`Ko + Kloga(D/S + 0.5) where T = time to position
`cursor using mouse {seck time}, Ky = constant time
`to adjust grasp on mouse, K = constant normaliza-
`tion factor (positioning device dependent), S = size of
`target in pixcls,,
`= distance in sereen pixels, helps
`explain our results because the ratio of the distance
`{D}} to target size (5} is emaller for pie menus. The
`fixed target distance and increased size of targets for
`pie menus decreases the rnean positioning time as com-
`pared with linear menus.
`In our experiment, the ac-
`tivation region for an item constitutes the target. All
`subjects were informed of the fact that their target
`was not necessarily the text, but the region containing
`the text target item. This was clearly understood by
`all participants. The font size for text items in both
`
`20
`
`25
`
`15
`
`16
`
`8
`7
`6
`58
`1 23 4
`Figure 5: Target location (x) vs. numberof errors (y)
`menu styles was the same, yet the target region size
`for pie menus (3500 — 6000pizels*) was on the order of
`2-3 times the size of linear menu activation region sizes
`(1000 ~ 2000pieels")}. ‘The distance from the center of
`a ple menu to an activation region is 10 pixels while
`the distance in linear memus varied fromm 13-200 pixels.
`
`Figure 5 displays the target location plotted against
`the total numberof errors across all subjects. Pie and
`linear menus seemtosuffer froma similar phenomenon
`~ errors are made niore often on items in the central
`
`CHI ’88
`
`Tens
`
`
` Using pte
` Using linear
`
`menus
`
` Meantirask
`Table 4: anmber of errors means per cell, menu type,
`and task type (all observations imcluding no errors)
`
`region of the menu display. These are the items with
`the most interaction with neighboring items [2].
`
`Repeated measure analysis of variance resulis on the
`error ratea show marginally statistically significant dif
`ferences (P = 0.087) between pie and linear menus
`(Tables 4 and 5}. No other statistically significant dif-
`ferences were observed.
`
`Subjective results obtained in the pilot shudy repeated
`themselves in the experiment. Subjects were split on
`preferring one menu type over another but those who
`preferred linear menus had no strong conviction in this
`direction and most agreed that with further practice
`
`‘Task type
`
`Menu type
`x
`
`|
`
`‘Table 5: repeated measures analysis of variance results
`for number oferrors
`
`they might prefer the pie menu structure. Those who
`preferred pie menus generally felé fairly confident ia
`their assessment and this is reflectcd in the question-
`naires.
`
`One subject complained of having a problem with menu
`drift which is the phenomenon which occurs as the re-
`sult of the cursor relocating to the relative screen lo-
`cation of the last selected target. With linear menus,
`this tends to “drift” the cursor towards the bottorn of
`the screen. This may explain the higher error rate for
`linear menus, but the same problem occurs to a lesser
`degree with pie menus. This, in fact, we believe to be
`another positive feature of pie menus: the cursor drift
`distance is minimized. Most subjects had no problems
`coping with drift in either menu style. One area of
`further research is measuring the extent and effect of
`this problern.
`
`CONCLUSIONS
`
`What does this mean? Should we program pie menus
`
`901
`
`99
`
`901
`
`
`
`CHI 88
`
`into our bitmmapped window systems tomorrow and ex-
`pect a 15-20% increase in productivity since users can
`select items slightly faster with pie menus. Pie menus
`seern promising, but more experimenis are needed be-
`fore issuing 6 strong recommendation.
`
`First, this experiment only addresses fixed length menus,
`in particular, menus consisting of 8 items - no more,
`no less. Secondly, there remains the problem of in-
`creased screen real estate usage, In one trial a subject
`complained because the pie menu obscured his view of
`the target prompt message. Fimally, the questionnaire
`showed that the subjects were almost evenly divided
`between pie and linear menus in subjective satisfac-
`tion. Many found it difficult to “home in on” a par-
`tieular item because of the unusual activation region
`characteristics of the pie memu.
`
`One assumption of this study concerns the use of a
`mouse/carsor control device and the use of pop-up
`style menus {as opposed to menus invoked from a fixed
`screen location or permanent menus}, Certainly, pie
`menus can and in fact have been incorporated to use
`keyed input [7] and fixed “pull-down” style presenta-
`tion (the pie mera becomes a semicircle menu). These
`variations are areas for further research.
`
`co]
`pie meaus is the lout on
`One contimung issae with
`the mumber of items that can be placed in a circu-
`
`f
`
`
`
`
`Paste
`
`Figure §: Advanced “pie” menus
`
`lar format before the size of the menu windowis im-
`practical. Perhaps,
`like the limiting factors in linear
`menus concerning their lengths, pie menus reach a sim-
`ilar “breaking point” beyond which cther menu styles
`would be more useful. Hierarchical organization, ar-
`bitrarily shaped windows (Figure 6), numeric item as-
`signment and other menu refinements as well as further
`analysis is contained in [7]. Pie menus offer a novel al-
`ternative worthy of further exploration.
`
`ACKNOWLEDGEMENTS
`The authors wish to thank the following people for
`their invaluable help in the preparation of the exper-
`iment, analysis of results and statistics, and this pa-
`per: dim Purtilo, Nancy Anderson, Ken Norman, John
`
`700
`
`Chin, Linda Weldon, Mark Feldman, Mike Gallaher,
`Mitch Bradley, and Glenn Pearson.
