulse width. The example
`demonstrated is for PWM encoding technique. All the other encoding techniques follow the
`same method. The actual tolerance would be affected by the application and the specific hard-
`ware used in the network. The variabilities introduced by the specific hardware will be needed
`to adjust parameters in Eq. (26.2) in orderto find the final node oscillator tolerance.
`
`Nominal Clock
`
`(Logic "1")
`
`|<———— PW max ————> | <—-PW min—>/|
`|
`
`|<—PW min—>|<——-—— PW max —————> |
`
`Nominal Clock
`
`(Logic "0")
`
`|
`
`|<—— PW max ——>|<-—PW min—>|
`
`Fast Clock
`(Logic "i")
`
`Slow Clock
`
`(Logic "0")
`
`|<—— PW min ——>|<—__—_——_—____ PW Max ——————_—_>|
`
`|
`
`1% =
`Tol%
`
`PW max — PW min -IT a
`PW max + PW min
`
`100
`
`|
`
`(26.2)
`
`FIGURE26.11 Exampleof oscillator tolerance calculations.
`
`There are generally three types of oscillators used with vehicle multiplex circuits: quartz
`crystal for very tight oscillator tolerances; ceramic resonators for low-cost, tight tolerances
`and fast startup time; and RC oscillators for very low cost and very fast startup time at a very
`loose oscillator tolerance.
`The noise filter used is usually a digital filter or some type of sampling process. RCfilters
`are usually not used because they are not precise enough.Forall of the encoding techniques,
`a 12.5 percent of minimum pulse width was used for IT in calculating the oscillator tolerance.
`
`543
`543
`
`

`

`26.12
`
`SAFETY, CONVENIENCE, ENTERTAINMENT, AND OTHER SYSTEMS
`
`In a single-wire network, ground offsets between nodes cause an added received pulse-
`width variability. This condition is especially acute when trapezoidal waveforms are used to
`reduce EMI. Figure 26.12 illustrates the pulse-width timing (T) variability introduced by
`ground offset. This pulse-width variability must be accounted for becauseit causes a reduction
`in minimum pulse width and an increase in maximum pulse width, thus having the effect of
`reducing oscillator tolerance.
`
`Offset
`
`Max
`Nom
`
`
`
`
`
`
`
`Min
`
`FIGURE26.12 Ground offset on pulse-width timing.
`
`The output drivers are the source of another pulse-width variability. The effect is the same
`as with ground offset only not nearly as acute. The problem is caused byline drivers used to
`permit arbitration having a longer delay time when going from the active to passive state than
`from a passive to active condition.
`Data integrity is judged on a scale from poor,fair, good, superb, to perfect. Data integrity is
`generally considered to be affected by the EMI. The ambient levels of EMIin a vehicle are
`very low. This condition must remain in order to operate a communications receiver (con-
`sumerradio) in a vehicle. The problem is that very high levels of “bursty” noise for a short
`period at random intervals can completely disrupt multiplex communicationsfor the duration
`of the noise. The only other effect of this bursty noise is a barely noticeable pop in the radio
`speaker. During these events, data integrity is compromised. The accepted practice for data
`communication (Class B) multiplexing is to simply detect this data corruption and throw out
`the full message rather than try to recover the data. This practice is acceptable because the
`amount of corrupted data compared to noncorrupted data is considered negligible, and/or can
`be retransmitted without causing bus bandwidth problems.
`A thorough noise immunity study is very complex, and the criteria for judging would con-
`sider manyfactors. If something is known about the noise environmentand the detector hard-
`ware, as is the case with the automotivesituation, a study of data integrity may be useful. For
`the purpose of this discussion, assume that the criteria for judging which data-encoding
`methodis acceptable is mainly dependenton its natural ability to detect corruption. Also,the
`corruption detection ability is often determined by the interface hardware capability andits
`message-handling protocol.
`A numberofvalidation tests can be performed on the messagelevel. Bit-error algorithms
`such as a parity bit, checksum, or CRC are the most commontest. Also, some protocols can
`perform message length by either message type or defining the message length in the data.
`These and other messagelevel tests are independentof the bit-encoding method and should
`notinfluence data integrity of the bit-encoding technique.
`The naturalability of the encoding technique to detect corruption is known as invalid bit
`detection. Usually three types of data integrity factors are considered for vehicle multiplexing
`because the effects of EMI environmentare basically known:
`
`1. Low passfiltering. For this factor, the data bit is passed through a low passfilter,i.e., an
`integrator; the longer the shortest pulse duration, the more effective the filtering.
`2. The bursty noise detection test checks for a short duration of EMI.
`3. Two independentdata bit tests confirm valid data. PWM,for example, has two unique sam-
`ple periods per bit and both periods must complement each other.
`
`544
`544
`
`

