Home │ Audio
Home Page |
Copyright © 2008, 2009, 2010, 2011, 2014, 2015
Updated November 28, 2015. See document history at
end for details.
SPICE verified where indicated.
Maximize your browser to allow room for the equations. Verified on Internet Explorer and Firefox browsers.
The s variable will
represent seconds or the Laplace s-domain variable according to context.
Typical RIAA Equalization Bode Plot |
The design cycle involves initially calculating capacitances
from resistors appropriate for the desired gain and noise, rechoosing
the nearest appropriate standard capacitor values in the correct ratio,
then recalculating the resistances from which the nearest standard
values are chosen. This is because resistors are available in
tighter tolerances than are capacitors.
I presume for now that you know how to set the gain for these circuits and that calculating the RC networks are the difficult calculation.
Op amps are used for ease of illustration only. There is no intent to disparage discrete circuit designs.
AV-1kHz
AV<50Hz |
= |
0.10103 | (-19.911018dB) |
AV<50Hz
AV-1kHz |
= |
9.89808 | (19.911018dB) |
This calculation applies to an RIAA impedance network that converts unequalized signal current into equalized output voltage. This includes inverting and current input stages.
TP1 = R1C1 = 3180µs
TP2 = R2C2 = 75µs
TZ = (R1||R2)(C1+C2)
=
318µs
TZ is met when:
R1/R2 ratio = 11.78
and when C1/C2 ratio = 3.6
Inverting series-parallel version verified by SPICE: Model.
(1) |
Z = R1||(1/sC1) + R2||(1/sC2) = | 1
1/R1 + sC1 |
+ |
1
1/R2 + sC2 |
= |
R1
sR1C1+1 |
+ |
R2
sR2C2+1 |
(2) |
Z = | sR1R2C2
+ R1
+ sR1R2C1
+ R2
(sR1C1 + 1)(sR2C2 + 1) |
Time constant of reduced term is seen in (sT + 1).
Poles are clear, solve for zero.
(3) |
sR1R2C2 + R1 + sR1R2C1 + R2 = s(R1R2C2 + R1R2C1) + (R1 + R2) |
(4) |
TZ = | R1R2C2
+ R1R2C1
R1 + R2 |
= |
R1(75µs) + R2(3180µs)
R1 + R2 |
= 318µs |
(5) |
R1(75µs) + R2(3180µs) = (R1+R2)318µs |
(6) |
R1(75µs-318µs) + R2(3180µs-318µs) = 0 |
(7) |
R1
R2 |
= |
3180µs – 318µs
318µs - 75µs |
= 11.777... = 11.7 |
(8) |
C1
C2 |
= |
TP1/R1
TP2/R2 |
= |
3180µs/11.7R2
75µs/R2 |
= 3.6 |
TP1 = R1C1 = 3180µs
TP2 = R2C2 = 75µs
TZ1 = 318µs
TZ2 = 3.18µs
R1, R2, and R3 in ratio 217.173913 :
17.67514356
: 1 will yield correct
time constants for zeros if C1 and C2 are chosen
to yield correct poles.
R3b = |
(235.8490566)R3 AV<50Hz |
= (0.2382778)R3
for RIAA
standard 40dB gain at 1kHz |
SPICE model of non-inverting series-parallel riaa circuit with extra
zero verifying correct adjustments for gain.
SPICE Model
See Low-noise
Hybrid
Op-amp/Discrete
Phono
Preamp for a complete
design example of this type.
(1) |
Z = R1||(1/sC1) + R2||(1/sC2) + R3 = | R1
sR1C1 + 1 |
+ |
R2
sR2C2 + 1 |
+ R3 |
(2) |
Z = | sR1R2C2
+
R1 +
sR2R1C1 + R2 + (sR1C1
+
1)(sR2C2 + 1)R3
(sR1C1 + 1)(sR2C2 + 1) |
(3) |
Z = | sR1R2C2
+
R1 +
sR2R1C1 + R2 + (s2R1C1R2C2
+
sR1C1 + sR2C2 + 1)R3
(sR1C1 + 1)(sR2C2 + 1) |
(4) |
Z = | sR1R2C2
+
R1 +
sR2R1C1 + R2 + s2R1C1R2C2R3
+ s(R1C1R3 + R2C2R3)
+
R3
(sR1C1 + 1)(sR2C2 + 1) |
(5) |
Z = | s2R1C1R2C2R3
+ s(R1R2C2 + R2R1C1
+ R1C1R3
+ R2C2R3)
+ (R1 + R2 + R3)
(sR1C1 + 1)(sR2C2 + 1) |
Time constant of reduced term is seen in (sT + 1).