`
`REFERENCES
`
`[1} Adobe Systerns Inc. Pastscript Reference Manual,
`Palo Alto, Calif., 1985,
`
`2] Card, S.K. User perceptual mechanisms in the
`search of computer command menus, In Proceed-
`sags - Human Factors in Computer Systems 1982
`(Gaithersburg, Md., Mar. 15-17}. ACM, New
`York, 1982, pp. 196-196.
`
`and Newell, A.
`3] Card, 5.K., Moran, T.P.,
`Lhe Psychology of Human-Computer Interaction,
`Lawrence Erlbaum, London, 1983.
`
`4] Dray, S.M., Ogden, W.G., and Vestewig, RE.
`Measuring performance with a menu-selection hu-
`rnan computer interface Proceedings of the Hu-
`man Factors Society: 25, Annual Meeting 1981
`(Rochester, N.Y., Oct. 12-16). Human Factors So-
`ciety, Santa Monica, Calif, 1981, pp. 746-748.
`
`{5} Gettys, J. and Newman, B., ¥ Windews. MIT,
`1385.
`
`[6] Gosling, ]. Ne WS: A Definitive Approach to Win-
`dow Systems Sun Microsystems Corp., Mountain
`View, Calif., 1986.
`
`and Weiser, M.
`J.,
`[7] Hopkins, D., Callahan,
`Pies:
`Implementation, Evaluation and Applica-
`tion of Circular Menus. University of Maryland
`Computer Science Departinent Technical Report,
`1988.
`
`{8] Kirk, BR. Experimental Design: Procedures for
`the Behavioral Sciences. Brooks-Cole, Belmont,
`Calif, 1968.
`
`{9] McDonald, J.E., Stone, J.D., and Liebelt, L.S.
`Searching for iterns in menus: The effects of orga-
`nization and type of target. Proceeding of the Hu-
`man Facters Society: 27th Annual Meeting 1983
`(Norfolk, Virginia, Oct. 10-14). Human Factors
`Society, Santa Monica, Calif., 1983, pp. 834-837.
`
`£10] Perlman, G. Making the right choices with menus.
`INTERACT ’84, First IFIP International Con-
`ference on Human Computer Interaction, North-
`Holland, Amsterdam, 1984, pp. 291-295.
`
`{ill Shneiderman, B. Designing the User Interface,
`Addison-Wesley, Reading, Mass., 1987.
`
`{12} Xerox Corporation, Inierpress Electronic Print-
`ing Standard, Stamford, Conn., 1984.
`
`902
`
`902
`
`
`
`
`EEE
`
`
` ee &
` 10:
`
`the watch face.
`Finger-write your figures on
`o-fne A pee
`SS howe
`
`WE INTRENOUCEFP AD
`
`
`
`alendie1ktaunt the
`* RGARARTESuY
`
`Agarad fi
`
`
`CAS83 ee =
`
`
`igenance amNeenie
`
`
`
`1adogesett
`
`tee RRA
`s
`
`_ waeeh: nae... kw dens
`pee
`
`ASSASSINS
`
`
`
`
`#Pramaeent Feeds Fatt Rangeurtant
`trenBieadnce nactesorebae
`
`
`
`
`
`
`seecanet
`
`Stat: A
`
`
`
`
`
`903
`
`REDIE CARD HOLDERS
`j ORDER POLL:
`
`
`
`
`
`Ou"te Rum
`
`AY Reterts Gt.. DagparimenPo: o
`PG. Bos €3, Parga, SE SOTO?
`tea 2.9800
`
` st gaix ide Bs
`
`
`
`HERE'S HOW THIS
`MARVEL WORKS
`
`RMSdaneStineabe
`
`
`
`903
`
`
`
`
`
`ae
`
`EasSTaaSCRneetaetaanaetaeaCe
`Rattaeenensaeanette
`
`LEeeeae:
`
`
`
`
`
`
`
`
`AALPALLAAPRIATDERLAAEAASELLLILIREASELSOORTELILEPLIDOSSREDTERRADRDOORDIDEGERDLELEDLLEERaaaatttanitaie
`
`Ceeereaennaeeaanaecarne:yRNngon‘aFRESEERIE
`8HnanSyeatssaat
`eee:aee
`SeSSEaUSraataaaSata
`aapa|SRR
`
`teataesaAnas2syite:aooPe
`Peeraeeeeeaa-esaneia
`ERICcESeaeaeSRB£23£
`eenaceea.
