V

f*

H

■'-: *s s « •

'Ki 1 J L ■ ■

if" fc"-^

£ a£. ?-" ■ * : * >

:;. > ■ r IT-. ; * ■

'<*. I 1 ' f ' l w *- 1 ?* i" ■ q h j f

i f '#■>■ J»" «,,

.;-: ."> ! P ™"f ^ \ P'

F% ^^^^k i ^ t ^^k ^ ^ ^ ^ ^ L

^^k ^ ^ ^ ^ ^ k ^ T ^ ^ ^ ^ ^ ^ B ^ ^ __ ^ ^ ^ ^ ^

V^^^J^k ^^^^^B ^ ^ ^ ^ ^ ^ l^m

^^^^^^^^^^ ^^^^^^^^ ^^^^^^^^H ^^K ^^^^^^^^^B ^^^^^^^^^^^^^^^^^^H ^^m^^

^ ^ ^ ^ ^^^ ^^^MT *

^^M^ ^ ^ H ^H_^^^^^^ ^^H

^^^ _^^^H .^^F _« ^^H^^^^^T ^^MT

^K ^m i ^ L_ ^ F J ^ ^ ^ ^ H

^C^H HR?SSB1^--^ ^ ^ ^ ^ ^ M ] u j | ^ ^ « r f Q r j i t

'S^^^^sS^**?? 3^? ^ j — B s t r V 3' J f c . V p j A i t f j t - ! ■ ■

^ ^ ^ 1 ^H. V I \ 1 1 k m * l ^ B B ■

L ^^b ^ f k ^^^M ^^^L ^^^H

^[ ■

^^^^H ^ |

.^■H ^ ^ ^ r ^ ^ ^ ^ H ^ ^ F ^^v^^^i ^r I^H J f ^ B ^B ^^B ^K

^^^^^^f ^^H

W^M W^ i^B I i

r ^B ^i 1 i ^V B ^^^^^^J

ifJiiS^^KR "fi*T^Si

j ~ f J P J T V

' ^^H

■ : .

■ ^^^H 1 MM I m ^f A [ ^L^r J Wr £

^M ^^^d ^r ^^r^ ^^^ . ^^^g^^

K^^fl

^^^_ ^^V^^^^H ^^^^H

^P^L. ^V ^ ^ ^ b

p ^ ^ L ^ ^ ^ ^ ^ ^ -^ ^ L ^ ^ ^ ^ ^ L

^ L ^ ^ ^ ^ ^ ^ k ^ C ^^^H^k ^^^L

I ^T^fc V ^ L ^^^L i ^ h ^ ^ L

1 K ^ K . ft ^L ^^^^^^^H

^L^H . : S K S

■

^^H

&ft ■

.

r ^H ^^^k ^r ^^fl ^ ^ ^ ^ ^ H ^^^^^^ ^^^^H

M^^V ^ ^ ^ ^ ^

^^^^^H ^^^^B

^^^^^^^^^H ^^^^^^H

^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ■ ^ 1 ■ ^^^H ^ ^_ ^^^B ^ ^ ^ ^ ^ ^ ^ ^ H

^ ^^B ^ ^ t

L ^H ^^^^^^^H

^^Br*i3 ^^Hsdt^ r^jSP^SfiT^G i&t^&SSi^^^B H J ^ f i " * i * i ^ J irf^L * "''■"'■ ^^ff*S^*^y-HUH

M <

15-039

Contents

General in t roduct ion 3

Measuring methods and experiences w i t h the object ive measurement of reference equivalents in Norway

by Trond Ulset 3

Review of current te lephonometr ic measurements in Finland by I. Jantt i & T. Tuisku 7

Acoust ic feedback in telephone sets by Einar Laukli 10

Factors determining the reproducibi l i ty and accuracy requirements of te lephonometry measurements: Review of some practical experiences

by I. S a lama 13

Frequency response measurements on the telephone sets type 7 3 D , F68 and 47E connected via a telephone line

by S. A. Jager & P. Schnack 17

What should be required of a Telephone Measuring Equipment? Product ion control experiences

by 0 . Larsson 21

General Introduction On the 12th and 13th November

1974 a special Telephonometry Symposium was held in Copen­hagen. Hosts for the symposium were KTAS (Copenhagen Telephone Company) and JTAS (Jutland Tele­phone Company), and initiative for the meeting was mainly due to the efforts of Mr. P. V. Arlev and S. A. Jager of JTAS.

and Sweden contributed to the suc­cess of the meeting.

The symposium consisted of lec­tures and discussions related to de­velopments in the field of telephon­ometry, and participants from Eng-and, Denmark, Finland, Norway

Of the papers presented, two sum­maries and six full texts were later printed in the B & K Technical Re­view No. 1 , 1975 , copies of which are available on demand from B & K . This Application Note prints a further five papers plus an extra paper which although not presented at the symposium received some dis­cussion there.

Papers printed in Technical Re­view No. 1, 1 975:

Problems in Telephone Measure­ments, by Norman Gleiss

Proposals for the Measurement of Loudness Ratings of Opera­tors' Headsets, by R. B. Archbold Comparison of Results obtained by Subjective Measuring Meth­ods, by lb Gilberg

Repeatabilities in Electro-Acous­tic Measurements on Telephone Capsules, by R. E. Walford

Stable Subset Measurements wi th the 73D, by K. Damsgaard Vibration Testing of Telephone Equipment, by R. Fluhr

Measuring methods and experiences with the objective measurement of reference equivalents in Norway by Trond Uiset, Sectional Engineer.

Norwegian Telecommunications Ad­ministration, Central Laboratory Acoustics Group.

Summary Information on the use of objec­

tive measuring of reference equiva­lents in Norway. Description of meas­uring methods and adjustment procedure for the measurement of telephone sets equipped wi th linear microphone and amplifier. Descrip­tion of future adjustments and im­provements. Description of addi­tional measurements made in con­nection wi th the telephone trans­mission measuring set. Experiences wi th the use of objective reference equivalents.

1 . Introduction The purpose of this paper is to

give some information on the meas­uring methods for objective tele­phone measurements used by the Norwegian Telecommunications Ad­ministrat ion.

Our first objective electro-acous­tic telephone transmission measur­ing set was purchased in 1959 . It

was a laboratory version of the Sie­mens Objektiver Bezugdampfungs-me&platz (OBDM). The system was used for equivalent measuring of te­lephone sets. Later on we became interested in measurements of ob­jective reference equivalents for spe­cial equipment, automatic answer­ing machines in particular, to en­sure that this equipment had a transmission efficiency comparable w i th that of telephones.

In 1 9 7 1 , we decided to procure a Bruel & Kjaer Electroacoustic Tele­phone Transmission Measuring Set Type 3 3 5 2 . Our work itself had not been considerably changed, but the amount had increased. Relatively soon we were using only the new measuring set.

As from 1st January, 1974 , a new unit has been established at our Central Laboratory. One of the main tasks of the new unit is to study measuring methods for subjec­tive and objective telephonometry and to participate in international co­operation. In the light of the con­stantly increasing amount of objec­tive measurements and the prepara­

tion of improved measuring meth­ods, the Central Laboratory has pur­chased another measuring set Type 3 3 5 2 .

2. Measuring Requirements In 1967 , the Norwegian Telecom­

munications Administration intro­duced a new telephone set. This set was equipped wi th an electrody-namic microphone and a two-step microphone amplifier. All special sets provided by the Norwegian Te­lecommunications Administrat ion to­day are also equipped wi th linear microphone and amplifier. This means that by far the majority of the objective measurements of ref­erence equivalents are conducted wi th sets without carbon micro­phone. In the preparation of stand­ard measuring methods it is conse­quently not necessary to consider factors such as non-linearity and packing. In the examination and con­trol of carbon microphone sets we fol low the prescriptions correspond­ing to OREM C, described in the In­struction Manual for the Type 3 3 5 2 .

n Norway the objective reference

3

equivalents are presented as a func­tion of inserted line length. Because of the linearity of the microphone 2 or 3 measurements are sufficient to describe the variation of the trans­mission/recept ion equivalent as a funct ion of inserted line length. Sidetone equivalents must be meas­ured for several lengths of inserted line in order to describe the varia­t ion.

The feeding system of most new types of telephone exchanges in Norway is 4 8 V , 2 x 2 5 0 O , and this has usually been used during measurements. The artificial line equivalent generally used has been 0 , 4 m m , 2 8 0 0 / k m , 4 5 n F / k m .

