Professional Documents
Culture Documents
INTRODUCTION
In a paper presented at the 110th AES +48V
Convention, Hebert and Thomas
described the “phantom menace,”
wherein phantom power faults can R1 R2
6k81 6k81
damage audio input circuitry [1]. Their V+
V+
was described as a “fault at both
inputs, simultaneously.” RG OUT Pre OUT
REF
However, this is not the only fault
condition that can occur. New fault
V-
R G2
C2
mechanisms are considered in this M ic I N+ I N+
3000
2500
mA
2000
1500
1000
500
0
V - P in C1 C2 D12 D22 Q3 Q4 D14 D24
P eak C urrent, mA 3770 1925 1925 1166 1166 720 720 720 720
-1.5
V R 4 - (V )
-20
-2
-25
-2.5 -30
22 mF -35 22 mF
-3 47 mF 47 mF
100 mF -40 100 mF
-3.5 220 mF -45 220 mF
470 mF
-4
470 mF Fig. 6. Voltage -50
0 5 10 15 20
0 5 10 15 20 across bias T ime - (ms )
T ime - (ms )
resistor R4
For the common-mode fault, no When SW1 is closed, –48 V appears tion), fault currents stop flowing
current flows through the base-emitter instantaneously across bias resistor R4, through the IC and the remaining
junctions of input transistors Q1 and as shown in Fig. 6. As in the common- charge on C1 bleeds off through R4.
Q2 because they are reversed biased mode fault, input capacitor C1 As in the common-mode fault, this
during the fault. In addition, no discharges to the V– rail primarily gives rise to the two time constants
current flows through the gain resistor through protection diode D12, but also evident in Fig. 6.
RG. As well, the fault currents in the through Q3 in series with D14. Fig. 7 shows the levels of these
input bias resistors R4 and R5 are Additionally, in this mode, since the currents. In the single-ended fault, the
relatively small (approximately 40 IN– pin initially remains near –48 V, simulation predicts that the input
mA peak for 1.21 kΩ resistors). current flows through Q3 in series with capacitor will deliver a peak current of
RG and D24, and through Q3 in series almost 2.4 A, almost all of which
SINGLE-ENDED FAULT with RG, the base-emitter junction of flows through the IC’s negative supply
A single-ended fault occurs when only Q2, and D22. For simplicity, we only pin (see Figs. 7 and 8). Individual
one input is connected to ground with describe the case in which C1 is components within the IC are exposed
phantom power turned on, and without connected to ground. The results are to significant currents.
a microphone connected. This type of symmetrical if, instead, C2 is To generate the simulation data
fault is simulated by closing one connected to ground. Once the IN– pin presented here, the gain resistor, RG,
switch in Fig. 1, SW1 or SW2, while reaches the negative supply rail was set to 10 Ω. This corresponds to a
the other remains open. (assumed to be –15 V in the simula- high-gain condition. RG plays a ➥
J. Audio Eng. Soc., Vol. 58, No. 3, 2010 March 199
The 48 -Volt Phantom Menace Returns
2000
1500
mA
1000
500
0
C1 V - P in D12 Q3 RG D14 D24 Q2 D22
P eak C urrent, mA 2308 2220 1166 1103 589 514 300 289 241
S ingle-E nded Fault Fig. 8. Current its associated input capacitor to fully
through input discharge before the second short
0
capacitor C1
occurs. However, the authors have
found that the largest currents flow if
-0.5
the first input capacitor is completely
discharged before the second input is
IC 1 - (A)
-1
shorted to ground. Accordingly, this is
the case analyzed in detail.
-1.5 The simulation starts at time t = 0
22 mF with SW2 closed. Input capacitor C2 is
47 mF
-2 connected to ground and contains no
100 mF
220 mF charge. At time t = 2 ms, SW1 is
-2.5
470 mF closed, which shorts input capacitor C1
0 5 10 15 20 to ground. The results are equally valid
T ime - (ms )
for the other possibility, C1 connected
very important role here because its sequentially, with phantom power to ground followed by C2.
value determines the amount of current turned on, and without a microphone When the differential fault occurs,
that flows through input transistor Q2 connected. At first glance, this fault a –48 V transient appears across bias
and protection diode D24. The higher might be considered a pair of single- resistor R4 (Fig. 9). Input capacitor C1
the value of RG, the lower this current ended faults in sequence, but in the supplies high currents (Fig. 10) as it
will be. For relatively high RG values single-ended fault, one input is shorted discharges. As in the other fault mech-
(low preamplifier gains), the current while the other remains open. In this anisms, the simulation shows that this
path to the V– rail is mostly through paper, the authors define the differen- current flows to the V– rail by splitting
protection diodes Q3 and D14, and the tial fault to occur when the second through several paths: protection diode
peak current is reduced. For single- input is shorted after the first input has D12; protection diodes Q3 and D14;
ended faults, the highest currents are been shorted, and where the first input Q3 in series with RG and D24; Q3 in
generated with the preamp set to the remains shorted during the second series with RG, input transistor Q2,
highest gain condition. input’s short. and protection diode D22; and through
Of course, the time between the two Q3, RG, Q2, and input capacitor C2.