`ee:
`eRepee
`
`
`
`ipeedbreccvorerrereccrrcereenevencettercsernngteciiinlennicenteninnirneniteriianisinutasianvenannnscnadsunsnpanpaanennnnimanamnnsreamanaenierrarenienintenneanietpeetnetD
`
`apparunneucensetecsianedterenicnaiecsenianetsnentattrtareiitiectsissiiiitcitenieiieriicnipiinripennccinrritiig
`
`Lata
`
`BRETTRRRARTTT
`ERSaeetSEREALSSRa
`
`i‘ARGRRTaRaRsuasCeapanuanntaeane
`
`E$.f¢éfe:é
`
`i
`
`ii|
`
`;
`
`
`
`
`
`ELUUTULARE
`
`904
`
`
`
`
`
`
`
`
`
`a_ee2
`
`_.SS ae
`
`905
`
`905
`
`
`
`
`
`LELLLLILISLITLE,
`
`AiTAALNLTAAUTELUTLLOLLLLMELL
`
` ES SSSA
` SSSR
`LELELLDS
`
`
`
`LaateTADLiePAULALAUDATLAEALALLLELDLATADOLDDATAADDDDDEUDERLOLLLOTLDEEE
`
`906
`
`906
`
`
`
`
`
`
`8e
`ts
`
`
`LLLLETELLDLUUILLELDLPLIDELELILTETD
`
`
`
`ALLTELLALITETTDLSPETITLLL
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Wraseradfe+:
`
`
`
`
`
`
`
`
`
`
`
`
`
` SLELLAEEEULALEDLTEDELLEULLTEDLELELOUCOELEOLEOCUECEDELIDLLTAIRUATTIESTIIDD
`
`pveeLEISSLDLLLLTLIDLITTLSTIDLTDLLLGDLDLLLLDLDLLSIDDLEDSLDIPUPILEELIEEEELEELEELLAELEELSEEELELLE:tt
`
`
` LLEIELLPULLELEOEDUALETELETUTELTILLITEOTIS
`
`
`
` SLLCLLEELALOUECELOUAMMOEEEEEEEOEY
`
`907
`
`
`
`Lee
`
`
`
` TILLLITLLIITTILTE
`
`
`
`
`
`S &&
`
`
`
`
` LULYEELELELECLLLELLSTSLLLSELTYELTAELLELTEELSSEDTLTEALIESELLTLLLLTL
`
`SOLEILALLELTULALULUEULEUUCUETOLOAUALTAULLOPEELSDLEODDEEPPEDSEDPEESEERESESEEESEETPE
`
`
` SLLLLELANLPEELPUTAAOLELELALEDLLLELLLEOELLULYELTELLEY
`
`ULELELUUTLEOL
`LL
`LELLELLEEEY
`
`
`
`PLELLULEarvone
`
`
`
`
`908
`
`
`
`ite&tite,agif
`Bi
`geeenieeyeeaera,
`2
`“espa
`
`“AWS
`
`SN
`SESEOS,
`SSse
`
`4n
`
`x
`AN
`Ss “
`
`®
`:
`*
`RE ESS
`
`TOT
`
`S&
`‘8
`\x
`,x
`\8
`\x
`\8
`\S
`\¥
`SS
`
`\\==&= <&x§x.S===&S¥¥¥x¥§S
`
`Xx
`
`¥xxsSy
`
`ge
`
`909
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`MyyZTbeUIIALIPATALLSEESASDSEDIREOTLEED.OFa
`
`
`
`
`
`
`
`
`
`x*‘
`
`
`RS
`
`SSUESUSNSRRSENERTANNANEENE
`3ss
`
`
`WLELEECCECELECLEECTEEETLEELECETECELLCLCLUE
`
`
` ¥:Sy=SsSxxsggsy=SsSxxs=gsy=SsS&&sggs
`
`ffOE ARRASSRRRRCCERSCRRT
`g==E&==§&==¥=£zx
`
`xxLALOILILSTLLSILISLLLLLLISLLLLLLELLLLLLELELEEEELLESLELLLLLLIESPOOLLEPLEALIPOLLILEVLALES
`
`MU
`
`saboobooopeabrodpeaarioeeerrdddrea,YY
`
`
`
`
`910
`
`
`
`
`
`
`
`Z
`AASMy
`
`i
`
`
`
` 5¢z%3
`
`
`
`eeatrereicone
`
`LRLIPELSELITERDS
`
`edeniaebicaaamnnersantasey
`MALTULILELLULLELLDELY
`
` EE
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Uaccccrvercancecereccrecescrcacchionaeatestsecnccersapneerseeventbetnegneestlecveeetbereeet
`Leccntectsccenctescenetecretoneetradnpancisssbapereriprbacrcsert:
`
`*
`§
`§
`:
`$
`3
`s
`
`fook
`Sook
`Eo
`wok
`Boog
`:
`
`2
`os
`toy
`§
`<
`<
`:
`$
`<
`<
`=
`©
`
`¢
`
`:
`
`=
`=
`<
`£
`=
`$
`$
`$=
`
`
`
`
` BSESecanara eevee
`
`
`
`
` a
`
`
`
`
`
`
`
`PULEELERELEETECE:
`
`IIITUTELETLTELLEPELEEEES
`
`
`OLITILOLLLOLLLOESITISITS.
