Apollo Experience Report Voice Communications Techniques and Performance

8/8/2019 Apollo Experience Report Voice Communications Techniques and Performance

1/18

APOLLO EXPERIENCE REPORT -VOICE COMMUNICATIONSTECHNIQUESAND PERFORMANCE

by John H. Dabbs und Oron L. Schmidt , ! 'M u n n e d Spucecray? CenterHouston, Texas 77058

N A T I O N A LE R O N A U T I C SN DP A C E- . . .

A D M I N I S T R A T I O N W A S H I N G T O N , D. C. M A R C H 1 9 7 2


2/18

TECH LIBRARY KAFB. NM

OL33bbb.... ~ .~ - ~ ~ ~~~-- -1. Report No.

5. Report Date. Title and Subtitle..NASA TN D-6739 3. Recipient's Catalog No.. Government Accession No..~~" ~~~~ ~ ~

APOLLO EXPERIENCE REPORTVOICE COMWNICATIONS TECHNIQUESAND PERFORMANCE '. Performing Organization CodeMarch 1972

I"_7. Author(s) " ~ ~~~ ~ . -~ 8. Performing Organization Repor t No.John H. Dabbs and Oron L. Schmidt, MSC MSC S-313. .. .. . - ~- - ~~ ~~ . . 10 . Work Unit No.9. Performing Organization Name and Address

~~ 914-50-17-08-72Manned Spacecraft CenterHouston, Texas 77058 11 . Contract or Grant No.

." . . "" ~ ~ 13 . Type of Report and Period Covered12. Sponsoring AgencyNameandAddress~~ Techn ical NoteNational Aeronautics and Space AdministrationWashington, D.C. 20546

I14. Sponsoring AgencyCode

I- . ".""" "" ~- ~~15 .~SuppternentaryN& ~~The MSC Director waived the use of the International System of Units (SI) forthis Apollo Experience Report, because, in his judgment, use of SI Units would impair the usefulnessof the report or result in excessive cost.

~. ~~ " - ~ . ~~ ~~16. Abstract

~ " _ . . . " ~

The prim ary per form ance re qui reme nt of the spaceborne Apollo voice communications system ispercent word intelligibility, which is defined in thi s repo rt and is related to other link/channelpar am ete rs. The effect of percent word intelligibility on voice channel design and a descriptionof the verificat ion procedure s are included. Development and testing performance problems andthe techniques used to solve the problems also are discussed. Voice communications perform-ance requirements should be comprehensive and verified easily; the total system m ust be con-sidered in component design, and the necessity of voice processin g and the associated effect onnoise, distortion, and cr os s ta lk should be examined carefully.

.. - . . ~~17. Key Words Suggested by Authori s) ) ~~ ~.'VoiceCommunication 'Clipping'Wordntelligibilityoice-Operatedelay* Signal-To-Noise Ratio* Automatic Volume Control'Voice-Operated Gain-Adjusting Amplifier

18. Distribu tion Statement

L19. Security Classif. (ofhiseport) 20. Security Classif. (ofhis p a g e )

None 1 None" ~~ ~ ~~ ~ . " -For sale by the National Technical Informationervice, Springfield, Virginia 22151


3/18

APOLLO EXPERIENCE REPORTV O I C E C O M M U N I C A T IO N S T E C H N I Q U E S A N D P E R F O R M A N C E

By J o h n H. Dabbs an d Or on L . Sch mid tM a n n e d S p a c e c r a f t C e n t e rS U M M A R Y

The Apollo voice communications system is composed of many individual linksthat are interfaced to meet overall requirements. These links can be grouped into wobasic types: ground to ground and spaceborne. The ground-to-ground links constitutea siza ble par t of the total system; however, because these links are basically commer-cial designs, they are not discussed in this report. Instead, the emphasis is on space-borne links and the inte rfaces of these links with the ground. The system-performancerequirement, design aspects, and performance testing are discussed. Of particularinterest is the use of percent word intelligibility as the primary performance require-ment, the precise defini tion of word intelligibility, the effect of the requirement onvoice channel design, the relationship of word intelligibility to other link/channel pa-rameter s, and the verific ation that the requirement has been met. Many aspec ts ofvoice channel performance have been defined during the Apollo Program; however, acomplete list of de sign cr it er ia and an expeditious method of verifying the performancestill a r e unavailable.I NTRODUCTI ON

The success of the Apollo Prog ram requ ired tha t a voice-communications capa-bility be establi shed across the 389 000 kilome ters between the earth and the moon.The capability had to includenot only direct communications but alsoa full conferencethat included the two crewmen on the lunar surface, the pilot in the orbiting commandmodule, and the flight controllers in the Mission Control Center (MCC). These voicechannels were implemented using links in the very-high-frequency (vhf) band (250 to300 megahertz) and channels in the unified S-band (USB) system. The USB also pro-vided for telemetry, command, ranging, and television functions.

Early in the Apollo Program, the perfor mance requir ement for voice communi-cations was established as 90-percent word intelligibility for the prime links and70 percen t for the backup links. This performance requirement, combined with therange and functional requirements, formed the basis for the system design. The re-quirements were met, but not without many design, analysis, and testing iterations.