Objective telephone measure­ments are conducted for the fol low­ing purposes:

— Type control of special sets — Type control of headsets — Type control of privately owned

equipment — Production control of telephone

sets and handsets — Production control of micro­

phones and earphone capsules — In connection wi th minor

changes on old models

4

3. Measur ing Methods 3.1. OREM A (NJ

Usually we measure the proper­ties of a complete telephone set; not just the microphone and earphone capsules. That was the reason for originally selecting the OREM A procedure. This procedure pre­scribes an artificial mouth in the REF-position. This led to a mechani­cal contact between the labial ring

and the mouthpiece. We therefore had to use the AEN position. We also chose to use the new IEC audi-ometric coupler instead of the NBS 9A coupler. The sound pressure was adjusted wi th a SFERT-baffle to 93 ,6 dB re 2 x 1 o ~ 5 N / m 2

(—0,4dB re 1 Pa). The receiving ref­erence level was also set at 93 ,6dB re 2 x 1 0 " 5 N / m 2 . The meter damping was OB DM. The rest of the procedure is identical wi th OREM A described in the In­struction Manual for Type 3 3 5 2 . We also chose to read the left de­flection on the reference equivalent meter because we thought that this was simpler and safer.

3.2. Orem C (Nj In the beginning of 1973 we

changed the measuring procedure.

41 mm 30 mm h H \*~ —*\

750472

Fig. 1. Voice pressure measuring j ig used in OREM C (N)

The main reason for the change was that we wished to use an un­weighted sound pressure from the artificial mouth. This meant that the SFERT-baffle for adjusting the sound pressure could no longer be used for calibration, so we designed a support for a 1 / 2 " microphone (Fig.1).

We chose a reference sound pres­sure of 94 dB re 2 x 1 0 - 5 N / m 2

(1 Pa). This sound pressure from the artificial mouth was measured 25 mm in front of the labial r ing.

The fol lowing parameters de­scribe the measuring procedure:

— Artif icial ear: IEC Audiometr ic — Artif icial mouth: AEN position — Sweep: 0,2 — 4,0 kHz — Meter scale: y = kx ° ' 6 dB — Meter damping: OBDM — Weighting network: Off

Reading of the left deflection of the objective reference equivalent meter.

Modern l ightweight headsets are loaded wi th 2 cm3 artificial ear dur-

•

n ""

K .■

"£ £ k -J , ^ '1 3 * r

^

^ \

s

s

s

^

ing measuring of receiving and side-tone equivalents.

3.3. Modifications of the existing measuring method

The purpose of further prepara­tion of objective measuring method is to

Describe the qualities of the tele­phone set in the most appropri­ate way w i th as few basic errors as possible Achieve the best possible corre­spondence between the subjec­tive and the objective measuring results.

rable peak l imiting may occur in the amplifier. Apart from this, such a high pressure wi l l not be necessary. Experience has shown that a sound pressure of 84 dB re 2 x 1 0 ~ 5 N / m 2 , adjusted according to procedure OREM C (N), results in an objective transmission equiva-ent deviating relatively little from

subjective transmission equivalents rel. NOSFER.

Because it is not possible to carry out subjective tests in the Norwe­gian Telecommunications Admin i ­stration today, we have concen­trated our effort on the preparation of the most suitable objective meas­uring procedures.

For sets w i th linear microphone and amplifier a sound pressure of 9 4 d B re 2 x 1 o ~ 5 N / m 2 (1 Pa) is too high for measurements of trans­mission and sidetone equivalents. When the conditions are unfavou-

We also re-considered the choice of sweep width and decided that a sweep of 3 0 0 — 3 3 0 0 Hz corre­sponds more to the bandwidth of a telephony channel.

The meter function and the meter attenuation for the objective refer­ence equivalent meter also have to be considered. A meter function of y = kx ' dB corresponds more to the presented results. The question of using the Slow meter needle damping or whether it would be more correct to use the so-called OBDM damping, i. e. a damping corresponding to that of the meter needle in the Siemens OBDM, is also something to consider.

The practical execution of objec­tive reference equivalent measure­ments is important and is given much consideration. In the prepara­tion of measuring procedures for telephone sets w i th linear micro­phone and amplifier, we may disre­gard phenomena like packing. The measuring procedure itself w i l l ac­cordingly be simpler, quicker and safer than a procedure for measur­ing carbon microphones. It is neces­sary to check that the handset is ap­propriately attached and that acous­tic connection between the ear­phone and the coupler is good. The acoustic coupling is most simply checked by means of a Frequency Response Tracer.

We must take due account of the conditions under which the meas­urements are carried out. The es­sential factors are those influencing the amplifier of the telephone set, the line current and the sound pres­sure. Too high a line current may cause part or complete destruction of elements in the amplifier; unfav­ourable combinations of line current and sound pressure may cause peak l imiting of the amplifier.

For the subscribers' protection most modern telephone sets are equipped wi th a limiter in the ear­phone to prevent aural damage caused by acoustic shock. In meas­urements of reception equivalents we must take this into considera­t ion when we decide the pressure eve I.

Finally, we must consider the sur­roundings. As far as possible reflec­tions must be avoided during meas­urements of transmission and side-tone, and noise must be prevented from disturbing the measurements.

3.4. Additional measurements In addition to the objective refer­

ence equivalent measurements we carry out the fol lowing measure­ments on the measured set:

(a)

(b) (c)

Characteristics of cu r ren t / voltage Harmonic distortion Adjustment attenuation (impedance)

(a) The characteristics of cu r ren t / voltage, i .e. voltage drop over the telephone set as a function of the line current, are found by varying

5

the voltage of the power-supply and reading off the line current and the voltage over the set connectors.

(The switch of the voltmeter is in position Line 2).

(b) We have considered different methods for the measurement of harmonic distortion:

Measuring of harmonic compo­nents by means of a spectrome­ter. Measuring of harmonic distortion by means of a slave filter. Measuring of harmonic distortion and products of intermodulation by means of sweep and highpass fi l ter.

The first two methods give infor­mation on harmonic distortion as a funct ion of the frequency. The re­sult may be wr i t ten out in curves. The last method gives information on a mean value of the sweep area.

We are usually interested in infor­mation on the total harmonic distor­t ion, not the harmonic components, and f ind that the use of a spec­trometer is not appropriate. In the opposite case, however, it is used.

The use of a slave filter requires a room wi th a good noise insula­t ion, usually special rooms like ane-choic chambers. It is possible that a combination of a slave filter and a frequency analyzer Type 2 0 2 0 or 2021 may prove useful.

The use of sweep and high pass filter gives an expression of mean generation of harmonic distortion and intermodulation over the fre­quency range of the sweep. This is a fast measurement, and we have found it gives adequate information. Today we use two measuring tech­niques:

Sweep 2 0 0 4 0 0 0 Hz, 1 sweep per /s , fi lter w i th limit fre­quency 5 0 0 0 Hz, sound pressure 8 4 d B re 2 x 1 f j - 5 N / m 2 , (— 10dB re 1 Pa) 2 5 m m in front of the lip ring of the artificial mouth.

Sweep 300 — 8 0 0 , 2,5 sweep per /s , filter wi th limit frequency 9 0 0 Hz, sound pressure 94 dB re 2 x 1 0 ~ 5 N / m 2 , (1 Pa) 2 5 m m in front of the lip ring.

(c) The return loss is expressed by the ratio between the impedance of the equipment and a reference im­pedance given in the equation.

z z

+ ZR

-ZR

Z = equipment impedance. ZR = reference line impedance

The measuring shown in Fig.2.

schematic is

zg«zR

u

3.5. Use of plug-in units For use in the examination of cap­

sules and as a reference, we have constructed two plug-in units. We have used two ordinary telephone sets, one of each of the most com­mon models in the Norwegian Tele­communications system.

In order to increase the benefit of the use of plug-in units we are con­structing a special reference set. From a transmission technical point of view, it wi l l be a perfect tele­phone set built w i th particularly strict tolerances in components and stability. Such a reference set con­structed as a plug-in unit has great advantages when used in type con­trols and environment tests of cap­sules as wel l as of telephone sets.

750473

Fig.2. Arrangement for return loss measure­ment

The ratio between impressed and measured voltage when Z-j = Z 2 is

U 0 = 2 Z + ZR

U Z — Z R

and the result may be read off the frequency response tracer or the level recorder.

4. Experiences Our experiences from the use of

objective measuring of reference equivalents of telephone equipment in the Norwegian Telecommunica­tions Administrat ion have been very satisfactory. An objective telephone transmission measuring set is indis­pensable in the production control of larger quantit ies. We have also

6

found that objective measurements of reference equivalents give us reli­able information on the trans­mission qualities of the equipment. The transmission qualit ies of a tele­phone set may be described in a simple and wel l defined measure­ment presentation. The measure­

ment precision seems to be satisfac­tory.

The determination of reference equivalents rel. NOSFER based on OREM measurements seems for the present to be somewhat insecure. It is therefore w i th great interest that

we are watching the work carried out by CCITT SG XII in this field be­cause an improved conformity be­tween the OREM and NOSFER measurements wi l l make the bene­fits of the objective measurements of reference equivalents even grea­ter.

Review of current telephonometric measurements in Finland by / . Jantti and T. Tuisku

Equipment used in reference equiv­alent measurements

As a national standard for measur­ing objective reference equivalents a Siemens OBDM is used. It was supplied to Telephone Laboratory of General Direction of Posts and Tele­graphs (PLH) in 1960 .