DIFFERENTIAL FAULT successive shorts will not necessarily As with the other mechanisms, after
A differential fault occurs when the follow any particular pattern. The first the voltage across R4 becomes more
two inputs are shorted to ground short need not persist long enough for positive than the V– supply (assumed
200 J. Audio Eng. Soc., Vol. 58, No. 3, 2010 March
The 48 -Volt Phantom Menace Returns
here to be –15 V), the current through Fig. 9 Voltage across
Diff
ffe rential Fault
Differential bias resistor R4
the IC drops to almost nothing, and the 5
remaining charge bleeds off through
0
R4. Fig. 11 shows the associated fault
-5
currents through these paths.
-10
In the simulation, approximately
40% (a little over 1.1 A) of C1’s -15
V R 4 - (V )
discharge current flows to the negative -20
supply rail through ESD protection -25
diode D12. The other 60% (almost 1.8 -30
A) flows through reverse Vbe protec- -35 22 mF
tion diode Q3. Only a small portion of 47 mF
-40 100 mF
Q3’s current flows through ESD
-45 220 mF
protection diode D14; the majority of 470 mF
current flows through gain resistor RG -50
0 5 10 15 20
(set to 10 Ω in this simulation). The T ime - (ms )
RG current flows through the base-
Fig. 10. Current
emitter junction of Q2 and capacitor Diff
ffe rential Fault
Differential through input capacitor
C2, and then to ground through SW2 0 C1
(which is permanently closed). No
current flows through protection -0.5
-1.5
formed by the reverse Vbe protection
diode at one input, the gain resistor, the -2
other input transistor, and the other
-2.5 22 mF
(previously discharged and grounded) 47 mF
input capacitor. The loop current is 100 mF
-3
primarily limited by the gain resistor, 220 mF
RG. As a result, the peak differential- 470 mF
-3.5
0 5 10 15 20
fault current is a function of the gain T ime - (ms )
setting of the microphone preampli- ➥
2000
mA
1500
1000
500
0
C1 Q3 RG Q2 C2 V - P in D12 D14
P eak C urrent, mA 2988 1784 1629 1629 1617 1319 1166 155
AVAILABLE
11 Sc
ree
n
12
14
13
17 15 Ad
Lis va
ten nc
ed
are ing 16 M ult
a 1 Scr ich
ee an
Up 18 19 2 n ne
l
pe 3 Au
r la 6 dio
1 yer 4 Sy
Sc 5 ste
2 ree ms
n 8 Lis
3 te n 11 S cr
Int
6 4 a re in g 7
ee
n
a 12
5 14
8 1 0
Lis Mi d 9
ro
ten dl e 13
lay 15
are ing 7 er 17 L is
ten
a
du are ing 16
1
Mid 10 9 2
Scr
ee Fu a
dle n ll
lay 3 Up 18 19
pe
ct
20 er 6 r la
S 4
S1 cree 5
yer
21 n
ion
8
T Lis 10
Mi d 9
ten dl e 13
are ing lay
er 1 7 L is
15
Fig L hi a ten
ure A
2 E se ows eanr are ing 16
20 : LSos pe tio tla c h 1 Scr a
Co 0 ued r n oyloer 2 ee 5.1
ndi o appr m4ualtnicriesspeackept s: hi gy—
n 18
3 Up 19
tionf oa d ha on erusa sto w 6 pe
r la
inc louLocuhe ednnelsp al rara ric ith 4 yer
the lud dsp ds.pPe ited aautdiai dngem n al pa 5
Fula i e a
a o
l q ere p
l n ge ng uaske rkt er by sosys ua ntvi f ers
8 Lis 11
na d q ner app edr- 2 cs D untemlity oew 2s 2 di te n S cr
s l p ua ati ro ba on ur d t e , s.2 vid a re in g 7
ee
n
5th.1yste roc lity e 1o-n 19 ach sed cent aInnd echn valu yste ed a
14
12
e ms ess va of es m rat stBru iq ati ms int 1 0
the pr a ing lu ap to A ull es ecgtio ue on a o Mi d 9
i r
in pa ma e2n - al atio pro th(e iclhoudon ulnt s. P —i nd hre tl a t dl e 13
F p x a t t la 1 1
in 3cahs per ry t ot 4i , 8g, or n. R ria ccrlud nnspehae , d art s P ech e b yer 7 L is 5
wh clud epar s co hem nclu 9 ithm eco te p leaatioingellos kfeurnsd eals 1, p art nol road ten
ing 1
de ich ed ate 2 -nce es ded s, c ITrdUin erc yern) o pwaetir am wi ub1l 2 o Sgycr o a rea 6
m sM criid thi to vol 4,r6ne-d of t he Roed -gRa eptu f a al ent th b ishe 2f an , een b
a
F3
ch 18
m erci bed s fie dem ume .1T9 his Srea, e ciongm, nBdSm77al c spa audi l pr ina d in 3 e
