ak4646 AKM Semiconductor, Inc., ak4646 Datasheet - Page 37
![no-image](/images/no-image-200.jpg)
ak4646
Manufacturer Part Number
ak4646
Description
Stereo Codec With Mic/spk-amp
Manufacturer
AKM Semiconductor, Inc.
Datasheet
1.AK4646.pdf
(79 pages)
Available stocks
Company
Part Number
Manufacturer
Quantity
Price
Part Number:
ak4646EN-L
Manufacturer:
AKM
Quantity:
20 000
Company:
Part Number:
ak4646EZ-L
Manufacturer:
PHI
Quantity:
2 500
Part Number:
ak4646EZ-L
Manufacturer:
AKM
Quantity:
20 000
(5) 5-band Notch
This block can be used as Equalizer or Notch Filter. 5-band Equalizer (EQ1, EQ2, EQ3, EQ4 and EQ5) is selected
ON/OFF independently by EQ1, EQ2, EQ3, EQ4 and EQ5 bits. When Equalizer is OFF, the audio data passes this block
by 0dB gain. E1A15-0, E1B15-0 and E1C15-0 bits set the coefficient of EQ1. E2A15-0, E2B15-0 and E2C15-0 bits set
the coefficient of EQ2. E3A15-0, E3B15-0 and E3C15-0 bits set the coefficient of EQ3. E4A15-0, E4B15-0 and E4C15-0
bits set the coefficient of EQ4. E5A15-0, E5B15-0 and E5C15-0 bits set the coefficient of EQ5. The EQx (x=1 ∼ 5)
coefficient should be set when EQx bit = “ 0 ” or PMADL=PMADR= PMDAC bits = “0”.
When gain of K is set to “-1”, the equalizer becomes notch filter. When it is used as notch filter, central frequency of a real
notch filter deviates from the above-mentioned calculation, if its central frequency of each band is near. The control soft
that is attached to the evaluation board has functions that revise a gap of frequency and calculate the coefficient. When its
central frequency of each band is near, the central frequency should be revised and confirm the frequency response.
Note 34. [Translation the filter coefficient calculated by the equations above from real number to binary code (2’s
MS0557-E-02
Register setting (Note 34)
A
Transfer function
H(z) = 1 + h
h
The center frequency should be set as below.
MSB of each filter coefficient setting register is sine bit.
fs: Sampling frequency
fo
fb
K
X = (Real number of filter coefficient calculated by the equations above) x 2
X should be rounded to integer, and then should be translated to binary code (2’s complement).
(n = 1, 2, 3, 4, 5)
n
n
(z) = A
1
= K
1
1
(n = 1, 2, 3, 4, 5)
EQ1: E1A[15:0] bits =A
EQ2: E2A[15:0] bits =A
EQ3: E3A[15:0] bits =A
EQ4: E4A[15:0] bits =A
EQ5: E5A[15:0] bits =A
(MSB=E1A15, E1B15, E1C15, E2A15, E2B15, E2C15, E3A15, E3B15, E3C15, E4A15, E4B15, E4C15,
fo
E5A15, E5B15, E5C15; LSB= E1A0, E1B0, E1C0, E2A0, E2B0, E2C0, E3A0, E3B0, E3C0, E4A0, E4B0,
E4C0, E5A0, E5B0, E5C0)
~ K
~ fo
~ fb
complement)]
n
n
/ fs < 0.497
5
x
5
5
: Gain ( − 1 ≤ K
n
: Center frequency
: Band width where the gain is 3dB different from center frequency
1
(z) + h
1− B
1 + tan (πfb
tan (πfb
1 − z
n
z
2
(z) + h
−1
− C
−2
n
n
/fs)
n
≤ 3)
3
n
z
(z) + h
/fs)
−2
1
2
3
4
5
, E1B[15:0] bits =B
, E2B[15:0] bits =B
, E3B[15:0] bits =B
, E4B[15:0] bits =B
, E5B[15:0] bits =B
4
(z) + h
,
B
n
5
= cos(2π fo
(z)
1
2
3
4
5
, E1C[15:0] bits =C
, E2C[15:0] bits =C
, E3C[15:0] bits =C
, E4C[15:0] bits =C
, E5C[15:0] bits =C
n
/fs) x
- 37 -
1 + tan (πfb
2
1
2
3
4
5
n
/fs)
13
,
C
n
=
1 + tan (πfb
1 − tan (πfb
n
n
[AK4646]
/fs)
/fs)
2007/05