4/18

Results of the Apollo voice communications developmental program are discussed alongwith brief discussions of the voice channel requirements, the configurations, and theradio-frequency (rf) links concerned.Unlike commer cia l ground-to-ground communications systems, the Apollo voice-communications links were constrained by the extreme range between transmitters and

rece ive rs and by the limitations on spacecraft transmitter power, antenna gains, andobtainable receiver noise figures. These constraints caused considerable difficulty indefining minimum acceptable signal-to-noise ratio ( S N R ) performance and in determin-ing optimum modulation. Because voice is a complex signal, the evaluat ion of which islargely subjective, specific minimum design criteria do not exist, and link calculationsbased on a sinusoidal modulating signal cannot be related directly to word intelligibility.However, considerable work has been done on defining general relationships amongword intelligibility and channel bandwidths, signal power, noise power, and so forth.On the basis of this previous work, the NASA and the prime con tractor s mutually agre edon a signal- to-no ise standard for the Apollo voice channels of 14 decibels for 90-percentword intelligibility and 4 decibels for 70-percent word intelligibility when both the sig-nal and the noise powers are expressed in root-mean-square (rms) values. This stand-ard served the purpose for which it was intended; but, as was expected and laterproved, specifying a single value of SNR for a specific percent word intelligibility isnot valid. The fact also was known that preemphasis followed by peak clipping wouldinc rease t he r ms valu e of the modulating signal, thereby increasing the output rm s SNRand th e resulting intelligibility. Therefore, for the links that had insufficient rf mar-gins, these voice-processing techniques were used to obtainthe required performance.However, many pitfalls exist in the u se of vo ice processing; these pitfalls must beunderstood fully if the required improvement is to be realized.

DISCUSSIONIn this section , the performance requi rement fo r the Apollo voice communica-tions, the functional requirements, the design considerations, the systems testing, andthe conclusions are presented.

P e r f o rm a n c e R e q u i r e m e n tThe use of percent word intelligibility as the performance requirementwas oneof the more controversia l aspe cts of the Apollo voice communications system. Themeaning of the requireme nt in te rms of end performance was not unders tood. Also, nostandardized, rapid, and economical method existed for verifying that the performance

had been met. No clear-cut guidelines existed on how a system should be designed tomeet the requirement, and, likewise, no direct relationship existed with other link/channe l parameters from which circui t margins could be calcula ted. Many questionsrema in; however, some have been answered and data pertaining to others have beenderived.The use of p erc ent word intelligibility is an attempt to place a quantitative valueon voice channel performance. Word intelligibility is defined as the percent of mono-syllabic words correctly understoodafter transmission by means of the communications

2


5/18

channel. The quality of the channel is not considered, and, under certain conditions,the channel can contain high noise levels, inter fering tones, and crossta lk, yet stillsc or e high on a word-intelligibility test. Such inte rfer ing signals can be annoying, but,if the signals do not mask the speech, the effect on word intelligibility is small. Inaddition, people usually communicate using sentences or phrases, not monosyllabicwords. Because of these factors, percent word intellig ibility can be misleading, and,used alone, is an incomplet e system perform ance criterion . However, the use of SNRmeasurements, spectral analyses, crosstalk measurements, and subjective listening,along with the word-intelligibility measurements, provided sufficient performance datato develop and verify the Apollo voice systems. Generally, the 90-percent intelligibilityrequirement was met or exceeded, and the significance of this figure and the limitationsinvolved are now well understood.

The Apollo specifications state that the percent word intelligibility shall be basedon the "American Standard Method for M easurement of Monosyllabic Word Intelligibil-ity , Test Guide. '( However, this standard merely establishes general guidelines, notspecific test standards. After considerable research on industrial testing methods andafter several attempts to perform in-house word intelligibility tests, an agreement wasestablished with the U. S. Army to score tapesat the Fort Huachuca Testing Center,Arizona. At Fort Huachuca, the standard phonetically balanced monosyllabic-word-evaluation technique, highly trained listeners , and an automatic grading system a r eused. This technique produces repeatable results; however, even with the automation,the procedure is time consuming. A minimum of 15 minutes of syst em tim e per datapoint is needed to produce the tapes, and as long as 4 weeks is required to score thetapes from a test ser ies . A more rapid method of verifying voice channel performanceis desirable, and efforts to develop such a method are recommended. A standardmethod, avai lable to all organizations involved in the design and testing of voice channelcomponents and systems, should be adopted early in any future programs.

The performance requirement has been di scussed only in te rm s of ve rificationcapabilities. Of equal importance is the ability to relate the performance requirementto channel design criteria. At the beginning of the Apollo Pr og ra m, no specif icchannel design criterion w a s known fr om which a channel could be designed that wouldprovide 90-percent word intelligibility under all conditions. However, previous (usuallyempirical) effortshad produced general relationships among word intelligibility andbandwidths, distortion, signal power, noise power, and s o forth. This previous workestablished a base for the determination of the requ ired channel parame ters. Fromthese relationships, it w a s determined that the SNR had the predominant effect on intel-ligibility. Also, because SNR is a simple measurement and is included easily in linkcircuit margin calculations, a conversion between word intelligibiiity and signal-to-noise ratios was chosen. The conversion, as ag ree d upon by the NASA and the Apolloprime contractors, was 14 decibels rm s/ rm s SNR for 90-percent word intelligibilityand was 4 decibels for 70-percent word intelligibility. This relationship served thepurpose of an established standard for use in design, analysis, and testing; however,the relationship w a s not valid for ever y channel (fig. 1)because only a mean channelbandwidth of 3 kilohertz and white noise were considered. The effects of specific band-widths, non-Gaussian noise, distortion, spectral distributions, and so forth were notconsidered in this conversion. In a few instances, less than optimum design resulted(either underdesign or, equally as undesirable, overdesign), and additional work isneeded to define minimum channel-design criteria for future programs. However, withth is standard, the Apollo voice system design was simplified because optimum total-link

3


6/18

-

1

I. Up-link combinat ion 6 n o rm a l m ice .2. Up-li nk combin ation 4 l n o r m a l m i c e

command, and pseudorandom-noise range codel

3. Up- l ink combina l ion 2 lnorma l M ice)and pseudorandom-noise range codel

J I I I I I10 20 30 40 MSNR at command and service module, dBFigure 1. - Word intelligibility as afunction of SNR or up-link combina-tions 2, 4, and 6.

command,and anging

30-fl 19.1 m iUSB antenna Cross-dlpolevhf antenna

(a) Launch through translunar injection.