Poles are clear, solve for zeros.
(6) |
s2R1C1R2C2R3 + s(R1R2C2 + R2R1C1 + R1C1R3 + R2C2R3) + (R1 + R2 + R3) = 0 |
(7) |
s2(3180µs)(75µs)R3 + s(R1(75µs) + R2(3180µs) + (3180µs)R3 + (75µs)R3) + (R1 + R2 + R3) = 0 |
(8) |
s2(238.5n(S2))R3 + s((R1 + R3)(75µs) + (R2 + R3)(3180µs)) + (R1 + R2 + R3) = 0 |
(9) |
s2 | R3(238.5n(S2))
R1 + R2 + R3 |
+ s | (R1
+ R3)75µs
+
(R2 + R3)3180µs
R1 + R2 + R3 |
+ 1 = 0 |
(10) |
= s2TZ1TZ2 + s(TZ1 + TZ2) + 1 = 0 |
(11) |
TZ1TZ2 = | R3(238.5n(s2))
R1 + R2 + R3 |
= 1.01124n(s2) |
(12) |
R3
R1 + R2 + R3 |
= 4.24m |
(13) |
R3 = 4.24m(R1 + R2 + R3) |
(14) |
R1 + R2 + R3 = 235.8490566 R3 |
(15) |
R1 + R2 = 234.8490566 R3 |
(16) |
TZ1 + TZ2 = | (R1 + R3)75µs
+
(R2
+ R3)3180µs
R1 + R2 + R3 |
= 321.18µs |
(17) |
(R1 + R3)75µs + (R2 + R3)3180µs = (R1 + R2 + R3)321.18µs |
(18) |
R1(75µs - 321.18µs) + R2(3180µs - 321.18µs) + R3(75µs + 3180µs - 321.18µs) = 0 |
(19) R1(-246.18µs) + R2(2858.82µs) + R3(2933.82µs) = 0
(19) |
R1(-246.18µs) + R2(2858.82µs) + R3(2933.82µs) = 0 |
The two equations (15) and (19) in three variables suggest R1, R2, R3 variable but in fixed ratio as intuition would suggest.
Choose R3 = 1 to solve for this ratio.
Then equation (15) becomes:
(20) R1 + R2 = 234.8490566
Then equation (19)
simplifies to:
(21) R1(-246.18µs) + R2(2858.82µs) =
-2933.82µs
Using simultaneous linear equations (18) and (19), solve for R1 and R2:
(22) |
Det = | 1 -246.18µ |
1 2858.82µ |
= 2858.82µ - -246.18µ = 3105µ |
(23) |
R1 = |
3105µ |
= |
(234.8490566)(
2858.82µ) - -2933.82µ
3105µ |
(24) |
R1 = 217.173913Ω |
(25) |
R2 = |
3105µ |
= |
-2933.82µ -
(234.8490566)(-246.18µ)
3105µ |
(26) |
R2 = 17.67514356Ω |
R1, R2, and R3 in ratio 217.173913
:
17.67514356 : 1 will yield
correct time constants for zeros if C1 and C2 are
chosen to yield
correct poles.
(27) |
AV<50Hz = |
R1 + R2 + R3a
+
R3b R3b |
= |
(217.173913 + 17.67514356 + 1)R3 R3b |
= |
(235.8490566)R3 R3b |
(28) |
R3b = |
(235.8490566)R3 AV<50Hz |
This calculation applies to an RIAA network normally used for passive equalization. Parallel topologies are also preferred for discrete transistor designs because stray circuit capacitances can be compensated in C2 and drain/collector resistances in R1. Application of Thevenin’s theorem moves R1 into parallel with C1 and R2, opening up the use of the same network for inverting and current input applications.