les div Up 19
mTaubltlic al a hav ld h1 ons . So hese volumxecae s m stoen ix 5.1ue tia o 6inc ura pe
o e p p r d a i s l s i l te 4 s a ide r la
w vie h3an pli e n as- 10tra me losu frm one t “aFu age tionng , re sou yst ple so ning 5ccur be
d yer
is ell d su nelRepcati ot a pass te th his udbjsepce taanl msi thlel”, ,baun tech pro n8df ems L s e un a ate tw
t d
in on ocu rrou spa roodnus ll fo ed, e e M ori tsaaker arddlye a td t niq uc eld , is te n s xte fie ea, ly1, en S cr d i d r 1 e
ab the cha me nd tial c. tTiohn und eve voliddclael re ls arisee ppl rans ues, tion , a re sin gtere nsio7 ld re by an a app e en
B ly b firs rting nted sou aud esecondth n i utio rleavyie ikel 3ucnha y di mis sig , a tr
o uc ph
o n
o p1re me co 1o2a
r
pe ecau e sa t de the by nd io a hel iteiior nws f the nar ewr sonlyyto void rect sion - M 10 f p t ac ony f t 4sen ans usti che
tec riod se id t cad tra rese sys s we p to ayfor sys y p and be abl ly to i dd illu hys ou9 . T he p tati of cal s th 1 3
s s
o t
we hni of f t hat e of il th arc em ll a do intoxpteerm oce pa cov y Fini e r
a
l t
eenlaey on cal tica h1e7 l syc L on, 15an even at ai
i i
th re que 197 his mu th Aat h pa s, w s pr cum m im s ae ss per ere gure e udi ’sre tha acc wa tte oa tennd ana t o m to r e l a h is a
an e A the s, fo 0 t mo ltich e 21 EleSad per hic ovi en ains ndts thro s ar
n d 3: Rnt o il ars t g ura ve r, w arceouingapp lyti1 6r so re
on d p ES topi r ex o 20 st o ann st c s1t1o6ts. T h a ding t the trea meth ugh e
ph epw lu to M en cy fro h a s ro ca un co
ch ysic raoydsu: sion coniderat , de nts 1 ile s tic p ach l ph dfie nstr
ste ly rpo in t c of am 00. f th el s entu whh C
u he re an or m c od an al ct . e p a8 o r1 e y l u
a h
to reo sm ses e 1 a n ple, So- e ke pati ry, ere nmvpe ot insi igin om
o e n s p rag ne sc theion Tche vinc s suUpepned nd s met inc9 s th sica d w ct, m
c l i s on s e f r o im ip at l ith o
na the pho al , st 97 ow are cal y p al a wh the hanstio art ght s o p - o m s e o de f l
pa u
t i m mpr atic pati ntist und iatpiop nshim icie riymer ndf es m es o are appr in a re o
l
mi l B birt nic l am arts 0s. T ou not led q ape udi en i fiel is, hn, iB cu int f
e o a e om en al ’s fie roa fo or nt ari iel a f ba oa l r
cro ell h o tec ou wh he t-o co ua rs o h t c d Pi a ow arrlil n
f s
l
p
pp s
r m ise gin aud wa ld ch r ehxepr per ly ds w nag two sed ch is-
M ph La f m ni nt h e - v d a a o g e y , o
Ge p erc ac o r ad eer io, o rec s c eorfim cep po ith ing -ch on to y e u
ult on bs u qu o re q curr prin ered raph re t s co uld locea 5veor,
ich e a e ltic es f o ua en t a in on ake me rea ted f 12 srmhailn ep he e op ’s
s w f c on an theent tu n so to an an
an nd xpe ha als ve dra t an nth m ic n s o osy tua , a s i m ted con hil on str be v s al a s m re ne
ne lo ri nn o rla ph th ol uc sy fro of a on- wi for o,ph20lly. nd e i l a in cep e th ceiv uctio dist alidi cues tere e de con l
l a ud m el co p. on ol og h
e st e m ge e th th. y a04It i ith r i the tio e ing n ing ty a op gr -
ud sp nt sp nta T ic og y det m th . n t M do Ms ap er t i e 19 n t ste o a u of t ho ee
io ea s ati in he s l y, pu ail s a e th ce s he os pt eyr a s 3 ha re f pp ish th the ni
sy ker on al a s m A eft for blis . T nd
ste c st u at ES of a he he a c e lou oun aim t rec ed i h8ap–1p1pro t h a 0s. T t gre oph “how roac ed i e sp lis c
ms han ere dio eri f l y t re d df of or n s s m ac n h w on h e n d ati -
a , l d h a sp ie d y h d e o ic o s t i a
an nel oph , in al th nth exh inten by ai etica tive eak ld w acc ing stem ore can i f f e re a ut o ap do are ffer l
d s.