`

`MULTIPLEX WIRING SYSTEMS
`
`26.13
`
`26.2.1 Pulse-Width Modulation (PWM)
`
`The PWM encoding technique® is composed of two sample periods or phases (T, and T;) per
`bit, as shown by Fig. 26.13. PWM encoding has the advantage that the time per bit remains con-
`stant, but has the disadvantage of generating more EMIbecause it has two transitionsperbit.
`This time per phase of PWMalso affects the generated EMInoise and, to minimize the EMI
`effect, one phase time is usually defined to be two times the other phase time in duration.
`
`Active
`
`Passive
`
`L_|
`
`| T,
`
`|
`
`T
`
`|
`
`LIL
`
`|
`
`Ty
`
`| 2 |
`
`"on
`Bit Logic Value
`FIGURE26.13 PWM encoded data.
`
`Wye
`
`Arbitration ofPWM. PWM hastheability to perform bit-by-bit arbitration. Figure 26.14illus-
`trates that a “0” bit dominates and takes priority when bit-by-bit arbitrating over a logic “1”.
`
`Logic "0" Bit
`
`Logic "1" bit
`
`Logic
`Results:
`"oO" has priority
`over a logic "1"
`
`Active
`
`Passive
`
`Active
`
`Passive
`
`Active
`
`Passive
`
`|
`
`FIGURE26.14 A logic “0”arbitrating with a logic “1”.
`
`|
`
`Data Integrity ofPWM. Consider a hardware sampler that has the capability of detecting (1)
`transition and (2) phase () every sample window,as shownin Fig. 26.15. The samplerstarts sam-
`pling at a transition, then sequentially samples window 1, window 2, and then window 3. If a
`transition is not detected by window3, then data has been corrupted and the messageis thrown
`out. When this type of sampler is used for PWM encoded data the sampler would sample five
`windows per bit and yield dual transition and phase information per bit. The transition and
`
`“7-704
`
`xN\foN
`
`
`
`Window
`
`1
`
`2
`
`3
`
`Error
`Short
`Long
`
`1+ ¢
`0
`0
`
`X
`1l1+¢
`Oo
`
`x
`xX
`1+@¢
`
`FIGURE26.15 A three-window sampler.
`
`545
`545
`
`