TP1 = 3180µs
TP2 = 75µs
TZ = 318µs = R2C1
R1C2 = 750µs
R1C1 = 2187µs
R1:R2 = 6.877358491
C1:C2 = 2.916
Passive parallel version verified by SPICE: Model, Inverse RIAA Subckt used in model.
(1) |
Z = R1||(R2+(1/sC1)))||(1/sC2) = | R1 || | sR2C1
+
1
sC1 |
|| | 1
sC2 |
(2) |
Z = | 1
|
|| | 1
sC2 |
= |
1
|
|| | 1
sC2 |
(3) |
Z = | sR1R2C1
+
R1
s(R1C1 + R2C1) + 1 |
|| | 1
sC2 |
= |
1
|
(4) |
Z = | 1
|
(5) |
Z = | sR1R2C1
+
R1
s2R1R2C1C2 + s(R1C1 + R2C1 + R1C2) + 1 |
Zero is clear:
(6) |
sTZ + 1 = | sR1R2C1
+
R1
R1 |
, therefore TZ = R2C1 = 318µs |
Solve for poles:
(7) |
s2TP1TP2 + s(TP1 + TP2) + 1 = 0 |
(8) TP1TP2 = R1R2C1C2 = (R2C1)R1C2 = TZR1C2 (from equations (5) and (7))
(8) |
TP1TP2 = R1R2C1C2 = (R2C1)R1C2 = TZR1C2 (from equations (5) and (7)) |
(9) |
R1C2 = | TP1TP2
TZ |
= |
(3180µs)(75µs)
318µs |
= 750µs |
(10) |
TP1 + TP2 = R1C1 + R2C1 + R1C2 (from equations (5) and (7)) |
(11) |
3180µs + 75µs = R1C1 + 318µs + 750µs |
(12) |
R1C1 = 3180µs + 75µs - 318µs - 750µs = 2187µs |
(13) |
R1
R2 |
= |
R1C1
R2C1 |
= |
2187µs
318µs |
= 6.877358491 |
(14) |
C1
C2 |
= |
R1C1
R1C2 |
= |
2187µs
750µs |
= 2.916 |
This calculation adds an extra 50kHz zero to an RIAA network
normally used for passive equalization. Parallel topologies are
also preferred for discrete transistor designs because stray circuit
capacitances can be compensated in C2 and drain/collector
resistances
in R1. The addition of R3 in series with C2
to add the
extra zero
somewhat diminishes this advantage, complicating the compensation for
stray capacitance. Application of Thevenin’s theorem moves R1
into parallel with C1 and R2, opening up the use
of the same
network
for inverting and current input applications.
(1) |
Z = R1 || (R2+1/sC1)
||
(R3 +
1/sC2) = R1 || |
sR2C1+1
sC1 |
|| |
sR3C2+1
sC2 |
(2) | 1 |
1 |
||||||||
Z = |
|
= |
|
|||||||
|
|
(3) |
R1(sR2C1+1)(sR3C2+1) | R1(sR2C1+1)(sR3C2+1) | ||
Z = |
|
= |
|
|
(s2R2R3C1C2
+
s(R2C1+R3C2) +1) + (s2R1R3C1C2 + sR1C1) + (s2R1R2C1C2 + sR1C2) |
s2(R2R3C1C2+R1R3C1C2+R1R2C1C2)
+ s(R2C1+R3C2+R1C1+R1C2) + 1 |
(4) |
s2TP1TP2
+ s(TP1+TP2) + 1 = 0 |
(5) |
TP1TP2 = R2R3C1C2 + R1R3C1C2 + R1R2C1C2 = TZ1TZ2 + TZ2R1C1 + TZ1R1C2 |
(6) |
TP1+TP2 = R2C1 + R3C2 + R1C1 + R1C2 = TZ1 + TZ2 + R1C1 + R1C2 |
(7) |
3.18µTA + 318µTB = 237.489n |
(8) |
TA + TB = 2.93382m |
(9) |
Det = | 3.18µ 1 |
318µ 1 |
= 3.18µ – 318µ = –314.82µ |
(10) |
TA = |
–314.82µ |
= |
237.489n –
(318µ)(2.93382m)
–314.82µ |
(11) |
TA = 2209.09µs = R1C1 |
(12) |
TB = |
–314.82µ |
= |
(3.18µ)(2.93382m) – 237.489n
–314.82µ |
(13) |
TB = 724.73µs = R1C2 |
(14) |
R1
R3 |
= |
TB
TZ2 |
= |
724.73µs
3.18µs |
= |
227.902 |
(15) |
R1
R2 |
= |
TA
TZ1 |
= |
2209.09µs
318µs |
= |
6.94682 |
(16) |
R2
R3 |
= |
R1/R3
R1/R2 |
= |
227.902
6.94682 |
= |
32.8066 |
Sometimes it is desireable to divide the equalization between two
stages. All practical circuits put the 3180µs pole and the 318µs
zero in one stage leaving the 75µs pole to the other.