(voice, tele metr y, command, anging,andtelevision) design was possib le by optimumselection of tra nsmission powers, modula-tion, cable losses, receive r-noise figure s,and so forth). ' Furthermore, preemphasisand peak clipping could be added when neces-sa ry to obtain the 14-decibel r m s /r m s or4-decibel SNR requirements.

Functional RequirementsThe basic functional requirementswere established early in the Apollo P ro -gram. A s the program evolved, so did therequirements. For the first lunar-landingmission, the voice requirements were asfollows .1. Launch through translunar injec-tion: Command and service module (CSM)duplex voice with the Manned Space FlightNetwork (MSFN) .using vhf/amplitude modu-lation (AM) or t he USB sys tem (fig. 2(a))2. Translunar and transearth coasts:CSM duplex voice with the MSFN using theUSB system (fig. 2(b))

Qf U S B antennasHigh-ga in and omnid i rect iona l

command,anging,Voice, elemetry.and elevlsion

t8 5 4 125.8 m l USB antenna

MSFN . . "~(b) Translunar and transearth coast.

Figure 2. - Functional voice requirements.4


7/18

3. Lunar-orbitaloperations:a. Command and service moduleand lunar module (LM) duplex voice with theMSFN using the USB syst em s (fig. 2(c))b. Command and service module

duplex or simplex voicewith the LM usingVM/AM (fig. 2(C))c. Command and service module/MSFN/LM voice conference using theUSBsystems and MSFN relay (fig. 2(c))

4. Lunar-surface perations:Sameas lunar-orbital operations with the additionof duplex voice conference among the woextravehicular astronauts(EVA- 1 andEVA-2) on the sur face of the moon, theMSFN, and the command module (CM) pilotusing the following systems.

a. Very high frequency/frequencymodulation (FM) from EVA-2 to EVA- 1withvhf/AM relay from EVA-1 to the LM(fig.2(d))

b. Very high frequency/AM fromEVA- 1 to both EVA-2 and the LM andvhf/AM f rom th e LM to both EVA- 1 andEVA-2 (fig.2(d))c. Unified S-band between the LMand the MSFN with LM relay between theMSFN and EVA-1 and EVA-2 (fig. 2(d))d. Unified S-band between theCSM and the MSFN with MSFN relay be-tween the CSM and LM (fig. 2(d))

5. Recoveryoperations: CM sim-plex voice with recovery airc raft usingvhf/AM (fig. 2(e))In addition to these requirements,requirements existed for communicationsamong the crewmen in each spacecraft; theCSM recording of CSM communications andthe LM vhf voice with subsequent playbackby way of the USB to th e MSFN; the pre-launch voice with the launch complex; and,

Notes: 111

12 )L M

13)

command, anging,

MSFN relays mice betweenthe CSM and LM when the CSMand LM are no t w i th in l ine ofsight of each olher.E i t h e r C S M o r M can relayo ther spacecraft f a fa i l uremic e between the MSFN and theoccurs in o ne USB l ink .CSM can record LM mice ont h e f a r s id e o f t h e m w n f o rplayback to MSFN on the nearside.

\/ Voice, elemetry,command, anging,and televisionk - f t 25.8 my ; antenna... .

MSFN(c ) Lunar-orbitaloperations.

85-11 125.8 m l o r 210-11 164 m l antenna

command,andranging

Voice. elemetry.command,andtelevision

Moon - vhf lFM --VA 2

Notes: 11) MSFN elaysmicebetween121 LM relays mice belweenCSM and LM.

13) EVA-1 relays mice from EVA-2MSFN and EVA-I an d EVA-2.to LM.

(d) Lunar-surfaceoperations.Figu re 2. - Continued.

5


8/18

r n

(e) Recoveryoperations.Figure 2. - Concluded.

for contingencies, the CSM relay of voicebetween the LM o r EVA- 1 and EVA-2 and theMSFN and LM re lay of voice between theCSM and the MSFN. These requirementswere met using different conf igurat ions ofthe CSM,LM, extravehicular communica-tions sys tem (EVCS), and the MSFN commu-nications links. In addition to the voicerequirements, these links had to accommo-date requireme nts for telemetr y, command,ranging, and television.

The CSM communications equipmentconsisted of three major groups: audio cen-ters, USB, and vhf. The audio centers (onefor each crewman) provided audio-channelselection, audio channel level control andisolation, voice operated relay (VOX), andpush-to-talk (PTT) functions. The USBequipment (premodulation processor, trans-ponders, power amplifiers, and so forth)provided modulation, transmission, recep-tion, and demodulation of S-band signalsbetween the CSM and the MSFN. The USBdown-link voice w a s transmitted in threebasic modes: the normal voice on the1.25-megahertz F M subcarrier that phasemodulated the ca rr ie r after being frequencymultiplexed with a 1.024-megahertz pulse code modulation ( P CM ) subcarr ier and the-turned-around pseudorandom noise (PRN) range code (fig. 3); the backup voice, which