Measurements w i th OBDM are made as instructed by the manufac­turer. Following points may, how­ever, make an exception to the com­mon practice:

the shape of the lip ring of the ar­tif icial mouth has been slightly modified in order to al low meas­urements of sets w i th modern short handsets. when reading the OBDM meter the average position of the nee­dle is observed. carbon microphones are condi­t ioned in principle according to IEEE standard 2 6 9 - 1 9 7 1 , but technicians show a tendency to have individual features in the method.

In addition to Telephone Labora­tory of PLH objective reference equivalent measurements are made by fol lowing institutions:

Helsinki Telephone Company (HPY), research and on-receipt in­spection. OY L . M . Ericsson AB (LME), pro­duction control. PLH/Division for Supply (HO), on-receipt inspection.

* General Direction of Posts and Telegraphs Telephone Laboratory, Helsinki.

#

(dB)

15 -

10

5

I- / / / Data points inside

these limits

5 10

LME 1/3/0.53 LME 2/3/0.62 LME 1/2/0.75 LME 1/1/0.84 LME 2/1/0.71 LME 2/2/0.74 PLH 1/3/0.95 PLH 1/1/0.92 PLH 1/2/0.82 PLH 2/3/0.95 PLH 2/1/0.96 PLH 2/2/0.67 HPY 1/3/0.93 HPY 2/3/0.94

15 OBDM (dB)

Telephones with carbon microphones, linear relations for sending

LME 1,2 : 2/LTM 357500 HPY 1,2 PLH 1,2

equipments based on OBDM

eg LME 1/2/.95 : LME 1, test 2, correlation coefficient .95

Correlation coefficient deviates significantly from other coefficients calculated to the same equipment 750474

F i g . l .

7

3

(dB)

"I

- 3

Data points inside these limits

OBDM (dB)

PLH 2/3/0.97 PLH 2/2/0.93 PLH 1/3/0.95 HPY 2/3/0.90 PLH 1/2/0.89 HPY 1/3/0.81 LME 1/3/0.86 LME 2/3/0.49 LME 2/2/0.48 LME 1/2/0.51

Telephones with carbon microphones, linear relations for receiving

LME 1,2 : 2/LTM 357500 HPY 1,2 PLH 1,2 equipments based on OBDM

eg LME 1/2/.95 : LME 1, test 2, correlation coefficient .95

Correlation coefficient deviates significantly from other coefficients calculated to the same equipment

750475

Fig.2.

I

LME 1/2/0.96 LME 2/2/0.97 LME 1/1/0.96 LME 2/1/0.96 LME 1/3/0.995 LME 2/3/0.995 HPY 2/1/0.99 HPY 1/1/0.97 PLH 2/1/0.99

PLH 1/3/0.96 PLH 1/1/0.99 PLH 1/2/0.99 PLH 2/3/0.96 HPY 2/2/0.996

HPY 1/2/0.97 PLH 2/2/0.996 HPY 1/3/0.97 HPY 2/3/0.94

Stable telephones, linear relations for sending

LME 1,2 : 2/LTM 357500 HPY 1,2 PLH 1,2 equipments based on OBDM

OBDM (dB)

eg LME 1/2/.95 : LME 1, test 2, correlation coefficient .95

* Correlation coefficient deviates significantly from other coefficients calculated to the same equipment 750476

Data points inside these limits

Fig.3.

8

J L OBDM <dB)

Data points inside these limites

PLH 2/2/0.99 PLH 1/2/0.99 HPY 1/2/0.998 HPY 2/2/0.99 HPY 1/3/0.95 PLH 2/3/0.95 HPY 2/3/0.90 PLH 1/3/0.95 LME 2/2/0.98 LME 1/2/0.98 LME 2/3/0.70 LME 1/3/0.89

Stable telephones, linear relations for receiving

LME 1,2 : 2/LTM 357500 HPY 1,2 PLH 1,2 equipments based on OBDM

eg LME 1/2/.95 : LME 1, test 2, correlation coefficient .95

correlation coefficient deviates significantly from other coefficients calculated to the same equipment

750477

Fig.4.

LME use their own design of transmission test equipment 2 /LTM 3 5 7 5 0 0 . The equipments used by HPY and PLH/HO are manufactured by HPY Research Institute. They are based on the principle developed by Dr. Braun and are calibrated against OBDM by means of stable telephone sets. Though the purpose has been to make these equipment essentially identical w i th OBDM, there are small differences in the frequency sweep, artificial mouths and time constants of the meters.

HPY also has a SRE working standard for subjective tests, but it has not been very much used re­cently.

A n Electroacoustic Telephone Transmission Measuring System B & K 3 3 5 2 has been ordered by the Telephone Laboratory of PLH, and was delivered in November 1974 .

Comparison measurements of ref­erence equivalents

During the last three years, objec­tive reference equivalent values ob­tained w i th OBDM and other equip­ments have been compared. Meas­urements have been made on both stable and carbon microphone sets, which have been circulated be­tween all the institutions mentioned above. The main reason why these measurements were started, was the use of completely different ob­jective R. E. measuring equipment at manufacturers production control and buyers' on-receipt inspections. In addition to that, correlation w i th the national standard was needed.

Some results of these compari­sons are shown in f igures 1 and 2 (17 sets w i th carbon microphones) and 3 and 4 (3 stable sets). These results do not encourage one to think that the difference between values could be predicted, even though always same types of sets are measured.

A review of the network The telephone network of Finland

has been planned in accordance with recommendations of CCITT. The reference equivalent f igures used in planning agree w i th those in P. 2 1 , but the reference equiva­lents have to be measured w i th OBDM.

The publication "Puhelinverkkojen rakennemaaraykset" (Regulations for building telephone networks) es­sentially states:

"As reference equivalent values those values shall be used, which have been obtained by loudness comparison to an inter­nationally accepted standard speech circuit (NOSFER). Until further notice an objective equipment for measuring refer­ence equivalents (OBDM), which is located in Telephone Labora­tory of PLH, shall be used as a national standard".

The regulations do not say any­thing about the difference in values obtained wi th subjective and objec­tive methods, and the practice has been to use only values measured w i th OBDM. This means that in­stead of the target value (mean value of SRE's of several sets SRE = 4 , 5, . . . 7 dB at GO line) the tele­phones have 6 . . . 7 dB higher SRE's.

Measurements of speech voltages were started in October this year in our network. The first results make us believe that the send reference equivalents really are somewhat high. The speech volume at the line terminals of the set has a mean value — 2 7 , 6 vu and a standard de­viation 4,1 vu at local calls.

An immediate question is, what , if anything, we should do to correct the send reference equivalents of the local telephone circuits. At pres­ent we feel, that it would not be wise to hurry, but to wait unti l the work on Question 1 5/XII is ready. After that we perhaps wi l l have a better view of the situation.

dB

0

- 10

- 2 0

- 3 0

- 4 0 x

Handset lifted 20 mm

Handset lifted 40 mm

kHz 750481

Fig.4. Ratio of receiver drive signal and microphone pick-up signal for various heights of hand set above subset

dB 0

- 10

- 2 0

30 kHz

dB + 10

0

- 1 0

Fig.5.

2 5 2 5 2 5 kHz 750482

Upper: ratio between receiver drive signal and microphone pick-up signal as the handset approaches a user Lower: Effect of a capacitance C on a frequency-dependent feedback path

difference varies according to fre­quency when the handset is kept in different positions: in free f ield, in normal positions on the telephone set, on table w i th holes parallel to table and on table w i th holes down.

In all examples shown, the set terminal is open and wi thout any line length, if not particularly men­t ioned.

The last mentioned position wi l l give howling at about 3 kHz, at zero degrees phase difference.

The telephone sets are chosen to i l lustrate the different measuring methods, and they are not typical for the Norwegian sets produced to­day by A / S Elektrisk Bureau.

If the handset is put on the tele­phone set in its normal position and then gradually l ifted, we can see the peaks in the stability responses get lower and the resonances move to other frequencies (Fig.4).

We found that the phase differ­ence between supplied and meas­ured voltage change about three t imes 3 6 0 degrees during the first couple of cm the handset is lifted from the telephone set. The signal are in phase three t imes, and if the loop gain is sufficient, unstable con­dit ions can occur.

Handset position In the upper part of Fig.5 is dem­

onstrated how the handset position influences the degree of stability: in free f ield, vertical in hand and held against the ear.

The lower part indicates ihe inf lu­ence of a frequency dependent feed­back in the microphone amplifier (a capacitor), C = 0 and C1 to C4 ( in­creasing capacity). The handset is here placed on a table w i th holes down.

Free field We have also kept the handset po­

sition constant, in free f ield, and varied other factors.

On the upper left in Fig.6 is shown the result of two different types of capsules w i th the same sensitivity but different frequency re­sponses. The middle curves show the effect of the capacitor in the feedback of the amplif ier, and on the right we have two different cord capacities (corresponding to the feedback capacitor).

In the middle response row, the left curves indicate how important the line matching (sidetone) is, here demonstrated w i th open line and terminated w i th 6 0 0 0 . The two re­sponses on the right show two dif­ferent line lengths, zero and one km (open line).