Circuit
1:
Series/Parallel |
(2) |
|
R1 |
sR1R2C1
+
R1 + R2 |
|||||
Z = R2 + |
|
= R2 + |
|
= |
|
|||
|
sR1C1
+ 1 |
sR1C1
+ 1 |
(4) |
TZ1 = |
R1R2C1
R1 + R2 |
= (R1 || R2)C1 = 318µs |
(8) |
(TP1/TZ1) | ||||
|
= R1 |
||||
|
(9) |
1
R1 |
+ |
1
R2 |
= |
(TP1/TZ1)
R1 |
, therefore |
1
R2 |
= |
(TP1/TZ1) –
1
R1 |
(10) |
R1 = |
TP1
TZ1 |
– 1 | R2 |
Circuit
2:
Parallel/Series |
(2) |
|
sR2R1C1
+
R2 |
|||||||
Z = |
|
= |
|
||||||
|
s(R2
+ R1)C1
+ 1 |
(4) |
TZ1
= |
R2R1C1
R2 |
= R1C1 = 318µs |
(8) |
R2 = |
TP1
TZ1 |
– 1 | R1 |
75µs
Lowpass
Stage
Common
to
Both |
If you have comments on the calculations email me at contact@waynestegall.com
The RIAA equalization calculations above presume that no
equalization is needed for the input loop itself. For voltage
input designs, correct RC loading at the preamp input meet these
conditions, although not perfectly. Reading reviews of current
input designs did not mention any special compensation that I
remember. However when I reexamined this issue I found the
following result: The virtual short at the input reduces the
equivalent input loop to a RL response with a low cutoff frequency,
which would greatly attenuate the treble. This is because the
complex pole pair of the full second order circuit diverges into two
real poles below a Q of 0.5 (The second pole is pushed far into the
ultrasonic region with the Q values involved). Therefore it is
necessary to add a zero of the same value to the compensation circuit,
perhaps in a second stage. I present this result only after I
verified it by examining a patent for this type of circuit issued in
1984 (That it expired 17 years later in 2001 makes this a public domain
design.) Because in the winding of an inductor both R and L
increase with the number of turns, cartridges whose windings are made
with the same gauge magnet wire may all have the same or similar RL
response. This suggests the selection of a fixed compensation
zero as a design compromise. Others may want to have an
adjustable zero to tailor the compensation to their desired cartridge.
TZ3 = | L
R |
fZ3 = | R
2πL |
Recommended
circuit
allows
placement
of
compensation
in
either
stage. |
R1 = |
3180µs
TZ3 |
– 1 | R2 |
R4 = |
TZ3
75µs |
– 1 | R3 |
ZGAIN = |
(R1 + R2)R5
R4 |
Series-Parallel
in
both
stages
allowed
by
adjustment
in
first
stage |
Discrete Full Series-Parallel Circuit |
Note: Incomplete
schematic only illustrates equalization topology, biasing is likely
incorrect |
R1 = |
3180µs
TZ3 |
– 1 | R2 |
Parallel-Series in both stages allowed by adjustment in second stage |
R4 = |
TZ3
75µs |
– 1 | R3 |
Current-input
equalization
at
a
plausibly
typical
cartridge
LR
frequency
of
212.2Hz
shows
much
flatter
equalization
than
voltage input designs. |
|