`

`26.14
`
`SAFETY, CONVENIENCE, ENTERTAINMENT, AND OTHER SYSTEMS
`
`phase information must be consistent for a correct PWM encoded data or corruption would be
`assumed and the message will be thrown out. If at any time a transition of either phase is
`detected in sample window1, the data has been corrupted and the messageis thrown out.
`PWMencodingis judged to have very good “perfect” invalid bit-testing capabilities even
`though the effectiveness of the low-passfilter is poor. Otherwise, it has two of the three (e-g.,
`dual periods confirmation of data and burst noise) validation tests. PWM is “perfect” encod-
`ing technique used in vehicle multiplexing when data integrity has the highest priority. How-
`ever, this encoding technique has multiple transitions per bit and would not allow operation
`at data rates near the natural EMI limits for single-wire or twisted pair transmission media.
`
`26.2.2 Variable Pulse-Width Modulation (VPWM)
`
`VPWM(sometimes referred to as VPW modulation) is a variation of PWM. Normal PWM
`has two phases per bit as shown by Fig. 26.16. T; is illustrated as a passive short and T, as an
`active long. This combinationis defined as a logic “0”bit. Notice that a logic “0”bit takes pri-
`ority when arbitrating over the opposite pattern of a passive long and an active short.
`
`bit n-1 —>|<—— bit n —>|<-— bit n+1
`
`Passive
`
`|
`
`TT,
`
`|
`
`T,
`
`|
`
`FIGURE 26.16 A PWM encoded logic “(0)”bit.
`
`Oneof the most attractive features of VPWMis that the pulse-width variability introduced
`by factors such as ground offset and output driver timing variabilities can be added to the
`pulse-width timing without severely reducing the oscillator tolerance.
`
`VPWMencodeseach phase as a databit. Figure 26.17 illustrates a
`Arbitration of VPWM.
`passive short arbitrating with a passive long. Figure 26.18 illustrates an active long arbitrating
`with an active short. In both cases, a logic “O” takes priority over a logic “1” bit. Therefore,
`arbitration using VPWM data encoding can be achieved.
`
`A passive Short
`
`is defined as a
`
`logic "0" bit
`
`long
`A passive
`is defined as a
`logic "1" bit
`
`Active
`
`Passive
`
`Active
`
`Passive
`
`Active
`
`Passive
`
`|
`
`|
`
`|
`
`Results:
`
`A passive
`
`short has priority
`
`over a passive long
`
`FIGURE26.17 A passive short arbitrating with a passive long.
`
`VPWM is utilized by SAE J1850; i.e., it uses a pulse width of 64 us for a short and 128 us
`for a long and approximates the same average data rate (10.4 Kbps) as regular PWM using a
`pulse width of 32 us for a short and 64 us for a long. The VPWM minimum pulse width for a
`short is 64 [ts and permitsa rise time of 16 us for T as illustrated in Fig. 26.12. Compare this
`rise time to conventional PWM wherea 32-us short permits only an 8-s rise time, The result
`of the proportionally longer rise time and wave shaping is an approximate 9-dBV improve-
`ment in EMI over PWM.The disadvantage of VPWM encodingis that the data rate per byte
`
`546
`546
`
`

`

`An active long
`
`is defined as a
`
`logic "0" bit
`
`An active short
`is defined as a
`logic "1" bit
`
`An active
`Results:
`priority
`long has
`over an active short
`
`MULTIPLEX WIRING SYSTEMS
`
`26.15
`
`|
`
`Active
`
`Passive
`
`Active
`
`Passive
`
`Active
`
`Passive
`
`FIGURE26.18 An active long arbitrating with an active short.
`
`transmitted will vary in time depending on the data value. The microcomputer interfacing
`transmitter/receiver polling rate with VPWM is required to be less than = 512 us per byte,
`whereas with PWM,less than ~ 768 ps per byte is required.
`
`VPWMhas gooddata integrity by sampling up to three times
`Data Integrity of VPWM.
`every pulse width as shown in Fig. 26.15. The sampler is designed to detect a transition and
`phase (@) every sample window and have an average numberof 2.5 samplesperbit. If at any
`time a transition of either phase is detected in sample window1, the data has been corrupted
`and the messageis thrown out.
`The sampling sequenceis initiated by a transition of either phase. A short symbolis then
`sensed by not detecting a transition (0) in window 1 and detecting a transition (1 + 6) and
`properphase in window 2. Sampling in window3is a “don’t care”(X), because samplingis ter-
`minated and the procedure is repeated. Window3 is actually window 1 on the next sampling
`sequence.
`A long symbolis likewise sensed by not detecting a transition (0) in window 1 or 2 and
`detecting a transition (1 + o) and proper phase in window3.
`The proper phase detection, when a transition is sensed, is used to ensure that the
`sequence does not get scrambled. It is also used to define which logic level, “1” or “0”, has
`been received.
`VPWMhasbeen judged to have “good”data integrity because low-passfiltering of the data
`pulse width is good. Every transitionis validated by an error-sampling window,and pulse dura-
`tion measurementsare validated by the proper phasetest, i.e., two of the three invalid bit tests.
`
`26.2.3 Standard 10-Bit NRZ
`
`This is an asynchronousserial I/O (standard UART)10 bits-per-byte of data. A start bit and a
`stop bit are added to provide data byte synchronization. The standard UARTused in RS232
`is bit-ordered least significant bit (LSB) first. Figure 26.19 illustrates this 10-bit NRZ wave-
`form. Vehicle multiplex networks that use 10-bit NRZ makeuse of the available hardware
`that have this asynchronousI/O,i.e., the serial communications interface (SCI) available on
`many microcomputers.
`
`START
`
`DATA
`
`STOP
`
`
`
`FIGURE26.19 A 10-bit NRZ waveform (LSBfirst).
`
`547
`547
`
`