was phase modulated directly onto the carrier after being multiplexed with the PCMsubc ar rier and PRN (fig. 4); and the dump voice, which was frequency modulated di-rectly onto a different carrier after being combined with PCM and analog telemetrysubca rr iers. The USB down-link normal voice was preemphasized 6 dB/octave and wasclipped 12 decibels before subcar rier modulation, and the backup voice was preempha-sized 6 dB/octave and was clipped 24 decibels. The CSM USB als o could relay receivedLM vhf voice or EVCS voice/data (fig. 5) to the MSFN by combining these signals withthe CSM normal voi ce before subcar rier modulation. When relaying EVCS voice/data,the CSM voice was filtered 300 to 2300 hertz to prevent interference wi th the EVCS datasubcarriers. The USB up-link voice was rece ived by two basic modes - he normalvoice by means of a phase-modulated ca rr ie r and a 30-kilohertz F M sub carr ier and thebackup voice by means of the 70-kilohertz FM command subcarrier. A squelch w a sprovided for the normal up-link voice subcarrier. The CSM vhf equipment consisted oftwo transceivers on different frequencies and provided for transmissions and receptionof vhf signals among the CSM and the LM, the EVCS, the MSFN, and the recovery air-craft. The vhf could opera te simplex on either frequency or duplex using both frequen-cies. During lunar landing and ascent, a ranging tone (approximately 30 kilohertz) w a stransmitted along with the duplex voice between the CSM and the LM. The vhftransmitted voice was preemphasized 6 dB/octave and clipped approximately 24 deci-bels, and the resulting waveform keyed the transmitteron and off to produce maximumrf efficiency. Between words, a 30-kilohertz noise suppression oscillator or the6


9/18

Microphoneicrophone

Audioenter

co n t ro lontrolontrol ga inai nai nAutomaticutomaticutomatic

centerudio centerudio

Lelt posit ion Rightosit ionosit ionenter

h n o d u l a t i o n p ~ c e s s r"_ "" - - -I Clipper module II

1

I IreemphasisI

I II

IClipper II I

L___-_--_..___--I I

-------- ---- --II Relay module II

MicmDhone

IIII 1.024-MHzI PC MJ subc;3rrier

tTo S-bandphase-modulat iont ra n sm i t t e r

Figure 3 . - Command module normal-voice signal flow.

ranging tone (depending on mode se lect ion)modulated the vhf transmitter and providedan rf signal that quieted the receiverat theLM, the EVCS, or the MSFN.The LM communications equipmentw a s simi lar to that of the CSM. Function-ally, this equipment w a s grouped as in theCSM: audiocenters, USB, and vhf. How-

ever, in the LM, the audio centers and thepremodulation processor were in one boxreferred toas the signal processor assem-bly. The major functional differences werein the USB ransmission techniques. TheLM down-link voice w a s transmitted inthree basic modes: the normal voice(fig. 6), the baseband voice (fig. 7), and

Microphone1 Microphone1Audio

TLeft

[ ce n t e rJutomatiggain

co n t ro lm s i t i o n

gainco n t ro lo n t m l

Centerosit ionighfosit ionL

tTo S-band phase-modulat ion ransmit terFigure 4. - Command module backup-voice signal flow.

Ext raveh icu la rco m m u n ica t o r - Imicrophone

Low-passf l l te r ClipperTelemetry I Telemetrv

oscillatorsommunicator-subcar r le r Ext raveh icu la r121 microphone

F V C 5 Automatlc gain controlI I . ..

outputreceiver

CM

C M m c eI II I Premodulationrocessor 1 1relay module

subcarr ie rosc i l la to r

1.024-MHz1 PCM S-band phase-t r a n sm i t t e rmodulat lon

Figure 5. - Command module re lay oftheackupoice (fig. 7). The normal EVCS signal flow.7


10/18

LMmicrophone

vhf receiver B

te lem et ry Automatic1.024-MHzPCMsubcarrier

oscillator14.5 kHz

1.024-MHz PCMsubcarrier, .Q- 1 4 y Acombination 8Baseband PreernphasisY "BasebandIClipper

Figure 7. - Lunar module baseband- andbackup-voice signal flow.S-band phase-Figure 6. - Lunar module normal-voice voice was frequency multiplexed with a

composite signal frequency modulated a1.25-megahertz ubcarrier.Thesignal flow. 14.5-k ilohertziomedicalubcarrier;he

1.25-megahertz subcarrier either phase modulated the c ar ri er after being multiplexedwith a 1.024-megahertz PCM subcarrier and a PRN range code or frequency modulatedthe car ri er a fte r being multiplexed with the PCM subcarrie r and the television. Thebaseband voice was frequency multiplexed with the biomedical and PCM sub carr ier , andthe composite signal phase modulated the carrier. The backup voice was similar tobaseband voice, except that the biomedical su bca rr ie r w a s not transmitted and addi-tional voice processing was used. The backup voice w a s preemphasized 6 dB/octave,clipped 24 decibels, and transmitted in a "hot mike" mode. No processing w a s usedfor the normal and baseband voice other than automatic volume control(AVC), and thevoice was transmitted in either a VOX or PTT mode. Also, in the normal and base-band modes, the LM could relay CSM voice o r EVCS voice/data to the MSFN.

The EVCS provided communications between EVA- 1 and EVA-2 and betweenthem and the LM or CSM. The EVCS w a s a replacement for the original space-suitcommunicator, which could support only one extravehicular astronaut. The EVCSconsisted of two extravehicular communicators (EVC-1 and EVC-2) and could operatein thr ee modes: the dual mode, with data from each extravehicular astronaut to berelayed by the LM or CSM to the MSFN and with voice between EVA- 1 and EVA-2 andbetween the extravehicular a stronau ts and the MSFN by way of the LM or CSM; thepr ima ry mode, with voice and data from one extravehicular astronaut; and the second-ary mode, with voice only between one extravehicular astronaut and the LM or CSM.In the dual mode, EVC-1 relayed EVC-2 voice and data to the LM or CSM, thus per-mitting the us e of only one vhf receiver on the spacecraft (fig. 5), which minimizedLM/CSM changes when the EVCS was implemented. The EVCS transmitted voice wasclipped approximately 12 decibels before modulation.