tec on clu at olo ibit ts go m a lly and er a he urat eng de a qu be r e n re, f a pro ” m th -
hn y din is gy ing
us g rel po od t acc and art nd n the ely inee sign esti ana c e s how com ach ulti e
olo in the ev on c sit in ur pe isti mic y rep rs, , r on lyz b eve m is -
g a io p a r c m r e o e e
gy m ori ant fo l ap n, ract te a cep ma roph ake ese for cor f th ed p t w r, so rcia a
ma
y
ul gi
tip - fie rwa plic whic ice, cous tual nne one rec ntin exam ding e pa hys e e n me l
ap le ld rd ati h o ti ly. r c or g p te ra ica t h -
pe l As of a , an ons. mak r ma cal r On to fa han ding som le, d chn digm lly e
ar a p d S e y e t s n s e o i o
to er rge a ru plic the om s th no pres he o hion els a of no no que, and r
m s, as liste le, atio mo ewh em t w ent the a m vai mu tion t se and
he ake is t ning an a n. st a ere hard ork ation r ha ix lable sic; al re t ou
nc su he a cc pp in t in s nd th t the fe t
e t ch c rea ur
he sy ase r ate ro the o u mo do , ap at w o th y u r-
pr
su ste w equ re iat mi se fo re t not pro ork em se
cc m ith ire co e dd r m han alw ach s a in
es s n ap le
s o im wa s a stru pr of ain on ays es th es-
v
f s pra ef er cti v oa al str e l so a
ter ct iel y l on
eo ica d s arg of
ph l f yn e
ch l th ea ist un t
de is m p eni d
pe lie ra ng
This anthology—with papers
on or th nu a so nd s a ct
ic m
red os s.
es m un
i
uc t ap Thi r of field
b e d
s o w i-
n ay
the
divided into three broad sections:
tio p s h lo o
ns lica as ud ve
tha tio te pe r a
t r ns nde ak
s historical reviews, systems and tech-
ely to d -
pa da to
rti te,
all
nology, perceptual and quality evalua-
y
tion—is Part 2 of an AES series on
spatial sound techniques. Part 1, pub-
lished in 2004, deals with binaural
approaches. Part 2 concentrates on
the fundamental principles of loud-
speaker-based multichannel spatial
audio systems, including approaches
to the creation of a spatial soundfield,
the generation of appropriate percep-
tual cues, reproduction, and quality
evaluation.
V peak mA peak
Common-Mode 48 48 1925 1925 1166 1166 720 720 0 0 720 720 0 3770
Single-Ended 48 23 2308 0 1166 241 514 300 0 289 1103 0 589 2220
fier when the fault occurs, and will be Due to their physical design, the width of the pin and socket, etc.) will
greatest at the highest gain. exposed contacts of male TRS plugs vary, so the order in which the three
are more likely to accidentally touch signal pins make contact is unpre-
FAULT SUMMARY metal objects than the partially dictable. Moreover, each individual
Table 1 provides a summary of the sim- shrouded XLR contacts. Thus, TRS contact may be intermittent or
ulated transient voltages and currents in plugs are especially prone to causing “bounce” as the connection is made.
each fault mode. The fault currents that phantom faults. However, XLR Thus, any of the described phantom
flow through the protection diodes and connectors are not immune to faults. faults can occur when devices are
input transistors are extremely high. In When an XLR connector is plugged plugged together.
the authors’ experience, no integrated into a signal output, contact is made to Some equipment provides “pseudo-
devices are capable of withstanding the outside shield first due to the phys- balanced” or unbalanced outputs via
such high currents without damage. ical design of the connector. Contact is XLR connectors. When a connection is
Since phantom power faults are essen- then made to each of the three signal made to the undriven signal leg, a low
tially inevitable, the most appropriate pins. The angle of insertion and condi- or near-zero impedance connection is
solution is to provide an external pro- tion of the connectors (oxidation, established between the microphone
tection network to divert these currents exact length of the pin and the socket, preamp input and ground, causing a ➥
safely around the IC.
+48V
+48V
R1 R2
6k81 6k81 R1 R2
6k81 6k81
C1 4x 1N4148
4 C1
M ic I N- I N-
M ic I N- I N-
V+
D1 D2
C2
D3 D4 V- ZENER
M ic I N+ I N+
C2 D1 D2
M ic I N+ I N+
R4 R5
D3 D4 R bias R bias
R4 R5
R bias R bias
Fig. 12. 1N4148 bridge clamping the input signal to the rails Fig. 13. Zeners back-to-back clamping the input signal to
(not recommended) ground (not recommended)
Bidirectional TVS
V-
Fig. 14 shows a protection scheme using C2
RS D3 D4
12–18 V bidirectional Transient Voltage M ic I N+
47R min
I N+
C1
RS 4x 1N4004G
1N4004GP
V4+ P 2x 1N4148 D5 1N4004 Bridge + Series Resistors
M ic I N- I N-
10R min RG
D1 D2
V-
D6
R G1 Fig. 17 shows a protection scheme
D3 D4
V+
D5 R G2
using 1N4004GP1 diodes in a bridge
C2
M ic I N+
RS RG I N+
configuration, with series resistors
10R min V-
D3 D4 D6 placed in front of the bridge to limit
R4 R5
C2
RS 1k21 1k21
R G2 the fault current. This circuit attempts
M ic I N+
10R min
I N+ to clamp the input signal to the power
R4 R5
rails. The 1N4004 diodes are inexpen-
1k21 1k21 sive and readily available.