`

`26.16
`
`SAFETY, CONVENIENCE, ENTERTAINMENT, AND OTHER SYSTEMS
`
`Arbitration of 10-Bit NRZ. Figure 26.20 illustrates 28H from transmitter #1 arbitrating
`with 44H from transmitter #2. An active “0”is defined to win arbitration over a passive “1”.
`Take note that bus has a 28H onit and, therefore, transmitter #2 shuts off its output whenit
`tries to transmit bit #2. This loss of arbitration action is the usual method employedbyall arbi-
`trating protocols.
`
`XMIT #1
`
`XMIT #2
`
`BUS
`
`1
`
`oO
`
`1
`
`0
`
`1
`
`0
`
`0
`
`0
`
`0
`
`QO
`
`0
`
`1
`
`0
`
`1
`
`QO
`
`1
`
`2
`
`1
`
`2
`
`0
`
`2
`
`3
`
`0
`
`3
`
`1
`
`3
`
`4
`
`0
`
`4
`
`0
`
`4
`
`5
`
`0
`
`5
`
`1
`
`5
`
`6
`
`1
`
`6
`
`Q
`
`6
`
`7
`
`0
`
`7
`
`0
`
`7
`
`QO
`
`0
`
`0
`
`FIGURE26.20 Arbitration of a 10-bit NRZ network.
`
`Data Integrity of 10-Bit NRZ. The bit-synchronized sampling technique used for PWM
`and VPWMcould be used for 10-bit NRZ encoded data, but it would be much more complex
`because it could be guaranteed to synchronize only on the stop-to-start transition. It would
`not sample only onebit, it may have to sampleall eight. The normal techniqueis to use a start
`bit detector to sense a valid start bit and then sampleall eight of the data bits sequentially.
`As with all decoding techniques, the data is passed through a low-passfilter. However,fil-
`tering for NRZ is very effective because of the long data pulse duration. 10-bit NRZ encod-
`ing has been judged to have “fair” data integrity, mainly because the addition of a transition
`between the stop bit and start bit is not unique in the sequence and the detector hardware
`could get scrambled. Adding an invalid bit detection for bursty noise between data bits would
`improvethe data integrity, but it could not validate every bit.
`
`26.2.4 Bit-Stuf NRZ
`
`Bit-stuffing is another way of synchronizing NRZ encoded data. The conceptis to insert a bit
`(.e., two transitions) after a specified number(X)ofbits of contiguous “1” or “O”bits andif the
`X + 1 bit is the same logical value as the other contiguousbits. The proper numberof contigu-
`ousbits is chosen as a compromise betweenthe oscillator tolerance and the generated EMI. The
`higher the number of contiguousbits before inserting a stuff-bit, the better the synchronizing
`oscillator tolerance must be, but the lower the EMI.Thereis also a receiver decoding complex-
`ity consideration with bit-stuf NRZ. Figure 26.21 illustrates waveform used by CANfor X = 5.
`The numberof stuff-bits in a frame is dependent on the data value, and the transmission
`time for a data byte is a variable quantity depending on this data value. A numberofthe fac-
`tors used to evaluate bit-stuf NRZ require a knowledge of the average numberof data bytes
`per stuff-bit (see Fig. 26.22). This average number can be derived because the nature of a data
`bit in a message has equal probability of being a “1” or “0”.
`
`Arbitration of Bit-Stuf NRZ. As with all arbitrating encoding methods, the bit-stuf NRZ
`transmitter utilizes a driver that has an active state anda passive state, thereby supportingbit-
`by-bit arbitration.
`
`548
`548
`
`