8


11/18

-The MSFN equipment applicable to voice communications usually consistedf adual USB system for simultaneous supportf two spacecra ft and vhf/AM tr an sm it te rsand receivers forearth orbital mission phases. No up-link voice processing was usedother than compression amplifi ers, which were used to preven t overmodulation. Trans -mitter or channel keying was under positive control of the capsule communicator at theMCC, Houston, Texas, by the us e of a n in-band phase-shift-keying tone-burst tech-

nique. On the down links, a 2.5-kilohertz low-pass filter (LPF) was installed on allUSB channels afterit was determined that most of the Apollo down-link voice was lessthan 2 .5 kilohertz and that he elimination of interf erence (no ise andEVCS data subcar-riers) above this frequency improved the quality without detectable word intelligibilitydegradation. Squelch was used on the vhf/AM receivers, but not on the USB, becauseof the requirement to use ll received USB voice regardl ess of the degree of noise andbecause of the ability of the ground crew to disconnect the channeloutput from the linesupon complete loss of signal. However, under carefully controlled conditions, a voice-operated gain-adjust ing amplifier (VOGAA) was used , with less than completely satis-factory results, to limit noise between spacecraft transmissions.

Channel Design ConsiderationsThe significant design considerations were in the areas of voice processing (VOX,preemphasis,andclipping),filtering,andsquelch.Generally, hepremodulationvoice processing in the spac ecra ft consis ted of VOX or PT T cir cuit s to e liminateback-ground noise and to key certain transmitters; AVC amplifiers to ensure sufficient mod-ulating levels and to prevent circuit overloads ha t could cause uncontrolled distortion;preemphasis networks and peak clippers ; and, in some instances, low-pass filters toprevent interfer ence with data channels. After this processing, the voice w a s appliedto the modulato rs and transmit ters . The modulation techniques used were FM/phasemodulation (PM); FM/FM; and baseband FM, PM, and AM. For th e CSM and LMvhftransmitt ers, the resulting clipped voicekeyed the carrier on and off so that the output

resembled a squar e wave. The amount of vo ice processing used, especially peak clip-ping, depended on the individual links and the need to maximize their performance. Oncertain links, no clipping was used; others used 1 2 to 24 decibels.The purpose of the peak clipping w a s to reduce the peak-to-rms ratio of the voicesignal and thereby increase the r m s modulation level. The objective w a s to meet the14-decibel rm s/ rm s an d 4-decibel SNR requir emen ts for 90- and 70-percent word intel-ligibility at the maximum requir ed ranges. Clipping does increase the intelligibilityunder marginal rf signal conditions. The improvement for the AM links is significant,but fo r FM links the improvement generally is much less because of the FM threshold-ing effects. Te st s conducted in January 1966 on the Apollo USB up link (FM/PM) withpreemphasis and 0-, 12-, and 18-decibel clipping indicated an improvement of 1 to

3 decibels at subcarrier threshold for12-decibel clipping and no additional improve-ment with 18-decibel clipping. If only the change in peak/rms ratios is considered, theimprovement should have been approximately 7 and 9 decibels, respectively . In AMlinks, the improvement also does not approach the expected value, presumably becauseof the added distortion products and ncrease in transmitted interword noise (causedyreducing the peak voice amplitude toa value closer to he peak background noise ampli-tudes). The original clipping lev el s fo r the CSM and L M vhf transmitters were greaterthan 40 decibels. Wi t h this heavy clipping and th e on/off carrier modulation technique,cabin noise as much as 40 decibels less than the normal audio levels could modulate9


12/18

the carrier 100 percent, resulting in a 0-decibel signal-to-interword noise ratio. Theprimary solution to this problem was to reduce the clipping level to approximately24 decibels. However, before this solution was found, considerable effor t was expendedon improving the noise-cancellation characteristicsof the microphones and on exploringmeans of suppressing the cabin noise on the audio-center input lines. The noise-cancella tion charact eristics were improved y the incorporation of improved directionalpickup, but attempts at developing a noise suppres sor were abandoned. The suppre ssorthat gave the most promise wasa modified center clipper that consistedof a pa ir ofback-to-back diodes shunted by a resi stor . The perfor mance of this device w a s satis-factory under normal conditions; however, the performance was highly sensitive to theinput audio level and presented reliability problem.

Another problem related to the on/off c a r r i e r modulation technique was high dis-tortion (unintelligible speech) when the CSM was first te sted with the MSFN vhf receiv-ers before the Apollo 7 mission. This distortion was traced to a misalinement of thereceiver and wa s correc ted by improving the MSFN alinement procedures. Because ofthe squar e wave modulation, the requ ired inte rmedia te freque ncy bandwidth w a s almostidentical to theactual bandwidths when the Doppler effect and the effects of tra nsm itt erand receiver stability were considered. Therefore, precise receiver alinement wasnecessary.