On the lower left, the caps are fastened wi th different t ightness, and on the right, the microphone cap holes are varied.

Studying the mechanical feedback wi l l clarify the need for elastic sus­pension of the capsules.

11

dB

0

10

- 2 0

- 3 0

- 4 0

Handset if ted 20 mm

Handset ifted 40 mm

777777777

kHz 750481

Fig.4. Ratio of receiver drive signal and microphone pick-up signal for various heights of hand set above subset

dB 0

- 10

- 2 0 -

- 3 0 kHz

dB + 10

0

10 5 kHz

750482

Fig.5. Upper: ratio between receiver drive signal and microphone pick-up signal as the handset approaches a user Lower: Effect of a capacitance C on a frequency-dependent feedback path

difference varies according to fre­quency when the handset is kept in different positions: in free f ield, in normal positions on the telephone set, on table w i th holes parallel to table and on table wi th holes down.

In all examples shown, the set terminal is open and wi thout any line length, if not particularly men­tioned.

The telephone sets are chosen to il lustrate the different

The last mentioned position wi l l give howling at about 3 kHz, at zero degrees phase difference.

measuring methods, and they are not typical for the Norwegian sets produced to­day by A / S Elektrisk Bureau.

If the handset is put on the tele­phone set in its normal position and then gradually l ifted, we can see the peaks in the stability responses get lower and the resonances move to other frequencies (Fig.4).

We found that the phase differ­ence between supplied and meas­ured voltage change about three times 3 6 0 degrees during the first couple of cm the handset is lifted from the telephone set. The signal are in phase three t imes, and if the loop gain is sufficient, unstable con­ditions can occur.

Handset position In the upper part of Fig.5 is dem­

onstrated how the handset position influences the degree of stability: in free f ield, vertical in hand and held against the ear.

The lower part indicates che inf lu­ence of a frequency dependent feed­back in the microphone amplifier (a capacitor), C = 0 and C1 to C4 (in­creasing capacity). The handset is here placed on a table w i th holes down.

Free field We have also kept the handset po­

sition constant, in free f ield, and varied other factors.

On the upper left in Fig.6 is shown the result of two different types of capsules w i th the same sensitivity but different frequency re­sponses. The middle curves show the effect of the capacitor in the feedback of the amplif ier, and on the right we have two different cord capacities (corresponding to the feedback capacitor).

In the middle response row, the left curves indicate how important the line matching (sidetone) is, here demonstrated w i th open line and terminated wi th 6 0 0 O. The two re­sponses on the right show two dif­ferent line lengths, zero and one km (open line).

On the lower left, the caps are fastened wi th different t ightness, and on the right, the microphone cap holes are varied.

Studying the mechanical feedback wi l l clarify the need for elastic sus­pension of the capsules.

11

dB

0 -

- 10

- 2 0 L

C = 0

Cord capacity

kHz

dB

0

10

- 2 0

30

Matching

open

60012

Line length

kHz

dB

0

- 10

- 2 0

i

P

Caps fastening

>— normal

I \t< /Til \

5

Jht

Micr. cap (holes)

nortma arger

kHz 750483

Fig.6. Effects of various parameters on howlback response

0,1 V u

dBSPL

70

60

mV

1

0,3

Fig.7.

2 b 2 5 2 5 kHz 750484

Direct measurement of voltage induced in a microphone output connection when a re ceiver is driven by a fixed amplitude signal

Measuring method 2 We can directly measure the in­

duced voltage in the microphone when the receiver is driven by a sig­nal (Fig.7). This method does not tell us if there is any chance of howling in the system, but for rela­tive measurements it can be of great interest to use such a simple method. Critical frequencies can be studied in this way. We can also measure the frequency response of the receiver at the microphone en­trance.

Below are shown examples of such measurements. On the left, the induced microphone voltage is measured when the handset is placed on a table w i th holes down. In the middle, the microphone volt­age and receiver (loudspeaker) re­sponses are shown when the hand­set is in free f ield, and on the right we have the same responses when the handset is placed on a table w i th holes parallel to the table.

Measuring method 3 A very simple method that can be

used e .g . in production control is shown in Fig.8. Varying the line

Tef. set Artif icial line 48 V 2 x 250 H

750485

Fig.8. Adjustment of line length to find mini­mum required to give a howlback-free system

length, this measurement tells us the min imum length that is neces­sary to give a stable system when the handset is lifted off the tele­phone set, when it is placed on a table in different positions or held in other ways.

12

Factors determining the reproducibility and accuracy requirements of telephonometry measurements. Review of some practical experiences by /. Sal am a

1 . Introduction The intention of this presentation

is to discuss the factors, which from the telephone company's point of view determine the accuracy and reproducibility requirements of the reference equivalent measure­ments.

Let me start by putting a simple question. Why do we have to meas­ure telephone sets? Two answers can be given: we have to check the ful f i lment of the transmission plan and we have to check the quality of the telephone sets.

To discuss how these factors de­termine the accuracy requirements of the measuring equipment, I have to say a few words about the trans­mission plan of the Finnish tele­phone network, and the quality con­trol methods used in the Helsinki Telephone Company.

2. The transmission plan In 1970 a new transmission plan

was adopted in Finland. It is based on the CCITT plan a), Recommenda­tion G 121 (p. 21). In this plan the maximum value of the nominal sending reference equivalent of the subscriber system of the local ex­change is 9 ,6dB (1,1 Np) and the corresponding receiving reference equivalent is 0 ,9dB (0,1 Np).

Based on these values there has been a tendency to draw up the ref­erence equivalent requirements of the telephone set, so that less than 3% of the incoming and outgoing calls in a local exchange area wi l l exceed the given values. There is further a desire to limit the lowest reference equivalent of the local connection to a mean value of 2,6 dB (0,2 Np). Based on these facts the reference equivalent re­quirements for the telephone set as seen in Figure 1 are derived.

The transmission characteristics of a set are determined by connect­ing the set to a subscriber cable of 0,5 mm and 3 7 n F / k m , and to a

feeding-bridge of 60 V and 2 x 5 0 0 O in such a way that the mean value of a suitably large sample of a consignment must fulf i l the specifi­cations in Figure 1. The length of the cable is 0 to 5 km. The standard deviation shall not exceed at any length the value 1,3dB (1,5dNp) on sending and the value 0,9 dB (1 dNp) on receiving. The measure­ments are made wi th an objective reference equivalent measuring equipment (OBDM).

These requirements imply that telephone sets w i th automatic level control have to be used. On the re­ceiving side the cable attenuation

must be compensated completely, whereas on the sending side feed­ing loss and only a part of the cable attenuation must be compensated.

The experience has shown that the combined reference equivalent of the telephone set and the local line is approaching the allowed max­imum value mainly on long sub­scriber lines. If the main objective is that only 3% of the calls of the en­tire exchange area wi l l exceed the limits 9,6 and 0 , 9 d B , a bigger ex­ceeding percent value for subscrib­ers far from the exchange may be al­lowed.

Reference equivalent

10 _ dB

9 -I

8 -

3 -

2

1 -

0

- 1 -

600

The mean value of the receiving reference equivalent

0 . . . 5 km

0 0,5 mm, 37 n F/km

I 2 x 500 n

60 V / / / /

^ j [ _

1 2 3 The mean value of the sending reference equivalent

Length of the cable 1 T

4 5 km

The specified mean values of the sending and receiving reference equivalents of a sample of telephone sets

Standard deviation at sending, 5 max = 1,3 dB Standard deviation at receiving, 5 max = 0,9 dB

750486

F i g . l .

1

SRE A (dB)

10 -

3. 2,5% 2. 34%

5 -

Standard deviation 5 = 1,3 dB

1 2 3 4 T

5 L/km

1 . no. of subscribers in 1 km region 2. no. of telephone sets exceeding 9,6 dB 3. no. of subscribers exceeding 9,6 dB

750487

Fig.2.

SRE dB

12 -

10 -

8 -

9,6 dB

1. 2 . 3.

8,5% 4,7% 65%

3,8% 7 , 3 ° / o ^^T 35% \^T 11% ^ f

1 1 1 3 4 5

5 = 1,3 dB

no. of subscribers in 1 km region no. of telephone sets exceeding 9,6 dB no. of subscribers exceeding 9,6 dB

3. 2. 1.

750488

Fig.3.

In Figure 2 a typical mean value curve of a telephone set is shown. It is supposed that the mean value at a distance of 5 km is exactly 9,6 dB. It is further supposed that the standard deviation is exactly the allowed 1,3dB. The upper end of the curve has been substituted by steps of 1 km. Over both steps there are three numbers. The num­bers shown come from a large in­vestigation made in Finland some years ago. The lowest number indi­cates the percentage of subscribers in the corresponding 1 km region.

The middle number indicates per­centage of telephone sets exceeding 9 ,6dB (assuming normal distr ibu­tion). The uppermost number indi­cates the percentage of the subscrib­ers exceeding the allowed limit 9 ,6dB . When adding the last two numbers it is found that the limit of 9,6 dB is exceeded by about 3,8% of the subscribers.

