Chapter 4 Proposed Building Blocks and Circuit Implementation
4.2 Building Blocks and Circuit Implementation
4.2.4 Calibration Circuits
Fig. 4-11 (a) shows the block diagram of calibration circuits. It contains two replica stages, a calibration comparator with offset calibration, and a digital loop filter.
As described in 3.4.3, because of incomplete-settling technique, there is sampling attenuation between stages. Sampling attenuation can be compensated:
( )⋅ =2
eff s y
G t G (4.8) From (4.8), Geff( )t is s 2
Gy . Because sampling attenuation happens between stages, if only the first replica stage with residue plot shown in Fig. 4-11 (b) is applied, a known analog input Vin known_ set 0 with D1_replica1 set 0 and D1_replica2 set -1 is given,
57
and an analog value set 2 Vref
which corresponds to Vin known_ is compared with the residue output Vout replica_ 1( )ts of the first replica stage at ts, assuming that the
comparator is calibrated and has no offset, from (4.1):
_ 1( ) ( )( _ 1_ 1 2 _ 1 )
out replica s eff s in known replica replica
ref
Gy . In other words, sampling attenuation cannot be compensated. If the second replica stage with residue plot shown in Fig. 4-11 (c) is added with D1_replica2
set 0 and D2 _replica2 set 1, and replica stage at ts, assuming that the comparator is calibrated and has no offset, from
(4.1) and (4.9):
out replica s eff s in replica replica replica
ref ref
eff s y out replica s replica replica
eff s y eff indeed partially compensates sampling attenuation. As described in 4.2.2,
58
dt of neighboring binary digital control bits
5, , , , , 4 3 2 1 0 To deserve to be mentioned,
2 Vref
used for calibration is generated in chip by a resistor-ladder which corresponds to 6-bit matching.
_ 0
59
The Second Replica Stage Calibration Point
(c)
Fig. 4-11 (a) block diagram of calibration circuits (b) residue plot of the first replica stage (c) residue plot of the second replica stage
Besides, offset of opamp of the replica stages will cause error when Geff( )ts is
determined. Assuming that offset of opamp of the fisrt and second replica stages is
60
1
Vos and Vos2, respectively, the residue output Vout replica_ 1( )t of the first replica stage s
at ts without considering sampling attenuation and offset of the calibration
comparator:
out replica s eff s in replica replica replica os
ref
out replica s eff s in replica replica replica os
ref ref
eff s out replica replica replica os
eff s eff error. Input-offset-storage (IOS) technique [18] is used to suppress the effect of offset of opamp of the replica stages. Fig. 4-12 (a) and (b) shows that input-offset-storage (IOS) technique is used in sampling phase and amplification phase, respectively. The output offset voltage of the outputs of opamp due to offset Vos is reduced to
0
61
Fig. 4-12 IOS technique is used (a) in sampling phase (b) in amplification phase
62
If offset of the calibration comparator after offset calibration is Vos cal_ , from
(4.13), equivalently, Vout replica_ 2( )ts is compared with ( 125 ) _ 2
ref
os cal
V = mV +V .
Assuming that sampling attenuation, Vos1, and Vos2 are not taken into consideration,
_ os cal
V with maximum ( 1.5 )
±V5LSB = ±
mV let Geff( )ts deviate from 2 by ±0.008 which is small enough refer to Fig. 3-8.
The circuit implementation of the calibration comparator with offset calibration is shown in Fig. 4-13. There are four binary-weighted current controlled by digital control bits , , , y3 y2 y1 y0 and z3, , , z2 z1 z0 at each side of the calibration
comparator to compensate offset of the calibration comparator before offset calibration.
φ1pb
φ1pb φ1pb
Vi+ Vi−
Do+
Do−
y3 y2 y1 y0
z3
z2
z1
z0
Fig. 4-13 circuit implementation of calibration comparator with offset calibration
63
Input-referred offset of the calibration comparator before offset calibration is
±17 mV ( 3σ ) in post-simulation. Fig. 4-14 shows embedded calibration offset v.s.
3 0
y ... y (or z ... z3 0 ) in decimal expression in post-simulation. Maximum
embedded calibration offset is ±17.9 mV with maximum 1.5 mV step.
0 2 4 6 8 10 12 14 16 18 20
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
embedded calibration offset
3, , , 2 1 0 ( , , , )3 2 1 0
y y y y or z z z z
Fig. 4-14 embedded calibration offset v.s. y ... y3 0 (or z ... z3 0)
Fig. 4-15 (a) shows the block diagram of the digital loop filter. It contains two cascaded sub-filter and back-end digital circuits. The function of the digital loop filter is to eliminate the disturbance of noise.
The first sub-filter shown in Fig. 4-15 (b) is composed of a front-end D flip-flop (DFF), two DFF chains, and two reset DFF. Each DFF chain has 5 DFF. The front-end DFF captures the comparison result of the calibration comparator. If the comparison
64
result is 1, the up DFF chain shifts 1 right and reset the down DFF chain, and vice versa. Once 1 is shifted to the most right of either the up or down DFF chain, both them are reset. For example, the outputs , , , , M4 _up M3_up M2 _up M1_up Up_1 of the up DFF chain are 11000 and the outputs , , , , M4 _dn M3 _dn M2 _dn M1_dn Dn_1 of the
down DFF chain are 00000 at this moment. If next comparison result is 1,
4 _up ... p_1
M U is 11100, and M4 _dn ... Dn_1 is 00000. And if next comparison result is 0, M4 _up ... Up_1 is 00000, and M4 _dn ... Dn_1 is 10000. Only that five consecutive 1 of the comparison result appears, Up_1 becomes 1, and vice versa.
The second sub-filter shown in Fig. 4-15 (c) is composed of a reset DFF and two DFF chains. Each DFF chain has 5 DFF. Before the beginning of sampling-point calibration, DFF of the DFF chains is reset by init. Then sampling-point calibration starts. The clock of the DFF chains is triggered by trig which becomes 1 when either Up_1 or Dn_1 becomes 1. When Up_1 becomes 1, the up DFF chain shifts 1
right, and the down DFF chain shifts 0 left, and vice versa. Once 1 is shifted to the most right of either the up or down DFF chain, both them are reset. For example, the outputs N4 _up, , , , N3 _up N2 _up N1_up Up_ 2 of the up DFF chain are 11100 and the outputs N4 _dn, , , , N3 _dn N2 _dn N1_dn Dn_ 2 of the down DFF chain are 10000 at this moment. If Up_1 becomes 1, N4 _up ... Up_ 2 is 11110, and N4 _dn ... Dn_ 2 is 00000.
And if Dn_1 becomes 1, N4 _up ... Up_ 2 is 11000, and N4 _dn ... Dn_ 2 is 11000.
65
Only that the accumulation of the number of times of 1 of Up_1 larger than that of
_1
Dn is five, Up_ 2 becomes 1, and vice versa.
The back-end digital circuits comprise the registers, a adder, and a subtracter. As described in 4.2.3, the registers is used to store current binary digital control bits
5, , , , , 4 3 2 1 0
66
Fig. 4-15 block diagram of (a) digital loop filter (b) the first sub-filter (c) the second sub-filter