The following f ac to rs should be considered in future programs f clipping isconsidered.1. Radio frequency link performance and the resulting need for clipping2. Increase in r m s modulation resu ltin g fr om clipping3 . Background cabin)noise4. Distort ion caused by clipping5. Type of modulation6. Receive r bandwidthsThe AVC amplifiers that were used to compensate for speaking level variationsand thus to maintain constant clipping/modulation lev els al so had some drawbacks.These amplifiers alsocould emphasize background noise and, when cascaded (as w asoften done in the Apollo systems for the relay and conference configura tions), couldproduce a ver y high gain system that could raise normally negligible crosstalk toobjectionable levels. (Thiswas one of the factors that allowed the echo to be heard

during the Apollo 11 lunar-surface operations. ) For these reasons, the use of com-pression in plac e of AVC amplifi ers in futu re p rograms s being conside red at the NASAManned Spacecraft Center (MSC). Also, in the early Apollo design, some problemswere experienced because of the combined VOX and AVC attack t imes; with the VOXpreceding the AVC, it took the combined at tac k tim es fo r the channel to stabilize. Thisproblem was resolved by putting the AVC and VOX detector circuits in parallel. OtherVOX relat ed prob lems suc has switching transients (banging) and first-syllable choppingwere r esolved by similar design refinem ents.

10


13/18

Low-pass filters were used in some premodulation voice-processing circuits toprevent interference in data channels (for example, EVCS 3.9-, 5.4-, 7.35-, and10.5-kilohertz subca rri ers ) tha t are transmitted along with the voice. The filter s thatwere used had cut-off frequencie s as low as 2.3 kilohertz. The us e of these low-passfilters did not degrade intelligibility significantly, because the speech from theApollomicrophones usually was limited (because of the spectrum of the male voice) to essthan 2.5 kilohertz and most of the power was below 2.0 kilohertz (fig. 8 ) . Althoughthis speech normally would not in te rfere with the data channels, the distortion producedby the clipping could produce interference.On the MSFN, low-pass filters were used on the USB down-link voice channels toprevent the data channels from interferingwith the voice and to minimize ut-of-bandnoise. The original MSFN signal data-demodulator system (SDDS) oice filtershad an upper cut-off at approximately3.5 kilohertz (noise bandwidth of 3.65 kil-ohertz). During tests in November 1967,the MSC Audio Techniques and Evaluation

Laboratory (ATEL) of the Teleme try andCommunications Systems Division (TCSD)determined that theSDDS filters permittedexcessive data channel and system noiseinterference to the voice. Therefore, asubsequent effort was initiated in the ATELto develop an optimum SDDS voice-channelLPF. The result s of this effort producedthe recommendation that a 0- to2.5-kilohertz LPF with a rolloff of36 dB/octave for frequenc ies above2.5 kilohertz be installed at the MSFNPANORAMIC SPECTRUM ANALYZER

LINEAR SWEEP WIDTH FACTORMODEL LP-laZ

LOG SWEEP FREOUENCY l CP S l

(b) Subject saying "board. ( (

1.00.8

0.6 z92

0.4

0.2

'0

LINEAR SWEEP WIDTH FACTORMODEL LP-laZ+O. 5

LOGSWEEPFREOUENCY ICPS)

(a) Recorded 1-kHz sine-wave.MODEL LP-laZLINEAR SWEEP WIDTH FACTOR

-0.5 '0.5

1.0

0.8

0.6 zz-

0.42

0.2A 0K

"LOGSWEEPFREOUENCY ICPSI

(c ) Subject saying "necessary. t tFigure 8. - A sp ec trum display of Apollo microphone output.

11


14/18

sites. Extensive tests were indicative that the 2.5-kilohertz LPF improved the speechquality by eliminating the annoying noise and data channel tones above the speech spec-trum without detectable effects in speaker recognition o r intelligibility. Subsequently,this filter was installed at all MSFN sites. However, although the 2.5-kilohertz LPFwas optimum for the Apollo down links, it may not be optimum for future programs.The sex of the speaker, microphone quality, premoctulation processing, link noiseCharacteristics, and adjacent channel interferenceall must be considered when select-ing a postdetection filter.

In the original Apollo designs, voice Channel squelch w as used only on the vhfreceivers . No squelch was implemented on the USB in either the spacecraft or theMSFN. However, i n syst em test s it was demonstrated that the loss of the voice sub-carr ie r or carr ie ron one channel could disrupt voice communications on other channelsand be annoying because of high noise levels produced in the earphones. This problemwas acute on the up-link relay modes, and squelch circuits were installed in both theCSM and the LM bef ore the firs t luna r landing. Subcarrier detection was used insteadof ca rr ie r detection because of the diffe rent subcarri ers and demodulators used fornormal and backup voice and because the voice subcarriers were not present forallup-link transmission combinations. Because of these factor s, a carrier-operatedsquelch w as unsa tisfactory. Because of the up-link signal levels, the sensitivity (oroperating point) of the up-link squelch circuits was not critical; usually, the up-linksubcarrier SNR was sufficiently high for good discriminator operation, or the signallevel was so low as to be almost nonexistent. Thus, the operating point was usually set1 or 2 decibels below discriminator threshold even though word intelligibility was usu-ally greater than 80 percent at this level . However, squelch "disable" was provided toextend communications capability to lower rf leve ls and to give the crewmen the ptionof operating with or without squelch. At times, certa in spacecra ft crewmen pre fer redto opera te without squelch because the level of channel noise provided an indication ofthe channel availability.