In reality the situation is, how­ever, not so simple. The mean value curve and standard deviation are varying from one manufacturing lot to the other. However, w i th this method we can roughly estimate the percentage of the subscribers who wi l l exceed the allowed l imit.

3. Quality control Helsinki Telephone Company buys

about 5 0 0 0 0 telephone sets per year. The telephone sets are deliv­ered in lots of about 8 0 0 pieces. Our Quality Control performs a sam­pling inspection of the telephone sets, where the most important functions of the telephone sets are tested and the reference equiva­lents are measured objectively w i th measuring equipment designed ac­cording to K. Braun's principle. The method of inspection is mainly the inspection by attributes whereas the reference equivalent is tested "by variables". The reference equiv­alents are measured w i th cable lengths of zero, 2 and 5 km (diame­ter 0,5 mm). Mean value and stand­ard deviation is calculated for each line length. However, in addition to this, if an individual reference equiv­alent differs more than 4 dB on sending and more than 3 dB on re­ceiving from the mean value require­ment, the telephone set is consid­ered defective and the inspection by attributes is applied. These values are excluded in calculations of mean value and standard deviation.

14

4. The accuracy requirement of a measuring equipment

The influence of the accuracy of a measuring equipment can be exam­ined according to the same method as above.

Let us assume that the measur­ing equipment reads incorrectly 1 dB in the direction that the real reference equivalent is 1 dB higher than the measured value. In Figure 3 the same situation as in

Figure 2 is shown, but now 1 dB higher. Again, it is easy to calculate from the reference equivalents and subscribers distribution given in Figure 3 that the percentage of the subscribers exceeding 9,6 dB is now 4 ,7 + 3,8 = 8,5%. If we fur­ther assume that the error of the measuring equipment is 2 d B , an ex­ceeding percentage of 14% is ob­tained. In addition to this the sub­scribers located nearer than 3 km must now be taken into account.

In practice the mean value of the incoming telephone sets is varying between the allowed limits from one lot to another. However, at 5 km, the mean value is more often approaching the upper than lower-limit. It can thus be estimated that the exceeding percentage wi l l be moderate, if the inaccuracy is about 1 dB. An inaccuracy of 2 dB begins to be critical from the standpoint of ful f i lment of the transmission plan.

Let's compare the result w i th the measuring accuracy of the other parts of telephone connection. The attenuation is usually measured us­ing level meters and signal genera­tors. Nowadays, very accurate lev­el meters and stable generators are available, but very often the ac­curacy of the measuring equipment in use is only about 0,5 dB. In addi­tion to this we must notice other varying factors, for example on car­rier frequency connections about 1 dB must be reserved for variation w i th t ime of the loss. Thus we can conclude that the other parts of the telephone connection cannot either be determined w i th greater accu­racy than 1 dB.

5. Other factors affecting the accu­racy of the measured results — the reproducibility of the results

As mentioned above, the mean value and standard deviation of the incoming lots of telephone sets are varying. This is partly caused by component parameters and produc-tional deviation. Total deviation is also increased by variations of envir­onmental factors, especially temper­ature and humidity.

To enable the comparison of the results of the different lots w i th each other, the reproducibility should be as good as possible. To in­vestigate the reproducibility of the

Fig.4.

Fig.5.

reference equivalent values meas­ured objectively, we made the fo l ­lowing test. 10 telephone sets equipped w i th carbon microphone and 5 telephone sets equipped wi th linear microphone were chosen. The sending reference equivalent of

them was measured on 20 succes­sive days and the mean value was calculated every day. The results can be seen in Figures 4 and 5. From these figures it is immediately noticed that the mean values for car­bon microphones may vary widely.

15

Telephone sets equipped wi th l in­ear microphones vary considerably less. It is diff icult to say, how this variation should be divided between telephone sets and measuring equip­ment. However, it is evident that for stable telephone sets the reproduci-bility of the results is acceptable if the deviation is of the magnitude 0,5 dB.

it should be possible to achieve the same reproducibil ity w i th differ­ent measuring equipment. Our expe­rience has shown that if the differ­ence between the mean values ob­tained by the manufacturer and the customer is more than 1 dB, it w i l l be difficult to decide who has got the correct value. Wi thout any fur­ther examination it can be con­cluded that results obtained using different measuring equipment must be w i th in the accuracy l imits.

6. Practical quality control As we know that reproducibility

of the results of the carbon micro­phone telephone sets is bad, we cannot immediately return a lot of telephone sets, if the mean value slightly exceeds the permitted limit. Our Quality Control logs cont inu­ously the mean value and standard deviation of the incoming lots in the manner seen in Figures 6 and 7. Steps are taken against the sup­plier, if the mean value remains con­tinuously outside the limit or if the mean value deviates considerably (1 dB) from the allowed limit.

The standard deviations (maxi­mum 1,3dB on sending and 0 ,9dB on receiving) are also very good quality criteria. If the standard devia­t ion suddenly increases, it is usu­ally a sign of some production error, usually when assembling the car­bon microphone or receiver.

It happens, however, sometimes that we have to discard the lot be­cause of our inspection. Our inspec­tion system enables us to prove to the manufacturer that the returning is just i f ied.

7. Conclusions I have tried to describe the accu­

racy requirements of the objective reference equivalent measuring equipment implied by transmission plan and quality control. The conclu­sion is that the accuracy ought to be about 1 dB (maximum 1,5dB) re­gardless of the principle of the meas­uring equipment. A different sub­scriber distribution and different re­quirements for reference equiva­lents might lead to a different accuracy requirement, but I think that our final result is quite realistic as it is based on both examination above and practical experiences.

1974

date of insp.

Jan. February March

S R E a t 5 k m

Mean value and standard deviation

April May June July August September October

30 5 7 12 13 14 15 18 25 1 7 7 14 18 27 1 5 16 18 22 25 30 7 13 22 23 28 29 5 10 13 12 20 26 28 2 6 5 15 12 16 22 29 3 5 ^3 16 18 27 3 14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52

The inspected tots

Fig.6.

dB

RRE at 9 km Mean value and standard deviation

750491

Fig.7.

16

Frequency response measurement on the telephone sets 73D, F68 and 47E connected via a telephone line by S. A. Jager and P. Schnack, Jydsk Telefon A / S .

en

en i -

n c

>

CO =3

<

o \— o

"O C H3

^ CO

1 —

> C/l t o 3 C"" L—

<

CD u , CD

" Q c CO

1-

CTR

N /

\ /

. /

Fig. 1. The measured chain 750160

Introduction From the State Audiology Centre

in Arhus, Denmark, the Danish tele­phone company J.T.A.S. received a request regarding the frequency re­sponse and sound pressure levels to be expected from normal tele­phone connections in Denmark. To answer this question experiments were made wi th commonly used telephone sets and an arbitrarily chosen trunk connection.

The Measuring Arrangement The trunk connection was set-up

over Tranebjerg — Arhus South — Skanderborg — Arhus South and back to Tranebjerg, involving five telephone exchanges. Measure­ments were made wi th the B & K equipment shown in Fig. 1.

Various separate calls were made using three different types of tele­phone set, and for each call fre­quency response curves were re­corded. A comparison of the results shows that the connection itself was almost identical in response from call to call. The curves shown in this report may therefore be re­garded as typical for the particular trunk connection chosen for the ex­periments.

Important Note At frequencies below 2 0 0 Hz the

response shown on the diagrams are not correct due to excessive dis­tortion from the Art i f icial Voice sig­nal.

Transmission Line Measurements Fig.2 shows the frequency re­

sponse of the "pu re " telephone-con­nection Tranebjerg — Arhus South — Skanderborg — Arhus South — Tranebjerg, loaded by a 6 0 0 O re­sistance in each end. As the con­nection between Skanderborg and Arhus South is of the carrier fre­quency type bandwidth l imitations are present. These can be clear­ly seen from Figure 2 (bandwidth: 2 5 0 H z — 3400Hz) . The peaks and valleys occuring in the 1 0 0 0 H z to 3 4 0 0 Hz region indicate artificia (Pupin) loading of the line.

Measurements with Different Telephone Sets

In the fol lowing f igures, various frequency responses are shown, in­cluding sender (A) set and receiver (B) set connected via the above de­

scribed line connection, see also Fig. 1. A brief outl ine of the charac­teristics of the telephone sets used in the experiments is given in Ap­pendix A.

Typical response using the tele­phone set 73D as transmitter (A) as wel l as receiver (B) i. e. A : 7 3 D — B: 7 3 D is shown in Fig.3.

Note that over the complete fre­quency range 3 0 0 H z — 3 4 0 0 H z the received sound pressure level is of the order of 80 dB, wh ich , accord­ing to several investigations is the preferred mean sound pressure evel for normal connections.