The authors’ experience is that this
circuit will protect the part against all
three fault modes, provided the series
resistors are large enough. The 47 Ω
Fig. 18. Proposed protection scheme that incorporates two 1N4148 to steer the
currents away from the base-emitter protection diodes resistors shown are effective, but they
compromise noise performance at high
to a low-enough value that the zeners Schottky Diodes + Series gains. Previously, THAT Corporation
survive the fault. However, the addi- Resistors had suggested this scheme, with only
tional 94 Ω in series with the source Fig. 16 shows a protection scheme that 10 Ω series resistors. The authors have
compromises the noise performance of uses Schottky diodes in a bridge config- found that without more current limiting
high-performance IC preamps [1]. uration, clamping the input signal to the (extra resistance) this scheme will not
A compromise might be sought by rails, with two resistors placed in front protect the IC from differential faults.
reducing the series impedance and of the bridge to limit the fault current.
increasing the wattage of the zener Bridge + Series Resistors + B-E
The authors have found that the mini-
diodes. However, in the authors’ expe- diodes
mum value of these resistors, required to
rience, if the zeners are large enough to protect against differential faults, is Table 1 reveals that very large cur-
handle the fault currents, they increase 47 Ω. Like the circuit of Fig. 15, this rents flow through the input transis-
distortion. Of course, if they are too compromises noise performance [1]. tors, RG, and internal reverse V be
small, they become sacrificial. Moreover, Schottky diodes used in diodes, Q3 and Q4 (Fig. 2), during ➥
3000
2500
mA
2000
1500
1000
500
0
C1 C2 D12 D22 D14 D24 Q3 Q4 V - P in
T ype A , No P rotec tion 1925 1925 1166 1166 720 720 720 720 3770
T ype B , S eries R es is tors 1198 1198 730 730 438 438 438 438 2337
T ype C , B P lus B ridge Diodes 3198 3198 11 11 0 0 0 0 21
T ype D, C P lus R ev-V be Diodes 3198 3198 11 11 0 0 0 0 21
Fig. 19. Current distribution during the common-mode fault for the circuits type A, B, C, and D
206 J. Audio Eng. Soc., Vol. 58, No. 3, 2010 March
The 48 -Volt Phantom Menace Returns
differential faults. Differential faults resistors at RS may be reduced to NEW PROTECTION SCHEME
are most destructive to Q3 and Q4. 10 Ω each, so noise performance is SIMULATION
Fig. 18 shows a protection scheme not unduly compromised. The authors After analyzing the three fault mecha-
that includes a pair of 1N4148 diodes have found that this protection nisms, and considering the various pro-
to steer currents around Q3 and Q4 scheme is effective against all three tection schemes presented above, the
during faults. With the addition of fault modes, and recommend it in authors focused on the circuit shown in
these diodes, the current-limiting preference to others. Fig. 18. This circuit was tested with ➥
2500
2000
mA
1500
1000
500
0
C1 D12 D22 D14 D24 Q2 Q3 RG V - P in
T ype A , No P rotec tion 2308 1166 241 514 300 289 1103 589 2220
T ype B , S eries R es is tors 1332 682 120 290 170 163 622 332 1261
T ype C , B P lus B ridge Diodes 3199 10 0 0 0 34 34 34 10
T ype D, C P lus R ev-V be Diodes 3199 10 0 0 0 36 2 36 10
Fig. 20. Current distribution during the single-ended fault for the circuits type A, B, C, and D
2500
2000
mA
1500
1000
500
0
C1 C2 D12 D14 Q2 Q3 RG V - P in
T ype A , No P rotec tion 2988 1617 1166 155 1629 1784 1629 1319
T ype B , S eries R es is tors 1519 770 614 85 794 878 794 698
T ype C , B P lus B ridge Diodes 3201 353 10 0 369 369 369 10
T ype D, C P lus R ev-V be Diodes 3202 610 9 0 630 19 630 9
Fig. 21. Current distribution during the differential fault for the circuits type A, B, C, and D
J. Audio Eng. Soc., Vol. 58, No. 3, 2010 March 207
The 48 -Volt Phantom Menace Returns
40
30
20
10
V 0
-10
-20
-30
-40
0 5 10 15 20 25 30 35 40 45 50 55 60 65
ms
Fig. 22. During the fault, in the circuit of Fig. 21, the negative rail (blue) follows the voltage across R4 (black)
a variety of microphone preamplifier away from the IC (and into the “stiff” differential fault is the most dangerous
ICs [2–6], and found to be both robust negative supply rail) by reducing the type due to the high currents flowing
in protecting the ICs and “transparent” impedance seen by the input capacitors through the input transistors and base-
in audio performance. while it clamps. This is why the emitter protection diodes. These high
To understand why this circuit discharge current is larger. However, currents flow through the resistor gain
worked so well, the authors ran simu- by steering currents away from the IC, RG, providing a convenient spot for
lations of all three fault modes with the diode bridge also significantly noninvasive current measurement and
different elements of the protection reduces the damaging currents flowing evaluation of protection effectiveness.
