
Data Sheet
AD5560
Rev. D | Page 39 of 68
10. Calculate FZ, the ESR zero frequency, using FZ =
1/(2πRcCr).
11. If FP > Fug, the load pole is above the bandwidth of the
AD5560. Ignore it with RZ[2:0] = 0, RP[2:0] = 0. This ends the
algorithm
12. If RC < (R0/25), then the ESR is negligible. Attempt to
cancel the load pole with RZ zero. Choose an ideal zero
frequency of 2 × FP for some safety margin and then
choose the RZ[2:0] value that gives the closest frequency on a
logarithmic scale. This ends the algorithm
13. Otherwise, this is a troublesome window in which a load
pole and a load zero can’t be ignored. Use the following
steps:
To cancel the load pole at FP, choose an ideal zero
frequency of 6 × FP (this is more conservative than the
2 × FP suggested earlier, but there is more that can go
wrong with miscalculation). Then choose the RZ[2:0]
value that gives the closest zero to this ideal frequency
of 6 × FP on a logarithmic scale.
To cancel the ESR zero at FZ, choose an ideal pole
frequency of 2 × FZ.
Then choose the RP[2:0] value that gives the closest pole
to this ideal frequency of 2 × FZon a logarithmic scale.
This ends the algorithm
ADJUSTING THE AUTOCOMPENSATION MODE
The autocompensation algorithm assumes that there is 1 Ω of
resistance (RC) from the AD5560 to the DUT. If a particular
application has resistance that differs greatly from this, then
it is likely that the autocompensation algorithm is nonoptimal.
If using the autocompensation algorithm as a starting point,
consider that overstating the CR capacitance and understating
the ESR RC is likely to give a faster response but could cause
oscillations. Understating CR and overstating RC is more likely
to slow things down and reduce phase margin but not create
an oscillator.
It is often advisable to err on the side of simplicity. Rather than
insert a pole and zero at similar frequencies, it may be better to
add none at all. Set RP[2:0] = RZ[2:0] = 0 to push them beyond the
AD5560 bandwidth.
DEALING WITH PARALLEL LOAD CAPACITORS
In the event that the load capacitance consists of two parallel
capacitors with different ESRs, it is highly likely that the overall
complex impedance at the unity gain bandwidth is dominated
by the larger capacitor and its ESR. Assuming that the smaller
capacitor does not exist normally is a safer simplifying assump-
tion.
A more complex alternative is to calculate the overall impedance
at the expected unity gain bandwidth and use this to calculate
an equivalent series CR and RC that have the same complex
impedance at that particular frequency.
DAC LEVELS
This device contains all the dedicated DAC levels necessary
for operation: a 16-bit DAC for the force amplifier, two 16-bit
DACs for the clamp high and low levels, two 16-bit DACs for
the comparator high and low levels, a 16-bit DAC to set a
programmable open sense voltage, and a 16-bit offset DAC
to bias or offset a number of DACs on chip (FORCE, CLL,
CLH, CPL, CPH).
FORCE AND COMPARATOR DACS
The architecture of the main force amplifier DAC consists of
a 16-bit R-2R DAC, whereas the comparator DACs are resistor-
string DACs followed by an output buffer amplifier. This
resistor-string architecture guarantees DAC monotonicity.
The 16-bit binary digital code loaded to the DAC register
determines at what node on the string the voltage is tapped
off before being fed to the output amplifier.
The comparator DAC is similarly arranged. The force and
comparator DACs have a 25.62 V span, including overrange
to enable offset and gain errors to be calibrated out.
The transfer function for these 16-bit DACs is
DUTGND
CODE
DAC
OFFSET
VREF
CODE
DAC
VREF
V
OUT
+
×
×
=
16
2
_
125
.
5
2
125
.
5
section).
CLAMP DACS
The architecture of the clamp DAC consists of a 16-bit resistor-
string DAC followed by an output buffer amplifier. This resistor-
string architecture guarantees DAC monotonicity. The 16-bit
binary digital code loaded to the DAC register determines at
what node on the string the voltage is tapped off before being
fed to the output amplifier.
The clamp DACs have a 25.62 V span, including overrange, to
enable offset and gain errors to be calibrated out.