Typical response using the tele­phone set F68 as transmitter (A)

] n □ Q a n a n n n a D D n n a D n D n n n D n a a a a D a n a □ D n n o n n n D a D a Q n a D n n c BKiei&KjEBr Briiel & Kjaer

Cop4nhiapan 50r25

dB Measuring Object:

Transmissions-linie

40

Tranbjerg-Ar, syd -Skanderborg 3 0

dB

■20

-15 BF t i l Ar. syd Tranbjerg

Rec No.: D a t e : _ 2 M 2 - 7 3 Sign.: ?S.,

20

RMS

-10

Rectifier.:. Zero Level: 10 m V . I Lower Lim. Freq. 20 Hz 5 Potentiometer:25 dB[_ Writing Speed:80 m/m s e k ~ Paper Speed:_10 m/m s e k "

Multiply Frequency L t ^ _ oL 0 5 =

1 0 V -Scale by- 1

4+r

I Oi75

d&

8 - 6 0

dB

6 -45

4 - 3 0

200 500 1000 2000 5000 10000

Q P 1 1 4 2 T50161

Fig.2. Characteristic frequency response of the plain transmission line

17

8L

8 9 d o\ i t * 9"6!d SSiOSl ZfrLL dD

0000 L

09 -8

s^kj

0009 oooz

: :

001 0 10 I ■ i "Aq eieos

>tas LU/UJ o i~;pe»ds j a d e j Has UJ/LU o g : p ^ d s S U S ^ M

~|8P S3^Joiaujoiiueiod q.*HOZb9ii rLun ««<n

■s-d "u6!S

01- 03

SL-OE

03-

8P

93 ^

89d — 3Z.t

Otr iDefqO Suunsee^j

SP H IN U

t a n n n n G n n n a n D D a n n n n D D n n n a n a n a a a a D D a n a a D D n n n n a a n D D n a JSef^ 19 | 9 n j a

Q£L o i 3Z.fr 9"6|d

9 £ K U

D a D a D n n a a D a a D D n n n n n n n a D a a n n n D a n D n a D a a a D a a a a a a a a n n D t l

S Z -* 0 S LI*. JB^U *dC3

aez_ 01 89=i "fr'Bu £91091

0000 L ro

3frU dD

0009 ' \' "Aq f>\*3$ AouenbeJj A|di»|nyy

>|&S a i / U J O L _ ; p o a d s J o d o d

>|3S UJ/LU 0 8 : P & e d S Su'lHM "lap 92" j^SLuo^uoi°d

^ z H 0 2 " b o j d LU ! l JO WOT ndsapgg^*^ ° i a

siAiy -^*!*3e

g*-Zl-OZ :eiB(

~ : O N 30]

Q£Z - 89 d

8P ^ ^

QZ.KH

t j a n a n n a n G D D D a n n u G a D G D n n D a D n a a a G a n n n a Q D n D n n n n n D D a a a c

Q £ £ 01 BujuiLUSUBJj. a e z . 9uoqda|sx '£"^!d Z9tO$£

0 0 0 0 1 ro

SI -3

oe - *

9t>

0 9 - 8

gp

000s =F 0003 000 L 009

- 9

8P

9 i k l L

J L 10 n o

"I Aouer

ZfrU dD 00 L

I :Aq Q|BOS lenbei j A|d<i|nftj

fas LU/UJ QL-:peeds jaded as UU/UJ Q8 :P s e dS 6 U I IMAA

"|gp g2:j&lsajo!lUB1°d 9 ZH 0 £ ' b e J d *"!"! JB^o i

" I d S 8P 99 : l0Aei OJOZ

~ ^ $ 3 — " u 6 | S

nL , "E*-ZL-oT~wa

QL-

03-

SP

oe aci - ati

ov : ioefqo Buunsea^j

92-*

gp

OS

*&rf*

JSe[>l 'S l 8 P j a jae!» * ia™a j n n a a a D n a n n D a a a a n n D n n a n n D D i n c n D a a D n D D n n n D n a a a a n Q n n D c

A|JOOC1 S|eu6fs Aouanbajj. M O | saAjao -aj 11 JEU,J pue ' jauj iusuBJi poo6 A|8A!ie|9j e SUJBIUOD 89=1 i9S auoqd - a p l BLJJ jeq i papnpuoo aq snqi Aew if a jnb j j 6ujo6ajo^ pue SILJI LUOJJ

"8P ZB J° uoiBaj aqj uj s| ZHOO^C 01 ZHOOOt a6uej Aouanb -aj^ UJ |3Aa| ajnssajd punos paAiao -aj aqj UJB68 ' jaAaMon zj-iOOOL °* Z H 0 0 £ 86UBJ Aouanbaj^ agj u ! 8 P 0 8 AAO|aq ieqAABiuos si |8A8| a jnssajd punos paAjaoaj aqj a jan

7_"6ij uj UMOtjs S| 3 9 J :g 89d : V 9 '! '(8) JQAjaoaj se ||3AA

se (v) J8UJLUSU9JJ se 89-j ias auoqd -a|aj a i j j 6ujsn asuodsaj |eoidAi

■;as 89d 8M1 UJ jaAiaoaj aq j p Ajueaun -uou Aouanbajj. a|qejapisuoo aq j pug jas 3/_-t? aqi ui j au j iusue j i /(>)BaMj(

jeqAAaiuos aq j p uoneujqujoo aqi 01 anp sj smj_ pauieiqo s\ aPS8 aixios p jaAa| a jnssajd punos paAiao -9J 8 ZHOOt^e 01 ZHOOOL s6uej aq i ui aj iqM 'gp g/_ p jap jo aqi p SJ |9A3| a jnssajd punos paA|ao -aj a iu 'ZHOOOI — Z H 0 0 £ ^^ue j Aouanbejj ei\\ u\ IBLJJ uaas SJ 1

•9-61-1 ui UAAOMS Sj 89d :a — 3Zlt7 : V a j '(g) jaAjaoaj se 89=1 J^s au,i pue

(V) jau iu jsue j j se ^p ias auoqd -a|ai aqj 6uisn asuodsaj leojdAi

"E 'Bld

si 3i_p jas auoLjdaiaj au,j u; j a i -lILUsueJi aqi se 'zHOOt^C — Z H 0 0 £ a6uej Aouanbaj^ aiajdiuoo au,i jaAO g p 0 8 ueMJ J9AAO| ;eu,Maiuos si |9Aa| a jnssajd punos paAiaoaj aqj_

■g-6!j u; UMoqs s; Q££_ :a — 3ZLfr ^V a 'j '(g) jaAjaoaj se Q£/_ jas aqj pue

(V) j au i i usue j ; se 3/_t7 l Q S ouou,d -a|ai aqi 6ujsn asuodsaj |60idAi

las Q£/_ au,i UBLJJ

j au ju j sue j ; ^ ja^eaM^ e seq 89d *8 S

auoqdajai aqj asneoaq si Sjq_L 'C'Bj-j UJ UMoqs ieq j ueq ; J3MO| jeqAAaiuos 'j3A3Moq 'SJ |aAa| s jnsss jd a6e -jaAe a q i Z H O O ^ C — zHOOC 96UBJ

Aouanbajj, aja|dujoo aqi JSAO g p 0 8 p jsp jo sq j p si |3A3| s jnssajd punos paAjaoaj aqi 'U IB6B ajaH

■t7'6|-j ui UMOL|S si a£Z. B — 89d : V a I '(g) jaAjaoaj SB Q£/_ jas aq ; pus

Typical response using the tele­phone set 73D as transmitter (A) and the set F68 as receiver (B), i. e. A: 7 3 D — B: F68 is shown in Fig.8.

Once again, the poor receive char­acteristics of the F68 apparatus re­sults in a response with a relatively large frequency nonlinearity. In the region 300 Hz — 1000 Hz the re­ceived sound pressure level is of the order of 77 dB, while in region 1000 Hz to 3 4 0 0 Hz it is approxi­mately 84dB. The overall received sound pressure level is, however, somewhat larger than that indicated in Figs.6 and 7 due to the fact that the transmitter in the 73D set is stronger than those used in the 47E as well as in the F68 sets.

Typical response using the tele­phone set 73D as transmitter (A) and the set 47E as receiver (B), i. e. A: 73D — B: 47E is shown in Fig.9.

The received sound pressure level is here somewhat above 80 dB over the complete frequency range 300 Hz to 3 4 0 0 Hz, as the set 73D contains a relatively strong transmit­ter, and the 47E set receives wel l . The receiver in the 47E apparatus does, although not too pronounced, show reception properties better in the frequency range above 1000 Hz than below 1000 Hz.

Typical response using the tele­phone set F68 as transmitter (A) and the set 47E as receiver (B), i. e. A: F68 — B: 47E is shown in Fig.10.

It is seen that the received sound pressure level is around 80 dB over the complete frequency range 300 Hz to 3400 Hz, although con­siderable frequency nonlinearity is present.

Typical response using the tele­phone set 47E as transmitter (A) as well as receiver (B), i .e. A: 47E — B: 47E is shown in Fig. 1 1.