`
`
`
`MeeoteE
`eth
`
`
`
`
`
`
`te
`
`ALLTEL
`
`
`ids
`
`cei
`
`
`ELLEELUPELIELEOLLUEETCOELELITOT
`
`
`911
`
`€
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`coy
`HnPiSswg
`
`
`3
`
`<<&<x&x<=
`=<<==<=S
`
`pivisioneit
`*<
`
` Praddtaerbebobinnaaabs
`asiasEeEs
`
`
`
` detecccantiuecooabbreanedesban
`hesehis
`
`
`
`
`
`
` LLLUEAUALAEETEAMIALAIETOLADIIAILILLILLAIAURTIAIAURIRUPAORALLPLAPLUPAPDILAPALMEAREGAIATOAhihi
`
`
`fits
`
`
`
`
`
` ire
`
`
`
`
`
`
`
`
`
`
`eehwaddesisaEETED
`
`SSAA
`
`
`912
`
`
`
`gDaren1TLLLELELTELELLLEELED
`
`
`
`912
`
`
`
`AAAAAAAAADAADEDADELEASDEEIAEDAAAApattaDdDEDEPLIDIEDET
`
`
`
`
`
`
`ainnisiiviinsissvnsivvvinsssvnvaivininsissst90RARLLIISIIDISIILDSFYSETISELDEAASESESBSBIDESEDS9891DWIADIGGSIRETEESEEDETLELLEAALESLELEESEELASLILILITITISLILIDSLISISISLLSSSISISLSESEAAIIIISSLISDAIIISIIDbbl
`
`ncccendccededdddtddtedddddeLecesttLlaneddesteededitteetatis
`
`ID:
`LESTE
`
`EOE,
`
`
`
`
`
`Lesnanaaeasooooopeanacecccenecriveceddatetedd
`
`913
`
`
`
`
`
`
`
`
`
`LISELIAEDESEIDPESEERAEIALESRASELISAESDSEO
`
`
`
`
`
`
`
`ADDAAbeaRLLEMELOOTELLIOTCCEEttLebiititiniitietiseiaeetistipibigs
`
`
`
`
`
`
`
`
`
`
`
`ARMAGAAaapre reSan
`
`oeAASORLEANS NG
`
`
`Lee
`
`a
`
`
`
`
`RAIDEDAETTDITDALISLIDIESIASIADITEYDDEDD.
`
`
`LLTIIILAIDALAPPIRELAAAMADAATEDNERAESTEE
`
`vecenectivenints
`Cydebubehmecterievenccedeiiire
`
`
`
`914
`
`
`
`
`
`
`
`ye
`
`
`
`AN SS
`cS
`
`
`
`
`
`
`
`
`
`
`
`‘tttttt‘t t::§::::::ty.s::::::.§ x<‘$g
`
`
`
`\FELEEELPEELEDIEESEESICELSEEEELOSEOELEE!
`
`915
`
`
`
`EEVLIVIGUIGPEVLIVIGTITIITTED
`LIEIIISIIELIIEDSIDEDSEEDEDEDELDULYOUT
`
`
`
`
`
`
`OSSSER ERN
`
`+ y2z
`
`SAAR:
`cS
`Se
`:
`
`
`IAATDARDADODEDDAaDAnobamnnacnenpebeepoi:
` t¢s¥:‘¥::s
`
`
`
`
`
`$
`4
`8
`t
`%
`
`$=
`
`:
`
`£:x
`¥38ys$8gs88
`38ys$8gs88
`$8=
`
`
`
`
`
`
`
`
`
`916
`
`
`
`
`
`
`
`Z,
`
`BerGEOGSIIELSTEES,yyod
`
`rfSorefyETTeaelid
`
`917
`
`917
`
`
`
`UDC 621.371.029.55: 621.976: 621.394: 681.32
`Indexing Terms: HA.
`radio communication systems, Modems,
`Microprocessor application
`
`Bit-slice
`microprocessors
`in hf. digital
`communications
`
`S. D. SMITH, 8.5."
`B®. G. FARRELL, @.Sc., Ph.b,
`C.Eng., M.LELE.,
`K. R. DIMOND, 8.Sc., Ph.D, CEng, MELEE"
`
`Pased on a paper presented at the 1ERE Conference on
`Micraprovessars in Autemation and Cammunications held
`in Londonin January 1987
`
`SUMMARY
`
`A 2-4 kbit/s baseband modemis being designed for use at
`hu,, incorporating modulation/demodulation techniques
`that are matched to those frequencies and the problems
`associated with them. Fed by a continuous serial data
`Stream, the modulator functions are implemented wholly
`by a bit-slice microprocessor, and controlled by another
`more conventional microprocessor. Analogue output
`waveforms are generated in a d/a converter, whish is
`driven by the bit-slice machine. Demodulation is per-
`formed in a similar device, using an a/d input and giving a
`serial output.
`
`*Flectvonics Laboratories, University of Kent at Canterbury,
`Canterbury, Kent CT2 7NZ
`
`The Radia and Electvanic Engineger, Voi. 51, No. &, ne. 299-307, dune 198T
`
`7
`
`introduction
`
`Over the past ten years, devices for transmission and
`reception of data have become more digital
`in their
`realization. Not only are these devices constructed with
`more digital
`circuitry, but
`also signals
`hitherto
`transmitted on analogue schemes have been modulated
`digitally. This mode of transmission requires a modem
`which will convert
`the baseband signal
`into a form
`suitable
`for
`transmission. Design of modulators/
`demodulators which convert between data streams and
`waveforms suitable for specific types of channel has
`accelerated in recent years, onc such channel being hf,
`radio,
`This paper describes a modem. of this type, which has
`been designed at
`the University of Kent
`for use
`specifically on voiceband channels at hf and also
`discusses methods of realization. The modem is fed by a
`serial data stream at 24 kblis per second, which it
`modulates into a 3kHz baschand channel.