`

`MULTIPLEX WIRING SYSTEMS
`
`26.17
`
`ORIGINAL BIT STREAM
`
`(TRANSMITTER)
`
`ofafo ofa 12222 afo of:]o 0 of:
`
`Sixth Contiguous "1"
`
`BIT STREAM ON BUS
`
`ofa]o 0/2 12 2 a]o[1 afo ofalo o of3|
`
`L Opposite Bit "Stuffed" in Stream
`
`ORIGINAL BIT STREAM
`(TRANSMITTER)
`
`0/11/00 000 0/1 11;0 0/1/00 0 O;1
`
`Sixth Contiguous "0"
`
`BIT STREAM ON BUS
`
`ofa]o oo o ofzjo[a 2 alo ofalo o of3
`
`L Opposite Bit "Stuffed" in Stream
`
`FIGURE26.21 NRZ bit-stuffing.
`
`The probability of there being a stuff-bit before 8; in the
`stuffing algorithm where X =5is as follows:
`
`0 when
`i<s
`_
`P; *|1/32 when i =5
`1/64 wheni>s
`
`Bor Bye ee B, ee Bh-1
`
`= No. of BYTES/STUF
`F(N)
`
`F(N) = _—*
`where
`yp
`n=WNo. of Bits
`s°
`N= 3 = No. of Bytes
`lim F(N) = 8
`N wo
`
`F(12)
`
`= 8.348
`
`FIGURE26.22 Derivation of average numberof stuff-bits.
`
`Data Integrity Bit-Stuf NRZ._ The bit-synchronized six-window sampling technique illus-
`trated in Fig. 26.23 could be used for bit-stuf NRZ encoded data and would be guaranteed to
`synchronize on the stuff-bit. It must accommodate sampling from oneto five data bits sequen-
`tially. All bits of data prior to sampling a transition are assigned the same logic level as the
`level detected by transition and phase detector. If transition is detected in window #1, which
`is due to bursty noise, the message would be thrown out. This detector could somewhat
`improvethe dataintegrity, but it could not validate every bit. As with all decoding techniques,
`the data is passed through a low-pass filter. However,filtering for NRZ is very effective
`because of the long data pulse duration.
`Bit-Stuf NRZ encoding has been judged to have “fair” data integrity, mainly because the
`addition of a synchronizing stuff bit is not unique in the sequence and the detector hardware
`could get scrambled.
`
`549
`549
`
`

`

`26.18
`
`SAFETY, CONVENIENCE, ENTERTAINMENT, AND OTHER SYSTEMS
`
`TOP WT TT UT TPE TTA
`
`SUNTRUS RL
`
`
`
`Window
`Data
`
`Error
`Bit 1
`Bit 2
`Bit 3
`Bit 4
`Bit 5
`
`1
`
`1+@
`0
`0
`0
`0
`0
`
`2
`Dy
`
`x
`1l+¢
`0
`0
`0
`0
`
`3
`Det
`
`x
`x
`1+¢
`0
`0
`0
`
`4
`Da+2
`
`xX
`x
`x
`1+¢
`0
`0
`
`5
`Da+3
`
`xX
`x
`x
`xX
`1it+¢
`0
`
`6
`Dass
`
`x
`x
`x
`xX
`xX
`1+¢
`
`FIGURE26.23 A six-window sampler.
`
`26.2.5 L-Manchester(L-MAN)
`
`The L-MANencoding technique is composed of two sample periods of opposite phases per
`bit, as shown by Fig. 26.24. L-MANencoding has the advantage that the time per bit remains
`constant, but has the disadvantage of generating more EMIbecauseit can have an average of
`one-and-a-half and maximum of twotransitionsperbit.
`
`CLOCK
`
`Active
`
`Passive
`
`|
`
`|
`
`|
`
`|
`
`|
`
`|
`|
`
`|
`
`Bit Logic Value
`
`won
`
`nyH
`
`FIGURE26.24 L-MANencoded synchronizing bits.
`
`A transition from active to passive level will be decodedasa logic “0”andatransition from
`passive to active level will be decodedas a logic “1”. A synchronizing transition is always gen-
`erated in the center of the bit period but may not be generated at the beginning of a bit
`period, depending on the data. Asillustrated by Fig. 26.25, when there is a “0” to “1” or “1”to
`“0” data sequence the transition is not generated.
`
`CLOCK
`
`Active
`
`Passive
`Data
`
`0
`
`oO
`
`0
`
`1
`
`0
`
`1
`
`1
`
`0
`
`FIGURE 26.25 One byte of L-MANencoded data.
`
`550
`550
`
`