Here the received sound pressure level in the frequency range 300 Hz — 1000 Hz is approximately 76 dB, while in the frequency range

] □ a u a o □ [ Bri ie l & Kjaer

Copenhagen c Q

• ^ dB

Measuring Object: — . A ("1

4 U

F68-F68 30

Rpr N n __ Jl IJ

Date :_ -20 - l 2 -73_ Sign.: P.S Rectifier.. RMS

■ i n D n a n a D D D n n D n n D D

Zb

"dB

-20

-15

-10

Zero L e v e l : 6 5 d B S P L . L o w e r U r n . F r e q . 2 0 H i 5 P o t e n t i o m e t e r : _ _ 2 5 d B

W r i t i n g S p e e d . 8 0 m / m St P a p e r S p e e d ; _ l 0 m / m se

1 j

Multiply Frequency L Scale by - 1 c\\ nl

OP 1142

T

^ 1

i

, .

a a a D [ Bri iel I

f

L j

■ I I J\

■k I / >k r^

1

[ I

1

j 1

1

u n □ Q i Kjaer

D n n n D n a a n D L

TT m m

" r t ' L 1

r 1 i I f 1

|

1 ■" F 1

■■ , ■ ■ ■ ■ ■ ■ ■ ■ ■ — ■ ■ ■ ■■■

■ m

J □ D □ □ D □ □ I

f i

|

[ f

■■ 1

1 I

1

\

1 I 1 1 [ _. . . .. . H ^

l | I 1 1

i= I

I

1

, - * - ^

| | [ ' | 1 ,—,

|

j _ , — i ■■ i

t 1

dB

8 -

1 ._..

[ E

1 L 1 ■

t T I

! i 1 1 j I I I

i o -

4 -

2-

n

/ 3

dB

60

45

30

15

100 Hz 200 500 1000 2000 L 5000 10000

750166

Fig.7. F 6 8 t o F 6 8

i n n n n n n D D D D D a n n D D D a D D a a n n D D n n n a n D a a D D D D D a D D n D D a n D D t Briiei &Kjt»r Briiel & Kjaer

copflnrmoBr 50 r25

dB

Measuring Object: 40

dB

■20

73D - F68 30-15

lee. No.: 20-1 D a t e : _ 2 0 : l ? : Z 3 _ [ Sign.: P S - ^ ^ectifier.:_RMS_ Zero L e v e l : 6 5 d B S P L . Lower Lim. Freq,20 Hz Potentiometer:-^ 25 dB Writing Speed: 80 m /m Paper Speed:- 1 0 m/m Multiply Frequency L Scale by.: . 1 Q L

K = T Oi75

II m

dB

8-

6 - 4 5

dB

60

4 - 3 0

- 1 5

100 200 500 1000 2000 5000 10000 0

QP 1142 750167

Fig.8. 73D to F68

3 □ □ □ D 0 D [ Bri ie l & Kjaer

ri " ■—■ PVafj n r i m n^tWi lrJ" ■ 'b

i . □ p4rini gtrn C n

^ ^ d B

M e a s u r i n g O b j e c t :

4 U

73D-47E ■ — ^ n

o u

R n r N n ^ n

Date: 20-12-73 Sig r . P .S .

^ e c t i f i e r . : _ R M S _

■ D D Q o n a n D n a D D D f :

25

dB

20

-15

-10

Zero L e v e l : 6 5 6 B S P L ^

L o w e r L i m . F r e q . 2 0 H z 5

P o t e n t i o m e t e r : - 2 5 d B L W r i t i n g S p e e d : 8 0 m / m 5E P a p e r S p e e d : _ J 0 m / m se

^^^m a^^ ^^ ^ ™ a^ ^^

_ _ L ^BB 1 1

M u l t m l v F r e n u e n r v 1 1

f

1 -^ - f

J I

!K ™ — p

,

— , — , , . 1 . . . .

■K: ^ 1

-1 1 E 1 / Sea e b y . : 1 nL n l ■ w -w -m— -m- j ' " ™ ■ ■■"■" %J "

QP 1142

11 30 F Jm t

— ■

: □ i : □ n n a n a n a a a n n n a a Bruel & Kjaer

1 r i T

— ^ —| |

— — - «

p - - " - T W - | ' V j i

1 ■ " t '

D D L

'

/ 1 1 /1 1 V i

i

— i — i — i

■

,

i

■

T I 1 j - - ■ . i.

.

i i - I \ 1 - —

/ ■■ '§]}lm ^f—

1

J 1

J 1 1 T

j D a a D n o n n n r

i i

~\ I I I I

I i

1 V J 1 — 1

1

1 1 1

1 I 1 LJ 1

. 1 1 F 1 I I j r \ \ r

'

dB

8-

1 |

1 1 | |

_ _

|

1 1 1

I i

6-

4 -

2-

i

n.

dB

60

45

30

15

\z** 200 500 1000 2000 5000 10000 v

750768

Fig.9. 73D to 47E

l a a a D D □ □ a n n n n n n a D D a Q D a D n a n a D n D n a a D n n D n a n a a D D D D D D D D r Bmei RKj r Briiel & KjaBr

Cop#Plhagan

Measuring Object:

F 6 8 - 4 7 E

Eec No.: a ie:^-20-12-73

Sign.: F \ S , _ Rectifier.: RMS' Zero Level: 65 dB SPL, Lower Lim. Freq. 20 H z ^ Potentiometer: 25 dB[_ Writing Speed:80 m/m sek Paper Speed:-10 m/m sek Multiply Frequency L Scale b y . : _ 1 . Q L 0

[QP 1142

Fig.10. F 6 8 t o 4 7 E

19

1000Hz to 3 4 0 0 Hz it is in the re­gion of 82 dB. This is due to a com­bination of the 47E sets relatively weak transmitter and the tendency in its receiver to receive low fre­quencies insufficiently.

The major results of the described experiments are summarized in the table.

Conclusion By looking at the figures 3 to 1 1,

as well as the summarizing table, it seems that the best and most pleas­ant sound picture, wi th highintel l igi-bility, is obtained when the tele­phone set 73D is used both as transmitter (A) and receiver (B).

Furthermore, in all combinations where the set 73D is used as re-

Fig.11. 47E to 47E

3 □ □ D D D D [ Bruel & Kjaer

^r3> dB

Measuring Object: , . — A C \

4 7 £ - 47E 4 0

■-■% #"%

— — — 30

■

"dB

■20

-15

P^i ft r

Ht*r. N o - I Q I * r t D a t e : „ 2 0 - 1 2 - 7 3 _ * " w

^ign P-S. Rectifier.:, RMS. Zero Level: 65 dB SPL. Lower Lim. Freq. 20 Hz 5 Potentiometer: 25 dB_ ■ B. ■ .■*-& C J f l _ ' _ _ J B-M-TM. . H .

T

J ,

1 | — T j .* i — ■ — — ■ — ^ " ™ - ■

n

P ^^™

J ^ r -

— 1

Wr i t i ng £p* f tH" OU m / m StfK \ 4 A /

Paper Speed: 10 rn/m seK |— Multiply Frequency L 1 1

Rc^nlft hy 1 n! nl J ■ ■ \ f

Q P 1 1 4 2

TOO y

~m ~ i

3 □ t 1 a

™ | 1

|

| 1 _

1 . — f —

J r - ~ M 1 J

1

^_^__.—„—

□ n a n C D n a n n n n D n n n n a r Bruel & Kjaer

■ I ■ j '

1 1 -i h*~^ ^ . f- .

£ 1 _ F — ■

f |

1 H — 4 - H ^ ■—■

yf

|

1

-"

(

' * '

r r r ^ B

a .

L_

■ ■ ■ ■ m J 1

I 1 1

] D 0 D D D D D C

^ 1

f ! 1

T - * * " ~ i

f ! 1

1

^ p

1

1 1 J

1

■-■ 1 ■■ 1 rwmj

^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ™ T

1 _ ^

"~ '

[ 1 —

[ 1 I

J ^ ~ — -

| | 1 J \ !

■ ■ ■■—# h

1

T

j 1

| — |

1

r L

1 ' 1

\ "

* r ■ * 1 N f "*" 1 ■—1 ^ X T V J

I r I

rT"1 1

- W I H J 1 _

J 1 \ 1

1

J

F

1

_ ^

. f .

** V_I_M_^H

1

._._. ^

| | |

J L 1

1

1

1 u -

8-

/a

dB

60

6 -

4 -

2^

n

45

30

15

[ T ^ ^ O O 500 1000 2000 ' 5000 10000

75G170

ceiver (B), good results are achieved, and the sound pressure level vs. frequency characteristics is

reasonably constant around the pre­ferred mean sound pressure level of 8 0 d B .

Transmitt. (A)

73D

F68

— ■ — —

47E

47E

1

i 1

F68

73D

73D

F68

47E

Receiver (B)

1 ■ - ■ " ^ ■ ■ n v

73D

! 1

1 1

1

i

73D

73D

F68

■ .