`In the
`receiver, afier mixing down to baseband, the second half
`of the modem uses the incoming signal to syachronize,
`and demodulates it back into a serial data stream (Fig,
`1). The modulation technique employed for this system
`is multi-channel four-phase differential p.s.k.,' both with
`and without pilot synchronization tones inserted in the
`band. Although other modulation schemes are under
`consideration to demonstrate the versatility of the
`moder, this technique is the one io be used at hftrials,
`Hf. transmission and reception have special problems
`associated with them. This is because h.f. channels are
`usually ionospheric and therefore suffer from mmiti-path
`propagation
`and
`both man-made
`and
`natural
`interference, prepertics which can cause unpredictable
`loss of data and synchronization. Unless modem
`parameters such as dala rate or bandwidth are altered,
`litth can be done to prevent
`loss of data, Loss of
`synchronization on the other hand results
`in an
`additional increase in data errors which can to some
`extent be controlled. Henee synchronization and the
`approachfar its implementation have been under careful
`scrutiny in the design of the demodulator.
`Until quite recently, nearly all modems would have
`consisted largely of analogue circuitry with a digital
`interface
`to the data
`source or
`sink. Utilizing
`microprocessors enables the construction of modems
`which are completely digital with just an analogue
`interface to the communication channel. The most
`obvious advantage in this case is the increased versatility
`of the modem, Whereas before, to change modulation
`type would have needed a major reconstruction of the
`hardware,
`the microprocessor realization reduces the
`problem te a modification in the program which it
`executes,
`
`2 Operation
`in the modulator, incoming data are packed into bytes
`which are used two or four at a time to provide sixteen or
`
`G033~7722/81/060299 4-03 $1 50/0
`@ 1987 Institution of Elactronic and Hadio Engineers
`
`918
`
`
`
`
`
`918
`
`
`
`
`
`
`
`5. D. SMITH, P.G. FARRELL and K. R. DIMOND
`
`
`
`deurBANDWIDTH SkisJe
`
`
`
`BASEBAND
`{ MOBULATOR
`
` jREONCSPaERICS
`
`}
`CHANNEL
`DATA
`
`SOURCE
`MICROPROGRAMMED
`OUTPUT
`MODEM
`
`éi
`
`Fig. 1, Schematic diagram of he modem.
`
`information. These
`thirty-iwo channels of parallel
`blocks of data arc modulated by a repeating real-tirne
`programme with period vr equal to 1/léth or L/32nd of
`the incoming serial data rate, into sixteen or thirty two
`parallel q.dip.s.k, channels all placed side-by-side in the
`3kH2 baseband. Hach channel is separated from its
`neighbour by 2/r Hz and is also at a multiple of the
`frequency2/t, In the 16-channel case, eight carriers each
`at multiple of the
`frequency 300Hz are phase
`modulated, carrying two bits of information on cach of
`four 90° spaced phases (Fig. 2). In the thirty-two-channel
`case, sixteen carriers are modulated at a time, but ihe
`period ¢ is doubled too.
`The individual carrier signals are generated from sine
`look-up tables, similar to those described in Ref. 2.
`These
`iables
`are
`sampled,
`scaled
`and summed,
`depending on the required frequency and phase, every
`1/9600th of a second. 128 samples at each frequency of
`the carriers are derived from the tables at the requisite
`phases,
`and summed ico obtain 128 samples
`for
`transmission. Another two or four bytes are taken from
`the incoming data stream and used to calculate the new
`phases for each carrier, so that the whole cycle may begin
`again. The resultant sampics are clocked through a d/a
`converter to produce the baseband modulated waveform
`(Fig. 3).
`fo contend with
`The demodulator, which has
`synchronization and. error decisions,
`is more complex
`than the modulator,
`(Error decisions
`consist of
`resolving the polarity of incoming data into ite most
`likely state, and possibly implementing any error
`detection/correction that might have been coded into the
`data.) The noise-corrupted incoming signal is sampled
`by an a/d converice at 9:6 kbaud. Samples are used in a
`synchronization algoritha: whichis arranged to provide
`the start pulses to a Fast Fourler Transform (PURE)
`routine. Output from this gives the phase and amplitude
`of each carrier, which may be compared with the
`
`AMPLITUDEI
`i
`, L
`
`ra
`Af
`
`PeoOY
`FREQUERCY IN UNITS
`ty
`Ba
`mapnnneem % TTF 204,
`
`FREQUENCY
`
`
`Big. 2. Amplitude and phase spectrum for nmulli-ch
`.
`dup.s.k.
`
`380
`
`previous phase and amplitude of the same carrier fo
`regenerate the two bits of data,
`Consider the sixteen carrier situation.
`The incoming data from the a/d converter consist of
`amplitudes of an analooue waveform sampled at discrete
`intervals of 1/9600th of a second. Without noise,
`this
`analogue signal is a sum of sixteen sine wayes of equal
`araplitudes at four possible discrete phases, At intervals
`of 1/32ndof the data rate G.e. 75 Hz) the phase of each
`carrier might change by multiples of 90°, depending on
`
`the two new bits of data if carries. Assuming it is highly
`probable that at least one of the carriers will change
`phase at every discontinuity, it is possible to gain data
`synchronization from the phase transitions. An output
`from this synchronization is used to keep an FFT. in
`step with the incoming data, A double 64-point radix-2
`F.E.T.