`

`Active
`.
`Passive
`xmit #1 Logic Value
`Active
`Xmit #2
`Passive
`.
`.
`Xmit #2 Logic Value
`Active
`
`Passive
`
`MULTIPLEX WIRING SYSTEMS
`
`26.19
`
`|
`|
`|
`
`|
`
`tom |
`
`|
`
`Arbitration of L-MAN. Figure 26.26 illus-
`trates how a logic “0”arbitrating with a logic
`“1” always wins arbitration. This situation is
`true because the active portion of the signal
`overrides the passive portion and also shuts
`off the output of the passive contender
`before it would becomeactive.
`
`ee |
`|
`
`Data Integrity of L-MAN. Consider a
`three-window sampler that has the capability
`of detecting a transition (1) and phase ()
`every sample window as shownin Fig.26.15.
`
`Ly Thecircuitstartssamplingatatransitionand
`
`then sequentially samples window 1, window
`2and then window 3.If a transition is not
`|
`"0"
`|
`Bus Logic Value
`FIGURE 26.26 A logic “0” arbitrating with a logic Cetected by window 3, then data has been
`ey.
`corrupted and the message is thrown out.
`When this type of sampler is used for L-
`MANencodeddata, the sampler would sam-
`ple four to five windowsper bit and yield complementary transition and phase information
`per bit. If D, and D,,, have the samelogic level—.e., “1s” or “Os”—then a transition must not
`occur at the bit boundary or a corrupted data would be assumed and the message will be
`thrownout.If at any time a transition of either phase is detected in sample window1, the data
`has been corrupted and the message is thrown out.
`L-MANencoding is judged to have very good “perfect” invalid bit-testing capabilities
`even though the effectiveness of the low-passfilter is poor. Otherwise, it has two of the three
`validation tests: dual validation of data by proper periodsat the bit boundary and bursty noise
`test every bit. L-MAN has “perfect” encoding technique used in vehicle multiplexing when
`data integrity has the highest priority. However, this encoding technique has multiple transi-
`tions per bit and would not allow operation at data rates near the natural EMIlimits for sin-
`gle-wire or twisted pair transmission media.
`
`26.2.6 E-Manchester (E-MAN)
`
`E-Manchester, or enhanced manchester, utilizes an L-Manchester encoded data bit for syn-
`chronization combined with three bits of NRZ encoded databits. Figure 26.24illustrates the
`L-Manchester synchronizing bit values and Fig. 26.27 is an illustration of four bits of EMAN
`encoded data.
`
`Clock
`
`Bit Logic
`Value
`
`Passive
`
`|
`
`|
`
`"wo
`NRZ
`
`|
`
`|
`
`wpe
`NRZ
`
`|
`
`{|
`
`wow
`NRZ
`
`|
`
`|
`
`wz
`MAN
`
`|
`
`|
`
`FIGURE26.27 Fourbits of E-MANencoded data bits.
`
`Arbitration of E-MAN. When arbitrating, an active (logic “O”) takes priority over the
`opposite pattern (logic “1”) for both the NRZ bits and the MAN encoded synchronization
`bit. This domination of a logic “0” over a logic “1” is easily understood for the NRZ encoded
`
`551
`551
`
`