F68

F68

47E

47E

47E

Receiver Sound Pressure Level 3 0 0 H z — 1 0 0 0 Hz

81 dB SPL

80 dB SPL

i

78,5 dB SPL j 1

7 4 , 5 d B SPL 1 1

1

77 ,5 dB SPL

1

77,5 dB SPL

1 1

j 1 i

81,5 dB SPL

1 m_j _ 1 1

1 1 1 1

80 dB SPL

76 ,5 dB SPL

1 0 0 0 H z — 3 4 0 0 Hz

81 dB SPL

8 0 dB SPL I 1

■

78 ,5 dB SPL

8 2 dB SPL

8 2 dB SPL

8 4 dB SPL

1

i

8 5 d B S P L

1

1 T

[ 85 dB SPL 1 1 >

L 1

■J ■ ■ ■

1

i

1 81 r5 dB SPL

Remarks 1 1 h-rf 1 1 1 t „ J ■ ■ \ " i _ "

Fig.3 Both high frequency and low fre­quency ends of the passband are somewhat pronounced, resulting in a pleasant and highly intell igible sound picture.

Fig,4. As Fig. 3 but w i th a somewhat lower overall level.

Fig.5 As Fig,3, The overall level is, how­ever, still lower than that obtained in Fig.4,

Fig^6. Fairly high attenuation in the low frequency region. This gives a sort of "meta l l ic" sound picture which is less pleasant for the listener.

Fig.7. As Fig.6. The attenuation in the low

i frequency region is less than in ! Fig.6, which results in a slightly

more pleasant sound picture. 1

Fig.8. As Fig,6, but w i th a higher average sound pressure level, result ing in a

; better intell igibil ity. i . „ . - . —_____ t ■ " ■ I - - - - - - . . . - _ . . . .-■■

I Fig.SL Fairly high overall sound pressure level and only slight attenuation in the low frequency region resulting in a pleasant sound picture and good intell igibil ity.

Fig.10. As Fig,9, the overall received sound pressure level isr however, slightly lower.

F ig .11 . As Fig.6, w i th less pronounced at­tenuation in the low frequency re­gion.

Appendix A

Brief Classification of the Telephone Sets used in the Experiments

The Telephone Set 47E is a con­ventional set containing a carbon microphone and an electromagnetic receiver cartridge. It was introduced into J.T.A.S. in 1 947 .

The Telephone Set F68 is again a conventional set w i th carbon mi­crophone and electromagnetic re­ceiver. It is, however, supplied wi th a varistor in the microphone circuit to compensate for differences in feed-currents between different lengths of lines. This should al low for relatively constant sending (transmitting) level. The classifica­tion No. F68 referes to the year of introduction into Denmark, 1 9 6 8 .

The Telephone Set 7 3 D is a set all-electronic trans-contaming an

mission system, and electrodynamic microphone and receiver cartridges. The apparatus was ready for produc­tion in 1 973 .

20

What should be required of a telephone measuring equipment? production control experiences by O. Larsson

Requirement

Accuracy + 0,5 dB Easy to calibrate Easy to handle Fast measurements Insensitive to low frequency room noise

Purpose

Laboratory

X

Manufacturing

X X

X X

X

Table 1

1 . Introduction Normally we require that tele­

phone transmission measuring equipments both for laboratory test­ing and production control shall measure characteristics close to true Reference Equivalents (RE). By the definit ion true RE's can only be measured subjectively (Recommen­dation P72 CCITT Green Book Vol. V). However, as loudness (in this case represented by RE's) is a main factor contributing to the trans­mission quality, we want to make a measurement that brings us as close as possible to the true RE's.

Other important transmission characteristics are for instance fre­quency responses, impedance and distort ion. At manufacturing it is both unpractical and uneconomic to measure all characteristics at the f i ­nal testing and we have to keep to the most important ones.

2. Measuring Equipment for Different Purposes

The requirements on laboratory equipment for objective measure­ments do not necessarily have to be the same as those for manufactur­ing control. In our opinion the differ­ences can be described by table 1.

The specific requirements for manufacturing control caused us to develop a special transmission tes­ter, 2 /LTM 3 5 7 5 0 0 , which is used in all LME factories w i th telephone instrument production. Sending, Re­ceiving and Sidetone w i th and w i th ­out local cable are tested.

In the laboratory we use both Bruel & Kjaer 3 3 5 0 - 5 2 and 2 /LTM 3 5 7 5 0 0 .

When the microphones and re­ceivers are manufactured as compo­nents, they are tested regarding-level and frequency response. To­gether w i th the f inal control of send­ing, receiving and sidetone on all telephone instruments we think we have a good check of the trans­mission quality. (Other characteris­tics are checked by random sample tests).

3. Transmission Testing Equipment 2 / L T M 3 5 7 5 0 0 (comparison with B & K 3 3 5 2 and NOSFER)

The 2 /LTM 357 5 0 0 Trans­mission Testing Equipment is de­scribed in LME/G 1 5 5 1 - 9 9 8 Ue (submitted to the participants in the meeting). In principle it works wi th whi te noise, speech weight ing net­work, artificial ear, artificial mouth, VU meter, amplifiers and attenua­tors.

As mentioned under 2 above we use both B & K 3 3 5 2 and 2 /LTM 357 5 0 0 in our laboratory work. The theoretics for loudness are not so wel l simulated in 2 /LTM 3 5 7 5 0 0 as in B & K 3 3 5 2 but

some confusing results have been obtained and are compared and commented below.

In table 2 are listed sending mea­suring results at 0 0 line for two types of telephones w i th carbon mic­rophones.

As can be seen 2 /LTM 357 500 gives a good estimation of the true SRE's for both sets whi le the OREM-A equipment shows a difference of 2,2 dB for set A and 5,6 dB for set B showing that a correlation figure must be related to the specific tele­phone sets concerned.

Another example of sending meas­urements is shown in table 3.

Method

Estimation by 2/LTM 357 500 Subjective tests rel NOSFER

Estimation by OREM-A

Measurements of Sending Reference Equivalents Tel set A

+ 4,4

+ 4 , 0 ( 9 5 % ±0,5)

Diff 2,2

1 + 1,8

Tel set B

+ 4,5

+ 4 , 6 ( 9 5 % cont. int. ±0,4)

Diff 5,6 t

- 1 , 0

Table 2

Method

Estimation by 2/LTM 357 500

Subjective measure­ments rel NOSFER

Estimation by OREM-A

Measurements of Sending Reference Equivalents Tel set B

0 km + 4,5

Diff 0,1

+ 4,6

Diff 5,6

- 1 , 0

3 km 0,4 mm line + 12,1

Diff 1,1

+ 13,2

Diff—4,1

+ 9,1

Table 3

21

Here we can see that the differ­ence between objective and subjec­tive measurements is not the same for 3 km local line as a Okm. Again 2 /LTM 3 5 7 5 0 0 is closer than B & K 3 3 5 2 . This is confusing be­cause the more accurate realisation of the loudness theory should in the­ory make the B & K 3 3 5 2 show more accurate values (except for a calibration constant). However, the t ime variation of the noise signal is closer to that of speech than the si-noidal, so we have a feeling that this can be an explanation. There might also be others.

4. The Customer -The Manufacturer

CCITT LAB

The discussion above does not prove that 2 /LTM 3 5 7 5 0 0 is a bet­ter laboratory equipment than B & K 3 3 5 2 . In fact we at LME use both. On the contrary it may happen that telephone instruments can be found, for wh ich B & K 3 3 5 2 gives the closest correlation to true SRE's. But the results imply that the importance of simulat ing the t ime-ampli tude characteristics of normal speech should be included in the study to make a good equip­ment for loudness rating measure­ments.

Very often the customer (the ad­ministration) requires tolerances of SRE and RRE of about ± 3 dB.

Manufacturer

In the CCITT documentation is shown that differences between lab­oratories for subjective measure­ments can be expected to be ± 3 d B . If we then add a transfer factor from subjective to objective values, we may land in a very bad accu­racy. Therefore, when an agree­ment is reached on the acceptance of a certain type of telephone instru­ment (of course based on true refer­ence equivalents) there must also fol low an agreement of objectively

measured values related to a spe­cific measuring equipment. Other­wise it wi l l be almost impossible to keep the tolerances wi th in the ± 3 dB wanted by the acceptance control. Thus the acceptance con­trol and the manufacturer shall pre­ferably use the same type of testing equipment to minimize the measur­ing problems. It is not said that the best laboratory equipment is the best one for this purpose too. On the contrary the requirements for manufacturing is likely to agree w i th those for acceptance control for instance for economical rea­sons.

The fol lowing example shows the practical arrangements used by L. M. Ericsson and the Swedish Ad­ministration (Tvt). In the laboratory LME and Tvt use subjective meas­urements to determine loudness rat­ings and other transmission charac­teristics. The laboratories are also equipped wi th OBDM and 2 /LTM 357 5 0 0 for supplementary meas­urements. A t manufacturing and for acceptance control Transmission Tester 2 /LTM 3 5 7 5 0 0 is used both at LME and Tvt.

C 1 9 7 9 B & K Ins t ruments , I n c . , C l e v e l a n d , 0 H P r i n t e d in U . S . A .

X$8cJ\ Instruments, Ina Bruel & Kjaer Precision Instruments 5111 W. 164th Street, Cleveiand, OH 44142 / Phone: (216) 267-4800