scheme shown in Fig. 18. The results in the IC. Simulations, although helpful, can
are shown in Figs. 19, 20, and 21. The external base-emitter diodes D5 miss phenomenon witnessed in the real
Each Figure shows a simulation for and D6 are in parallel with internal world. For example, the data in Figs. 19
four different circuits, as follows: diodes Q3 and Q4. They serve to through Fig. 21 assumes the bridge
A) circuit of Fig. 2 reduce the current flowing in the inter- diodes clamp the IC inputs exactly 0.6 V
B) circuit of Fig. 2, with two 10 Ω nal protection diodes. D5 and D6 allow volts from the ±15 V power supplies
input series resistors (RS) the input series resistors to be reduced (which are themselves modeled as
C) circuit in B, plus the diode bridge from 47 Ω to 10 Ω while still protect- perfect batteries). In practice, the power
(D1–D4) to the power supply rails ing the part. This is important to main- supplies are not batteries, and the clamp-
D) circuit in C, plus base-emitter taining a low noise floor. ing is not ideal.
diodes (D5 and D6). This is the cir- The equivalent input noise of the In particular, simulator power
cuit shown in Fig. 18. microphone preamplifier IC in ques- supplies have low Thevenin impedance
Other circuit parameters are: R4 = tion with 60 dB gain and its input for both sinking and sourcing current. In
R5 = 1.21 kΩ, RG = 10 Ω, and V+ / shorted, is 1 nV/√Hz. With 10 Ω input the real world, most positive-power-
V– = ±15 V. resistors, the input noise is increased to supply regulators are only able to source
The addition of 10 Ω input series 1.15 nV/√Hz, a compromise of 1.2 dB. current, and do a poor job of sinking it.
resistors reduces fault currents, but With 47 Ω input resistors, the input The negative-supply regulator will likely
20 Ω (total) is not sufficient to protect noise goes up to 1.6 nV/√Hz, a be a good sink but a poor source. ➥
the IC from differential faults. The compromise of 4.1 dB. Using
experiments suggest that all preampli- typical 150 Ω microphone
fier ICs will fail without some input source impedance and 10 Ω
series resistance. As well, although resistors, the dynamic range is
beyond the scope of this paper, the degraded by 0.4 dB. For the V in V in -15V -15V
authors suspect that discrete designs same source impedance and
C1 GND C2
will be similarly challenged by differ- 47 Ω resistors, the dynamic
ential faults. range is compromised by 1.6dB.
Adding the diode bridge to the rails Therefore, by using 10 Ω resis-
appears (in the simulations) to provide tors instead of 47 Ω resistors,
signal clamping and current steering, the dynamic range is increased Fig. 23. Simple representation of a bench
but it also increases the fault current. by 1.2 dB. power supply or three-terminal regulator.
The bridge steers voltages and currents As shown in Figs. 19–21, the Typically, C1>>C2.
208 J. Audio Eng. Soc., Vol. 58, No. 3, 2010 March
The 48 -Volt Phantom Menace Returns
+48V
V+
R1 R2 D7
6k81 6k81 1N4148
4
C3
C1 4x 1N4004GP
1N4004G
4 P 2x 1N4148
4
RS
M ic I N- I N-
10R min 100n
47uF R G1
V+
D1 D2
D5
RG OUTT
OU OU T
R EF
D3 D4 D6
V-
R G2
C2
RS C4
M ic I N+ I N+
10R min
47uF * 18V
8 TV S optional
opt
p ional 100n
TV S
* * R4 R5 D8
C5 C6 R bias R bias 1N4148
4
47uF min D9 D10 47uF
min V-
Fig. 24. New protection scheme with diode-isolated rails (D7, D8) and bypass capacitors (C5, C6) to ground
Often, if a power-supply output is driven To see how this circuit actually negative than the regulator input volt-
past the regulated voltage by a phantom- behaves, the authors observed various age get steered to the filter capacitance
related fault, it has no means to correct waveforms in the circuit. A digital at the regulator’s input. With the linear
the error and will go out of regulation. oscilloscope was used to capture the supply used in the tests, the unregu-
Consequently, the power supply will not waveforms. lated supply voltage will determine the
instantly absorb the capacitor’s transient Fig. 22 shows the voltage across the clamp voltage.
discharge. In this case, the negative rail bias resistor R4 (black trace) and the Other loads connected to the regula-
“moves,” and the clamping voltage is no voltage at the V– rail (blue trace) tor can also absorb the transient.