`routine®’
`is applied to each block of 128
`samples to produce two frequency domain samples for
`each carrier
`frequency. These
`are
`averaged
`and
`converted from complex coordinates to amplitude and
`phase coordinates from which net only the data may be
`determined, but also the rate of fading of the incoming
`signal and the frequeney/phase shift caused by b-f.
`interference.
`
`ps Ro
`
`(a) SINE SAMPLES FOR A FAEQUENCY
`fy AND PHASE d,
`
`fc} RESULTANT SANELES FROM AGDING
`(a) + (b}
`
`Thoth ft.
`Wea py
`
`ib) SINE SAMPLES FOR f,atie b
`Pig. 3. Example ofsine sample summation for two carriers,
`
`ia) OUTPUT AFTER OFA AND LOW-PASS
`FILTERING
`
`Wore the F.P.T. to take its 128 samples so that a phase
`discontinuity boundary was sorsewhere in the middle,
`the resulting data would be completely useless. In fact
`the errors rise fairly quickly with the number ofsamples
`at the wrong side of a phase transition, so it is essential
`that there is good data synchronization. This requires
`accurate data rate recovery from the incoming signal,
`which is achieved by a sliding filter algorithm in
`association with 4 local ‘flywheel’ clock. Whether this
`focal clock or the generated synchronization pulses are
`used to synchronize the transform depends on the depth
`
`of fade orthe frequency/phiaseerror, as ascertained from
`previously decoded data blocks.
`An additional technique is available for improving
`
`The Radio and Electronic Engineer, Yal. 51, No. &
`
`919
`
`919
`
`
`
`BIT-SLICE MICROPROCESSORS IN H.F, DIGITAL COMMUNICATIONS
`
`
`1K » Bd MICROMEMORY
` ———»
`
`| useo Fe
`i POP 11780
`FOR
`MICROPROSRAM CONTROL
`i SOETWARE
`
`
`LOADING too
`p DEVELOPMENT
`
`
`
`BP
` 4
`
`E900 BU-SCE
`
`6500
`1
`SERIES
`BICHO~
`
`PROCESSOR
`
`
`24 RTS
`SERIAL
`
`FAST PROCESSOR
`DATA
`
`
`
`ANALOGUE
`SIGNAL GUT
`
`{vou
`SLOW PERIPHERALS
`
`{aoe
`>
`
`beveLorend /
`a
`SIME LOOKUP TABLE
`dl
`
`GALA MEMORY
` eee
`
`Fig. 4. Modemstructure (block diagrasn),
`
`synchronization, and for minimizing the possibility of
`performing an FPLF.T. across a discontinuity. This
`involves spreading the carrier frequencies out so that
`they cover the complete bandwidth ofthe voice channel,
`rather than their frequencies being integral multiples of
`the data rate. The sampling frequency of the receiver is
`increased proportionately so that it is still an integral
`multiple of the carrier frequencies. However, the period
`between phase discontimuities in the transmitted signal
`remains a simple fraction ef the data rate. Hence the
`period during which the 128 samples are taken for the
`F.F.Y. is shorter than the time between discontinuities
`by approximately [6% for a data rate of 2-4 kHz ina
`3 kHz bandwidth.* This meansthat there is a fairly long
`period of time, acress the phase transition, over which
`no sarnples are taken for use in the FLF.T. This is
`advantageous for two reasons: (a) to allow a greater
`margin
`for
`syachronization
`error
`before
`trans
`discontinuity samples
`cause errors
`in the FLF.T.
`algorithm, and (b) the F.F.T. dees not employ samples
`near te the discontinuity where the 3 kHz bandlimiting
`causes rounding’ of the signal on citherside.
`
`3 The Modem Structure
`In both the modulator and the demodulator there are
`iwo microprocessors, A slower, one-chip microprocessor
`fromthe 6800 farhily is used to interface the modem io
`the serial data source or smmk. Hts responsibility is for the
`slower dala processing, such as packing the incoming
`serial data into bytes and encoding Ht, some of the
`synchronization mechanism in the demodulator, and the
`control functions for the fast processor. (Fig. 4.)
`This fast processor consists of a 2900-series bit-slice
`microprocessor
`to
`perform the modulation
`and
`emodulation of data, and is connected directly to the
`analogue port via its data bus. Its purpose is to convert
`
`data to samples of summed sine waves at the correct
`phases in the modulator, and to perfoun the F.FUT. and
`clock
`recovery
`in
`the
`receiver.
`Since
`as
`a
`is
`modulator/demedulator it
`repeatedly executing a
`edicated routine of known duration, there is no need
`for macro-coding and a mapping pom. as in the
`conventional bit-slice machine.* Hence all programming
`is at the microcode level. Microcode is boctstrapped into
`
`the writable microcode memory on power-up by the
`6800 processor, which in turn is fed by a host mainfrarne
`computer during microprogram development. Por a
`completed portable modem,
`the bootstrapping is from
`€.p.r.0.m.5 in the 6800's memory map.