`

`26.20
`
`SAFETY, CONVENIENCE, ENTERTAINMENT, AND OTHER SYSTEMS
`
`portion of an E-MANencodedbyte. The dominanceof a logic “OQ” synchronization bit, which
`seems confusing, can be easily realized by considering the fact that the active portion of the
`signal overrides the passive portion and also shuts off the output of the passive contender
`before it would becomeactive.
`The main advantage of E-MAN over PWM or VPWMisthat the shortest pulse width at
`10.4 Kbpsis = 76.8 us, and permits a 42 percentrise time of 32.3 us compared to a 32 us short
`and 13 us rise time with conventional PWM.Theresult of the proportionally longerrise time
`and wave shaping is an approximate 11-dbV improvement in EMI over PWM. Another
`advantage of E-MAN over VPWMis that the data rate per byte transmitted is constant. The
`microcomputer transmitter/receiver polling rate with E-MANis required to be a constant of
`less than ~ 768 1s per byte.
`
`If the same sampling technique is used for E-MANaswas used
`Data Integrity of E-MAN.
`for the bit-stuf NRZ case,it is capable of detecting (1) a transition and (2) a phase (@) every
`sample window. The E-MAN sampling hardware can sample upto six times every pulse width
`as shownin Fig. 26.23.
`The sampling sequenceis initiated by a transition of either phase.If there is a transition (1
`+ ©) in window2, then the value of D, is determined by the logic level and phase.
`If there is a transition (1 + ©) in window 3 in the sampling sequence, then D, = D,,,, and is
`determined by the logic level and phase.If thereis a transition (1 + ) in window4 in the sam-
`pling sequence, then D, = D,.1 = D,+2 and is determined by the logic level and phase.
`If there is a transition (1 + 6) in window5 in the sampling sequence, then D,, = Djs; = Dns
`2 and is determinedby the logic level and phase. The value of D,,,3 1s determined by the phase
`() because it is the L-MANencodedbit and will be confirmed by a transition (1 + 6) in win-
`dow 2 on the following sampling sequence.
`If thereis a transition (1 + 6) in window 6 in the sampling sequence, then D, = Drs, = Dns
`2 and is determined by the logic level and phase. The value of D,,.; is determined by the phase
`() because it is the L-MAN encodedbit.
`The L-MANencodedbit is not unique in the sequence and the detector hardware must
`keep track of where the L-MANbit should be, because if a transition (1 + ©) for the L-MAN
`encodedbit falls in a window other than window6, then the value of D,,.3; is determined by
`the phase (6) of the transition.
`E-MANencoding has been judged to have “fair” data integrity, mainly because the L-
`MANencodedbit is not unique in the sequence and the detector hardware could get scram-
`bled. If a transition is detected in window 1, an invalid bit was detected, which helps data
`integrity but it could not validate every bit. As with all decoding techniques,the data is passed
`through a low-passfilter. However, filtering for E-MANis very effective because of the long
`data pulse duration.
`
`26.2.7 Modified Frequency Modulation (MFM)
`
`Modified frequency modulation (MFM), a modulation technique developed duringthelatter
`1960s, was used in disk drives and is adaptable to vehicle multiplexing. The advantage of using
`the MFM encoding techniqueis that it would be synchronous with an average of 0.75 and a
`maximum of 1 transition per bit. The encoding technique permits a transition rise time that
`can be maximized and wave-shapedto significantly reduce EMI. Disk drives have a similar
`requirement where the modulation technique allows pulses to be recorded onadisk at maxi-
`mum density. The diagram shown in Fig. 26.28 demonstrates one method of applying MFM
`encoding technique to a data communication network.
`The rule for encoding simply causes a transition at the data time when the dataat that time
`slot is a logic “1”. A transition Is also generated at the clock time slot when the data before and
`after the time slot wasa logic “OQ”(or two “0”s in a row).
`
`552
`552
`
`

`

`Bit #
`Time
`Data
`
`Waveform
`
`Cc
`
`PoN
`
`c
`
`FOO
`
`c
`
`oun
`
`Cc
`
`Poe
`|
`
`MULTIPLEX WIRING SYSTEMS
`
`26.21
`
`c
`
`oUwW
`
`c
`
`OUN
`
`c
`
`oUP
`
`c
`
`HOO
`
`c
`
`D = Data Pulse
`
`Cc = Clock Pulse
`
`FIGURE 26.28 MFM encodedbyte of data.
`
`Arbitration of MFM. A requirement for vehicle multiplexing is that the data from one
`device shall bit-by-bit arbitrate with the data from another device. The arbitration bit-order-
`ing is defined MSBfirst, and 00H hasthe highest priority. To support arbitration, the output
`driveris defined and designed to have an activestate that has priority over a passive state. Fig-
`ure 26.29 demonstrates the four encoding rules that can be used to generate a waveform “A”
`that always winsarbitration.
`
`Waveform "B"
`Rule 1 - Waveform "A":
`Active
`If the previous bit was
`Passive i
`a passive "0", coding the
`Waveform "A"
`0
`0
`next bit
`as
`a
`"0" will
`Active of
`always dominate.
`Passive
`Results "A" Wins Arbitr