longer V–. Under these conditions the during a phantom power fault. Uncontrolled, a phantom fault applied
currents flowing through the IC can be Initially, the input voltage is zero and to the circuit of Fig. 18 can corrupt the
significantly different from those V– is –15 V. When the fault occurs, entire V– supply. The authors expected
predicted by the simulations. the input voltage is not clamped at V–, that this could damage the IC preamp,
as predicted by the simulation, but as well as other circuitry, unless it can
REAL-WORLD TESTING drops to approximately –35 V, driven tolerate the transient increase in supply
Given that the SPICE simulations pre- there by the discharge current in C1. voltage.
sented above do not represent the real Substantially the same behavior mani- One means of clamping the negative
world of regulated power-supply fests regardless of the fault mode. excursion is to add a bypass capacitor
behavior, the authors measured the rec- Initially it was suspected that the from the IC’s V– pin to ground (the
ommended circuit shown in Fig. 18 on voltage to which the supply is driven same might be used for V+ as well) to
the bench. The results presented earlier would depend on the capacitance on form a capacitive voltage divider. (See
show that the differential fault is the the negative supply. But, after some C5 and C6 in Fig. 24.) During a fault, a
worst-case scenario, so the authors investigation, the authors observed that portion of the input capacitor’s charge
concentrated on that one. As well, the the circuit was clamping to the bench is transferred to the bypass capacitor
current flowing through the IC’s front- power supply’s regulator input voltage, through the bridge-protection diode
end stage is essentially equal to the which has a much larger filter capaci- (D2 or D4). Bypass capacitors give the
current flowing through the gain resis- tance than its output. energy in the input capacitors (C1 or
tor RG and the internal base-emitter Fig. 23 shows a typical three-termi- C2) a less destructive path to ground.
protection diodes. Thus, the differen- nal regulator with reverse-polarity But, in order to limit the negative-
tial voltage across RG permits indirect protection. The regulator output, typi- supply transient to a safe level
measurement of internal junction cur- cally an emitter follower, cannot (perhaps –18 V), a large capacitor is
rents. Limiting transient currents source the current required to clamp required. The charge distributes based
through these junctions is the key to transients. Regulator reverse-polarity on the size of the bypass capacitor rela-
protecting against differential faults. protection assures that transients more tive to the input-coupling capacitor. ➥
J. Audio Eng. Soc., Vol. 58, No. 3, 2010 March 209
AES is now a scitopia.org partner
search
60
40
20
V 0
-20
-40
-60
-80
0 5 10 15 20 25 30 35 40 45 50 55 60 65
ms
Fig. 25. Isolation diodes without capacitors C5 and C6. Input voltage (black) and isolated negative rail (blue).
40
30
20
10
V 0
-10
-20
-30
-40
0 1 2 3 4 5 6 7 8 9 10 11 12 13
ms
Fig. 26. Voltage at IN+ (black) and V– (blue), for the circuit of Fig. 24 as is.
40
30
20
10
V 0
-10
-20
-30
-40
0 1 2 3 4 5 6 7 8 9 10 11 12 13
ms
Fig. 27. Voltage at IN+ (black) and V– (blue), for the circuit of Fig. 24 including 18 V TVS diode at D10; C6 = 47 µF.
To limit the voltage increase on the found, somewhat surprisingly, that the circuitry, and so should be at least
supply-bypass capacitor to, say, 3 V, IC preamplifiers were not damaged by isolated, if not eliminated altogether.
the bypass capacitor must be many the brief negative supply transient Accordingly, the authors sought a
times larger than the input-coupling to –35 V shown in Fig. 22. None- means to isolate them from the –15 V
capacitor. This is often impractical. theless, a large disturbance on the supply and other circuitry.
However, on the bench, the authors negative supply might damage other In Fig. 24, diodes D7 and D8 ➥
J. Audio Eng. Soc., Vol. 58, No. 3, 2010 March 211
The 48 -Volt Phantom Menace Returns
40
30
20
10
V 0
-10
-20
-30
-40
0 1 2 3 4 5 6 7 8 9 10 11 12 13
ms
Fig. 28. Differential fault. Voltages across the gain resistor (black) and the isolated V– voltages (blue), for 0.1 µF and 47 µF
bypass capacitors.
isolate the IC’s V+ and V– pins from mend that the input-coupling capaci- Fig. 14, which is not recommended. It
the power supply rails supplying the tors be no larger than 100 µF. As is also possible to share elements from
rest of the system. D7 and D8 are shown in Fig. 25, unless a sufficiently the circuit of Fig. 24 among several
forward biased during normal opera- large bypass capacitor is used, during preamplifiers, isolating each individual
tion, but reversed biased when a phan- fault events the negative rail can be input with its own bridge (and provid-
tom power fault occurs or an external driven above the maximum rating for ing each input with individual D5 and
voltage source is applied to the typical IC preamplifiers. While the D6), but sharing the supply isolation
preamplifier inputs (+48 V discharg- authors’ experience is that IC preamps and bypass components. The authors
ing from the output capacitors of an seem to survive this abuse, this does consider this one of many areas for
AC-coupled line output previously raise concerns over reliability. One further investigation.