Squelch on the USB down links constituted a prob lem that could not be solvedentirely by either a carri er or subc arrie r detector . The down-link signal levels usuallywere not as strong as the up-link levels, and a requirement existed to use all down-linkvoice regard less of the degree of noise. The ground-station operators could handle acomplete loss of su bc ar ri er o r ca rr ie rby disconnecting the objectionable channel outputfrom the lines. The real problem occurred during mission phases when both the CSMand LM USB links were active and the down-link audio signals were combined at anMSFN sta tion fo r tr ansmissio n through one landline to the MCC, and, in the MSFN re -lay mode, when the audio from each spacecraftwas transmitted on the up links to theother spacecraft. During these mission phases, a noisy but usable down link couldseriously degrade the otherwise good voice from the other spacecraft or, in the MSFNrelay mode, could seriously degrade the MCC voice transmissions received in the otherspacecraf t. This problem could not be solved with a carr ie r or subcarrier squelchbecause the squelch would mute the output (noise plus usable voice) of the objectionablechannel. In an attempt to solve the down-link squelch problem, voice-operated gain-adjusting amplifiers were installed at the MSFN sites. These dev ices were intended topass the "voice plus noise" and mute the "noise only, v t theoretically, a good solution.However, the VOGAA had some bad characteristic s: it dropped syllables and wordsand produced objectionable banging sounds. Because of these charac teri st ics, test swere performed in the ATEL to determine the precise operating characteristics f theVOGAA. A s a result of these tests, detailed procedures were developed and12


15/18

implemented by the MSC Flight Support and Fl ight Control Divisions for the useof theseamplifiers. In the procedures, specific mission phases were defined fo r th e us e of theamplif iers, requir ements to main tain correc t input leve ls, requirem ents for the MSFNopera tors to monitor the input/output, and procedures for bypassing the amplifiers. Afollow-on task also was under taken to optimize the OGAA speech detection circui tsfor Apollo speech. An analy sis of the Apollo speech using power-spectral-densitymeasurement equipment was indicative hat the greatest concen tration of power fo r eac hspoken word was in the frequency band of 400 to 800 her tz. Because this band of fre-quencies would have the highest speech-to-noise power ratio, it was decided to applyonly th is pa rt of the speech spectrum to the VOGAA speech-detection cir cuits. A te lem-etry filter with a 540-hertz center frequency and an 80-hertz bandpass waschosen. Theincorporation of th is filter and the extension of the VOGAA "hold" time af te r speechwas detected resulted in fewer speech dropouts and less impulse noise. However, anyamaunt of voice dropout is unsatisfactory, and the development of an ideal syllabic de-tector, perhaps using digital techniques, is desirable.The voice process ing (AVC and clipping) and modulation indexes were optimizedto obtain the required circuit margin s fo r eac hdown-link tr ansm ission combination.

One of the adverse byproducts of this link optimization w a s changes in received r m saudio levels at the MSFN when different down-link voice modes were selected. Thisphenomenon was first noticed in an Electronic Systems Test Laboratory (ESTL) systemdemonstration conducted in July 1967 and was subsequently evaluated and verified inMSC document EB68-3709, "Evaluation of Apollo USB Voice Downlink Audio LevelVariation, '' dated December 1968. The variat ion w a s large (approximately 35 decibels);however, most of the variation (approximately 19 decibels) could be fixed by a one-timeMSFN adjustment. The only satisfac tory solution to the remaining variation was real-time level control by the g r m d station operators. In future programs, the effect onaudio levels of audio process ing and changes in modulation causedby mode changesshould be a design consideration.Audio Test Techniques

The major Apollo total systems compatibility testing w a s performed in the ESTLusing spacecraft hardware (simulated hardware, developmental models, and flightmodels) and a configuration controlled MSFN USB stat ion. For the voice channels,bas ical ly this testing consisted of estab lish ing the up-link and down-link signal combina-tions, simulating the rf ranges by means of adjustment of rf path loss, measuringvoice-channel SNR (predetection and postdetection), making word-intelligibility tapesfor subsequent scoring, determining VOX/squelch opera ting characte ristics, determ in-ing system crosstalk, and conducting functional tests to determine operational problems.The ATEL was established at MSC in 1966. This labora tory provides a capabilityfor detailed evaluation of audio characteristics and developmentof voice communicationsand test techniques. For the Apollo Progr am, the majo r ATEL functions were asfollows.1. Perform or coordinate intelligibility tests2. Maintain an audio- tape library of master source tapes and test tapes

13


16/18

3. Perform detailed analyses of selected ESTL and flight audio tapes to resolveanomalies4. Develop audio-test techniques and standards to be used in theATEL and ESTL5. Develop recommended system modifications to improve total system

performanceInitially, word-intelligibility source tapes were prepared in the ATEL soundroomusing standard auditory test word lists, and the resulting data tapes (made by playingthe source tapes through the ESTL systems) were played back to subjects in the ound-room for intelligibility scoring . Because of the limited availability of test subjects,the results were not reliable. To improve the test results, arrangements were madeto perform a word intelligibility test in the MCC using flight controllers as subjects.The re su lt s of thi s test were good; a sufficient number of trained and motivated sub-jects were available. However, the results of a subsequent test in the MCC were notas good. Because of inconsi stent results, effort s were intensified to obtain an outsideorganization to score word-intelligibility tapes for the NASA. The resu lt was that an

