Circuit Surgery
Regular clinic by Ian Bell
Electronically controlled resistance – Part 7
T
his month, we continue our
Example designs
The regulator in Fig.1 uses an internal
series on electronically controlled
The typical value for R1 in reference
voltage reference circuit to provide a
resistance by considering another
point of comparison with the (scaled)
designs (for example, in device datasheets)
application of digipots (digital potentiis in the order of 100Ω to 300Ω, with
output voltage. A feedback control
ometers), which are digitally controller
121Ω and 240Ω being very commonly
circuit, typically implemented using
variable resistors, often, but not always
used. For example, R1 = 121Ω and R2 =
an internal op amp, regulates the voltage
interfaced to microcontrollers. The prebetween the Out and Adj pins to be equal
365Ω and Vref = 1.25V gives Vout = 5V.
vious three articles covered the basics of
to the reference voltage (Vref). If the Adj
These examples typically use R1 as a
digipots, their characteristics, example
pin is connected directly to ground the
fixed resistor and R2 as a single variable
devices, modelling in LTspice and their
output is regulated to be equal to Vref.
resistor, which implies use of a digipot
use to control amplifier gain.
in rheostat mode.
If adjustment resistors (R1 and R2) are
The control of amplifier gain, which
The LM317 is probably the best-known
included (as shown in Fig.1) then the
we discussed last month, is not the only
device of this type. The LT1085 is similar,
output voltage is regulated to the voltage
application for digipots. There are of
but is ‘low drop out’ (LDO), which means
at the Adj pin plus Vref.
Electronically
controlled
7 difference in voltage between
course many possible specific applications
that–aPart
smaller
The voltage across
R1 is equal
to Vresistance
ref,
– common circuits which could benefit
input and output is required to maintain
so the current through it is IR1 = Vref/R1.
from digipot-based control include voltage
operation (Vin needs to be about 3V above
If we ignore the current required by the
regulators and analogue filters. This
Adj pin (Iadj) the same current flows in
Vout for the LM317, but only 1V for the
𝑅𝑅'
month, we will look at using digipots
R1 and R2, so the voltage across R2 is
LT1085,
giving
efficiency). The
𝑉𝑉!"# = 𝑉𝑉$%&
#1 +superior
'
controlled resistance – Part 7
𝑅𝑅(
with adjustable Electronically
linear voltage
regulator
LT1085 can provide
up to 3A output
IR1R2 = VrefR2/R1. The output voltage is
circuits. Replacing the adjustment resistor
current and has a Vref value of 1.25V.
the voltage across R2 plus the reference
with a digipot may seem straightforward,
voltage, so:
As an example design, consider
controlled
but there are some Electronically
issues that need
to resistance – Part 7 𝑅𝑅
adjustable output up to 15V using an
𝑅𝑅'
'
𝑉𝑉!"# = 𝑉𝑉$%& #1 + '
𝑉𝑉!"#LT1085.
= 𝑉𝑉$%& #1If+we use
' + 𝐼𝐼the
be considered and which may not be
value of R1
)*+ 𝑅𝑅typical
'
𝑅𝑅(
𝑅𝑅
immediately obvious. These are related to
of 121Ω then (15V requires R2 = 1.33kΩ,
the current and voltage ratings, available
which we can find by rearranging the
The current in the adjustment pin adds
𝑅𝑅'
component values and tolerance, which
Vout equation above:
an additional
contribution
𝑉𝑉!"# = 𝑉𝑉$%&
#1 + ' to the voltage
𝑅𝑅( also adds to the
are parameters that may to be of concern
drop across R
𝑅𝑅'2, which
𝑉𝑉!"#
= 𝑉𝑉$%&voltage,
#1 + ' giving
+ 𝐼𝐼)*+ 𝑅𝑅'a more accurate
𝑅𝑅' = 𝑅𝑅( )
− 1+
in other types of circuit where digipots 𝑉𝑉!"#
output
𝑅𝑅(
𝑉𝑉$%&
are considered.
formula for Vout, as:
We also need a minimum input voltage
𝑅𝑅'
of about 16V.
Three-terminal regulators
𝑉𝑉!"# = 𝑉𝑉$%& #1 + ' + 𝐼𝐼)*+ 𝑅𝑅'
𝑅𝑅(
𝑉𝑉!"#
A common form of adjustable regulator
1
1
1
𝑅𝑅' = 𝑅𝑅( )
− 1+
=
−
circuit is shown in Fig.1. Regulator
First
example
However,
the
𝑉𝑉$%&adjustment pin current is
𝑅𝑅,simulation
𝑅𝑅' 𝑅𝑅integrated circuits of this type are often
typically in the order of 50 to 100µA, and
For simulation purposes we can select ten
referred to as ‘3-terminal regulators’,
with the typical
values of the resistors
different values, stepped from 133Ω to R2 =
𝑉𝑉!"#
𝑅𝑅( )
− 1+ of ohms range
reflecting the minimal pin count. We
being 𝑅𝑅in
the
hundreds
1.33kΩ
in 133Ω steps. The specific values
' =
𝑉𝑉$%&
1
1 IadjR
1 2 is small and can often
𝑅𝑅( much for illustrative
will look at using a digipot for R2 (which
the value
of
do not matter too
=
−
𝑉𝑉!"# = 3.75 #1 + '
𝑅𝑅, 𝑅𝑅' 𝑅𝑅purposes
– we are
adjusts the output voltage).
be ignored.