`All data/address buses on the bit-slice are {2 bits
`wide, together with the a/d, d/a converters, while the
`width of the micropropram word is 64 bits. Pwo-level
`pipehmng and parallel hardware stacks, together with
`fast dala paths and devices isolatedfromslow data buses
`by registers alow minimizauion of processor cycle imes,
`The bit-skice machine is connected to the slow processor
`by an 8-bit bidirectional data register which is directly
`addressable in the memory map of cach machine.
`
`4 Conclusions
`
`This paper describes a modem which uses only digital
`processing to accomplish its operation. When used for
`hf.
`trials the modem demonstrates the viability of
`microprecessor controlled modulation and demodula-
`iion. Tt also reveals its versatiily to be reprogrammed
`with ease to a completely different modulation scheme.
`
`& Acknowledgments
`S. Dy. Smith would like to acknowledge the support of
`the S.R.C, and of the Ministry of Defence (Procurement
`Executive};
`the
`authors
`are prateful
`for helpful
`discussions with Mr J. Pennington (A.S.W.E.)
`
`8 References
`1 Ziemer, R. E., and Tranter, W. H., ‘Principles of Communication:
`Modulation and Noise’, Sect. 7.5 (Houghton Miffin, Boston,
`1976),
`2 Gorski-Popiel, J, (Ed), ‘Frequency Synthesis: Techniques and
`Applications’ (IEEE Press, New York, 1975}.
`3 Rabiner, L. RB. and Gold, B,. ‘Theery and Application of Digital
`
`Signal Processing’ (Prentice Hall, Englewood Chis, NL, 1975).
`y, G. 1, ‘Errer Contre! for Baia Multiplex Systems’, Ph.D.
`Thesis, University of Kent at Canterbury, 1975,
`5 Brigham, ©.
`‘The Fast Fourier Transform’
`Eaglewood Clits, NLL, 1974)
`‘Bulld a Microcomputer’, Advanced Micra Devices, Sunnyvale,
`Cab, 1979
`
`(Prentice Hall,
`
`6
`
`Manuscript received bythe Institution infinalform on 27th March 1981
`(Paper No. 1993{Comm 220)
`
`June 1987
`
`301
`
`920
`
`
`
`
`
`920
`
`
`
`
`
`
`
`A Nove} use fer Microprocessoxa in Designing Single and Hulti-tene Generators,
`Ringing Genarators and Inverters
`Authors
`
`26-3
`
`Ray Bonnett
`Engineer ing, Consultant
`Permace Assoclatas
`ai Rickey Drive
`Maynard, Ma OL754
`
`INTRORUCTION
`
`and Tone generators
`Caurxsenthy Ringing
`consigts of anaieg devices such as osciLia-
`tors and linear amplifiers, ox non-addustable
`digital oscillators and
`bandpass filters
`te
`generate
`the
`required
`frequency and wave-
`shapes.
`Some of
`the more important
`analog
`design considerations are
`signal Linearity,
`symmetry,
`frequency stabliity,
`and tempera
`ture variations that effect all of
`the above.
`
`Frequency is derived from standard os
`Later
`
`
`ewizgenits, which
`contain xesistors,
`capaci~
`tora, andfor inductors.
`in adjusting the fre-
`quency, fine
`tuning
`is dene with variable
`resistors, while more coarse
`adjustments are
`
`done
`by
`switching capacitors
`and/or
`duc~
`cers. The frequency selective components used
`in ringing generators
`have large values
`and
`are physically large.
`
`Sone
`of
`the more
`important digital de~-
`sign elements are the master
`clock and count
`dovn circuits. The
`use of digital circuits
`usually solves the problem of stability and
`symmetry but
`introduces sone
`flitering re-
`quirements because digital signals sre square
`waves which, by definition,
`ars rich in har-
`
`mnias, Propex £iltex design
`reduces harmen-
`
`es to the desixed output level, Digital fi4-
`ars at
`ringing
`frequencies contain large
`capacitance
`and/or
`inductance values
`and,
`ke the analog oscillaters, are physicaiiy
`arge. A cesonant
`transformer may be used
`as
`a filter bub it
`is physically large, and can
`only work with
`a single frequency.
`The f£il-
`ters
`fox tones need to have high 9 Values in
`orfer
`to
`guppvess the
`harmonica
`below the
`DEME band ceguirements.
`In the curfvent tech-
`noLogy,
`inkexrapters are net
`synchronized ts
`thea tone wave shapes being interrupted.
`of
`Yoday's
`technology allows
`the use
`wlerocontrollers to genezate signals. Micro-
`contrelier generate the vequired data and D/A
`converters
`transform the digital words
`te
`
`analog signals, eliminating the
`need Zor RC
`
`or Lt oscillators. Ali the required functions
`are executed in firweware which calculates the
`sine functions for magnitude and duration of
`wail
`statements necessary to obtain the fre-
`quency. @ince the mierecontroller is driven
`
`by a crystal oscillater,
`frequency stability
`can be a8 qoad as a standard quarts watch fin
`the
`order of
`092%). Frequency and voltage
`adjustments
`can
`be made
`by ReCaICULAEIN
`Micgocontroller data, The user may input the
`gata in wany ways, such
`ad a key pad, seleo-
`tor switches, cz it may be down
`loaded via a
`RS232 computer/modem port.