connected to a phantom power source, way to address this without requiring It was found that bypass-capacitor
for example). The protection bridge such large capacitance is to place a values should be at least equal to the
(D1–D4) terminates on the IC side of TVS diode across the rail as shown at value of Cin (C1 or C2, Fig. 24), and
the supply diodes D7 and D8, contain- D10 in Fig. 24. Fig. 27 shows the ideally greater than two times Cin, to
ing any overvoltage from faults to same voltages if an 18 V TVS is provide protection from common
within the preamp system itself. A included at D10. mode faults where the effective value
modest-sized bypass capacitor (C5 Fig. 28 shows the voltages across of Cin is doubled.
and C6, shown as 47 µF) serves to the 10 Ω gain resistor (black traces) To date, most of this work’s experi-
absorb some of the charge during the and the isolated V– pin (blue traces) ments and simulations have been
fault, limiting the voltage excursion at using 0.1 µF and 47 µF bypass capaci- focused on faults based on shorting the
the supply pins. tors at C6. With a 0.1 µF bypass inputs. These faults share the common
Fig. 25 shows the IC input voltage capacitor, the peak current flowing feature of drawing the IC input pins
at Rbias (black trace) and the isolated through RG is approximately 1.2 A. negative, with consequences as
V– pin voltage (blue trace), with With a 47 µF bypass capacitor, the discussed in this article. While this is
capacitor C6 = 0.1 µF, during a phan- RG current drops significantly to clearly a common cause of faults,
tom power fault on the Mic IN+. The approximately 700 mA. This current another class consists of those where a
voltage at the IC’s IN+ pin drops to can be reduced even further if larger source of phantom power is mistakenly
approximately –48 V. Without a capacitors and/or TVS are used. The connected to the input of a microphone
means to absorb the input capacitor’s use of bridge diodes, TVS devices, preamplifier. The authors have not yet
charge, V– tracks the input voltage. and capacitors has been explored by studied this class of fault sufficiently to
Fig. 26 shows the voltage at IN+ and others to protect sensitive low-voltage reach any firm recommendations.
V– for C6 = 47 µF. Note, of course, programmable gain amplifiers against However, at a minimum, similar protec-
that if C1 and C2 are scaled, the tran- phantom faults [7]. tion against excessive positive input
sient peak will scale directly unless C6 TVS devices, isolated in this way by excursions and on the positive power
is also scaled similarly. Partly for this their use on the power-supply lines for supply seems appropriate to guard
reason, and considering low-frequency the preamplifier, do not contribute against this category. For this reason, Fig.
response versus typical preamplifier signal distortion. This makes them 24 includes components symmetrically
input impedances, the authors recom- applicable, in contrast to the circuit of arranged toward positive excursions.
212 J. Audio Eng. Soc., Vol. 58, No. 3, 2010 March
CONCLUSIONS AND system. Amsterdam, The Netherlands
SUGGESTIONS The authors have tested the (2001 May).
This work presents three phantom proposed protection scheme exhaus- [2] THAT1500 Series datasheet, Doc.
power-related fault mechanisms: tively, but understand that there are 600031 Rev.05, www.thatcorp.
common-mode, open-circuit single- numerous conditions and applications com/datashts/1500data.pdf.
ended, and differential, and pro- that were not covered in this work. [3] TI INA217 datasheet, SBOS247B,
poses a new protection scheme that Designers should consider these limi- 06/02 Rev. 02/05, http://focus.ti.
works under a variety of conditions tations and modify circuitry as they com/lit/ds/symlink/ina217.pdf.
with minimal impact on noise deem necessary. [4] TI INA163 datasheet, SBOS177D,
performance. 11/00 Rev. 05/05, http://focus.ti.
Similar previously published works ACKNOWLEDGEMENTS com/lit/ds/symlink/ina163.pdf.
have focused on common-mode faults, The authors thank THAT Corpora- [5] AD SSM2019 datasheet, Rev.0,
but the authors’ observation is that tion for encouraging and supporting 02/03, www.analog.com/static/
differential faults are the most danger- this work. Also, the authors thank imported-files/data_sheets/
ous to an IC. Les Tyler, Bob Moses, and Fred SSM2019.pdf.
Many of the protection schemes in Floru for their discussions and sug- [6] W. Jung and A. Garcia, “A Low
use today are ineffective at protecting gestions, and David Lail for editing Noise Microphone Preamp with a
the active elements against phantom the figures. Phantom Power Option,” Analog
power faults. Devices Application Note,
The proposed new protection REFERENCES AN-242, 11/93.
scheme is more robust than previous [1 ] G . H e b e r t a n d F . T h o m a s , [7] Gries et al., “Protection Circuit for
recommendations, steers destructive “The 48-Volt Phantom Menace,” an Input Stage, and Respective
currents away from the IC during paper 5335 presented at the Circuit Arrangement,” Patent
fault conditions, and isolates the 110th Convention of the Application No.: US 2008/
phantom menace from the rest of the Audio Engineering Society, 0007884 A1.