interagency agreement was made with the U. S. Army to score the tapesat the FortHuachuca Test Center. The source tapes for these tests were made using trainedspea kers and Apollo microphones in Apollo suits. These source tapes were playedthrough the ESTL system according to a procedure developed for this purpose. Theresulting data tape s were given a cursory examination in the ATEL and were forwardedto For t Huachuca for scoring. The maste r sou rce t ape s and the returned data tapeswere filed in the ATEL tape library. On specific problems, ATEL personnel performeda more detailed evaluation of the tapes. This more detailed evaluation included S N Rmeasurements using test tones that were recorded on the tapes, speech-to-noise meas-urements using a technique developed by the TCSD, spe ctr al ana lyses us ing cathode-ray tube (CRT) spectral analyzers or a power spectral-density X-Y plotter, andtime-domain analyses using oscilloscopes and a high-speed beam oscillograph.The speech-to-noise measurement technique was developed because the wordintelligibility results did not agree closely with the SNR (using a sinusoid as the signal)measurements. It was believed that signal-to-noise rat ios based on the actual speechwould be more representative , and, therefore, more relatable to intelligibility. Un-fortunately, the use of availa ble me ters to measure rms sp eech l evels w asoo timeconsuming and inaccurate because the results depended on visual calibrations. There -fore, an effort was initiated to develop a speech-to-noise measurement technique thatwould produce reliable and repeatable results.A digital computer was programed to detect speech in the presence of noise.Measurements of speech-plus-noise power and between-word-noise power were madeand used to calculate the speech-to-noise ratio for 1 second intervals. The speech-to-noise ratios for the 1 second intervals and the long te rm speech-to-noise ratio wereprinted by the computer. Analog test equipment was used to develop and verify thespeech detection and speech measurement techniques applied in the digital computerprogram. A strip-c hart recorder and a printer were added to the analog system toallow simulation of the digital system for compari son purposes. Both systems providedreliable data. The analog system was retained in the ATEL. Subsequent use of thi stechnique proved the invalidity of specifying that all links (or all link configurations) will

14


17/18

provide a specific word intelligibility percent score for a given SNR. However, a re-liable, repeatable method of determining actual r m s speech-to-noise ratios does nowexist (ref. 1).A method also was developed to estab lish and measure voicemodulation indexesaccurately. Previously, such setups and measurements depended on vis ual est ima tes

(from an oscil loscope) of the voice levels. Because speech is a dynamic signal, humanjudgment was required in making the determination. Also, the actual voice modulationlevels of a system can be greatly different from the design levelsf sine waves areused in t he modulation-index (MI) setup. To overcome th is problem, a static represen-tat ion of the voice signal was produced by performing a spec tral analysis of the voiceoutput of the Apollo microphones and by simulating the spectrumwith a noise generatorand variablefilters. The simulated spectrum can be r ecorded on tape and used as thesys tem input in making MI adjustmen ts or measurements. The need for this techniquewas demonstr ated by the problems experienced with the M backup down-link voice.On thi s channel, the modulation design and setup were based on sine waves; however,when voice was used, the actual peak modulat ion was much greater. Thi s excessivevoice modulation caused serious degradation inhe PCM telemetry Channel.One of the major tasks in the postflight evaluation of Apollo voice performancewas the determination of the actual spacecraf t/MSFN configuration: backup voice asopposed to normal voice, CSM or LM or both remoted to MCC, up link to CSM or LMor both, LM or CSM relay of EVCS, and s o forth. The MSFN recordings were themajor source of data available for making this determination. The 14-track recordersprovided a recor d of all site voice channels: CSM USB air-to-ground up and down link,LM USB air-to-ground up and down link, vhf air -to-ground up and down link, CSM USBnormal-voice down link, LM USB normal-voice down link, CSM USB backup-voice downlink, LM USB backup-voice down link, CSMUSB veri fica tion receiver up link, LM USBverification receiver up link, and "Ne t 1" (up and down links remoted to MCC). Be-cause of the amount of data and the complexity of possible configura tions, the approachused in theATEL to determine the actual configurationw as to use a multichannel beamoscillograph to make a time-corr elated record of the re cord er tra cks, enabling moreaccurate determina tion of the configuration with mor e economica l useof manpower.The use of power-spectral-density plots and CRT spec tral analyzer sw a s usefulin identifying interfering tones (data subcarrier intermodulation products, power supplyelectromagnetic interference, and s o forth) and in determining optimum low-passfiltering.

CONCLUDING REMARKSExperience gained in the development of the Apollo voice-communications systemhas produced some conclusions, which are listed here as an aid to thos e involved indefining or developing voice systems for future programs.The performance requirement shouldbe comprehensive and shouldbe stated insuch a manner that it can be verified economically by means of standard techniques.Although percent word intelligibility defines a major aspec t of channel performance, itis not concerned directly with quality, crosstalk, and certain types of interfering tones

15


18/18

and noise. Also, the techniques used in measuring percent word intelligibility arenumerous and usually expensive; and, if percent word intelligibility is specified, theprec ise method of measuring it al so should be specified. Additional work in definingan optimum performance requirement specification, including applicable design crite-ri a and related test specifications, would be beneficial.The tota l sy stem, talker to l istener, should be considered in the design of anycomponent of a voice communications system. The effects of background noise andcascading of automatic-volume-control amplifiers , clippers, filters, and voice-operated relays should be well understood. It is not valid to design two independentlinks to meet the link-performance requirement and then cascade them and expect thatthe system requireme nt a lso will e met.The us e of voice processing should be examined carefully as to the need for t andits effect on noise, distortion , and crosstalk. If needed because of radio-frequencymargin limitations, approximately 12-decibel peak clipping is a suitable level; mostof the system improvement occurs with the fi rs t 1 2 decibels, and 12 decibels generallya r e not enough to produce severe distortion or to emphasize cabin noise. Automatic-

volume-control amplifiers should be avoided whenever possible because (dependingndynamic range) the amplifiers too can emphasize noise and raise negligible crosstalkto objectionable levels.

Manned Spacecraft CenterNational Aeronautics and Space AdministrationHouston, Texas,December 3, 1971914-50-17-08-72

REFERENCE1. Schmidt, 0. L. : Measurement of Speech-to-Noise Power Ratios by Using a DigitalComputer. Paper presente d at Fir st Annual Houston Conference on Circuits,Systems and Computers (Houston, Tex. ), May 1969.

16 NASA-Langley, 1912- 1 s-313

Apollo Experience Report Voice Communications Techniques and Performance

Documents

Transcript of Apollo Experience Report Voice Communications Techniques and Performance