`DIGETAL SYNTHESIS.
`
`are
`signals
`analog
`tone
`and
`Ringing
`as a
`independent, continuous signals ywarying
`function of time, Digital signals
`are dls-
`cxete time varying
`signals. A digital signal -
`is a sequence of numbarst.
`
`614
`
`GH2028-6/9079000-083 1801.00 G1600 TREE
`
`Bari &hyne
`president
`Permace Associates
`Y Galnut Street
`Millis, Ma O2054
`
`ABSTRACGY
`
`paper describes a method for pro-
`This
`ing precise
`single
`or malti-frequency
`es
`fex
`telephone offica
`ringing genera-
`,
`tone generators, and general purpose
`nverters, by using wicxoprocessers and digi-
`tal technolody.
`Recent technical developrents have pro-
`engineers new tools for generating the
`ded
`gnais used for
`telephone equipment anc for
`permlbting remote access to the equipment for
`supervisory and diagnostic purposes. Figure 1
`illustrates
`a
`system in which wiere-
`
`controliers,
`counters,timers, Random Access
`Memory
`(RAM}, Analog to Digital Converters
`
`(A/D), and Digital to
`log Converters (0/A}
`are
`combined to produce
`tone signals, These
`signais are then amplified to produce the ce-
`quired singing or tone power.
`fhe microcontroller is programmed with a
`mathematical
`eguatten
`to derive
`timing and
`voltage
`levels fer
`the catgut
`signal, This
`equation
`is converted to digital woxds that
`aré passed to the data portion of RAM and is
`converted ta a digital word that sets a timer
`
`controlling the address portion to RAM. The
`RAM output
`is converted to
`an anaicg signal
`by the D/A converter. This signal
`is used by
`a power amplifier to condition the siqnal £or
`ugea on the telephone lines.
`Hy using a Microcontroller,
`the terms of
`the equatian (i.e. Exequencies and
`the volt-
`age levels of
`«ach Ereguency
`independently}
`ave ingut as variables,
`thus giving the
`user
`compiete
`contrel cf
`the output aignal. The
`BSSYy CAN USE a hannal control {such as a key~
`pad/readeut} cor a remote computer (through an
`RS232
`port}
`to change
`the variables. The
`yanae of
`the
`signals ia
`determined by
`the
`yesolution of
`the D/A converter and the fre-
`quency
`response of the
`power amplifier. Be~
`cause
`the Hicrocontroller is
`crystal cont-
`yelled,
`frequency vesponse and accuracy aré a
`function of
`binary resolution, fhe
`gutput
`level
`is also a
`function of Che binary reso-
`Lution and reference voltage. Muitiple
`fre-
`quency tones are generated
`by the same meth~
`ed.
`Since the
`gqvantizing freguency is much
`higher than the
`tone Erequenciles, simple low
`pass £iltering is used to eliminate unvanted
`freguencies, Using
`an &/f
`converter,
`the
`output is monitored, This same signal may be
`uged as a built-in dilaqnestic test,
`x
`a
`The
`output
`of the
`power
`amplifie
`w
`
`sensed for voltage and current output.
`IAo
`frequency applications,
`tha microcontr
`Br
`L
`can monitor
`for
`inductive
`and
`capac
`v
`loads, and make adjustments.
`The versatility of
`the microcantrolier
`
`allows a single design
`to eover a wide wari~
`ety of uses, The
`power amplifier can custom-
`ige
`the application. Other options
`such ag
`zero
`crossing interrupting would be
`undex
`contrel of the microcontroller.
`
` cad
`
`5&
`
`ii
`
`ko
`
`3
`Li
`ti
`
`921
`
`921
`
`
`
`
`
`
`
`
`
`26-3
`
`f { ?
`
`al
`
`| }
`
`z G
`
`
`
`
`
`
`cc~
`Fa | tt
`L
`——o
`ANGNG/ TONE GENERATOR:
`Figure 1
`
`NOE
`
`
`
`wt /
`
`1 i
`
`csnerated mathematical“ tormula
`The computer
`a digitized analeq signal.
`is
`tuxyned ints
`This analeg signal has quantized errear
`terms
`that
`cause Gistortian with
`to
`an
`analog generated signal.
`if the carrech oun-
`ber of points are
`chosen the error terus are
`low. These error terms contain high frequency
`signals that are
`casily Ellteread ont with
`a
`simple bandpass filter, This process yields a
`signal
`that
`is amplified and applied to the
`telephone circuits,
`Dial
`tone and vingback tone are produced
`adding
`together
`two precise
`frequency
`by
`tones. Ringback tones comprise 440 Hz and 480
`Ha; dial tones
`comprise 35
`he and
`446 Hz.
`
`The "Tone Cycle" is defined as follows: with
`all
`frequencies
`starting at
`zero degrees
`phase angle (sera volbs
`and zero current), a
`"Fone Cycle"
`is complete when
`aki
`the tones
`azxive at 8 3607 phase angle at the same time
`{i.e. mero yalts}.
`For a
`single
`frequency
`tone, quantized data
`is calculated for
`only
`ome 3607 cycle, For
`xingback tone,
`the
`com-
`puted
`da