𝑅𝑅'not aiming to show all
Input
IN
Regulator
Adj
Output
OUT
Vref
IAdj
R1
Cin
Cout
R2
Fig.1. Typical 3-terminal voltage
regulator circuit.
56
1
1
1
=
−
𝑅𝑅, 𝑅𝑅' 𝑅𝑅 𝑅𝑅(
𝑉𝑉!"# = 3.75 #1 + '
𝑅𝑅'
𝑅𝑅(
𝑉𝑉!"# = 3.75 #1 + '
𝑅𝑅
𝑅𝑅'
𝑅𝑅./'
𝑉𝑉!"# #
' = 𝑉𝑉$%& #
'
𝑅𝑅( + 𝑅𝑅'
𝑅𝑅./ + 𝑅𝑅0/
𝑅𝑅
𝑅𝑅./
𝑅𝑅'+ ' ' #
𝑅𝑅./
𝑉𝑉!"# = 𝑉𝑉$%& #1
'
𝑉𝑉!"# #
'
𝑅𝑅'(= 𝑉𝑉𝑅𝑅$%&
./#+ 𝑅𝑅0/
𝑅𝑅( + 𝑅𝑅'
𝑅𝑅./ + 𝑅𝑅0/
𝑅𝑅'
𝑅𝑅./
𝑉𝑉!"# #
' = 𝑉𝑉$%& #
'
𝑅𝑅( + 𝑅𝑅'
𝑅𝑅./ + 𝑅𝑅0/
𝑉𝑉!"# = 𝑉𝑉$%& #1 +
𝑅𝑅'
𝑅𝑅./
'#
'
𝑅𝑅( 𝑅𝑅./ + 𝑅𝑅0/
𝑉𝑉!"# = 3.75𝑁𝑁 #1 +
𝑅𝑅'
'
𝑅𝑅(
Fig.2. Possible use of a digipot with a 3-terminal regulator (see text for issues).
𝑅𝑅'
𝑅𝑅./
'#
'
𝑅𝑅( 𝑅𝑅 𝑅𝑅./ + 𝑅𝑅0/
'
𝑉𝑉!"# = 3.75𝑁𝑁 #1 + '
𝑅𝑅(
𝑉𝑉!"# = 𝑉𝑉$%& #1 +
Practical Electronics | March | 2023
Fig.3. Simulation results from the circuit in Fig.2 showing output for various resistance values.
designs given in datasheets, where it
is assumed that a standard mechanical
potentiometer is used for R2. The digipot
is modelled using a stepped-resistor
parameter – the behavioural resistors
discussed previously are not required
here as only a single resistor is required to
model the digipot in rheostat mode. The
simulation ramps the input voltage from 0
to 20V over two seconds. The time is not
particularly important, but two seconds
gives a simple time-to-voltage relationship.
The results of the simulation are shown
in Fig.3 – the step command plots a set of
different results as Rdigipot is varied.
The result show that the output voltage
is regulated once the input voltage rises
about 1V above the set output voltage, so,
for example, the 15V output (Rdigipot =
1.33kΩ) appears after about 1.6s and the
other voltages at proportionally earlier
times. This can be seen more clearly in
Fig.4, where the input voltage is also
plotted (but this makes it harder to
identify the individual stepped traces).
Digipot current
Fig.5 shows the current in the digipot.
Once regulation is established (for a
sufficiently large input voltage) the
current is constant at about 10.3mA.
This is equal to V ref/R 1 = 1.25/121 =
10.33mA. The problem with this is that
this current is higher than the maximum
continuous current rating for many
digipots, which may be 5mA or 1mA.
Low resistor values are used in standard
versions of this circuit, partly to ensure
that the resistor current is much larger
Fig.4. Simulation results from the circuit in Fig.2 showing input-output relationship
than the adjustment pin current (55µA
(input: red trace; outputs: green traces).
Electronically controlled resistanceto– 120µA
Part 7 for the LT1085) and partly to
take the minimum load current from the
regulator output (5mA to 10mA for the
LT1085). The resistors can be made larger
𝑅𝑅'
to reduce the current,
and, if necessary,
𝑉𝑉!"# = 𝑉𝑉$%& #1 + '
𝑅𝑅( resistor could be added
an additional load
to ensure minimum current.
It may be possible to overcome the
maximum current problem by using a
𝑅𝑅'
with
𝑉𝑉!"# =parallel
𝑉𝑉$%& #1 +resistor
' + 𝐼𝐼)*+
𝑅𝑅' the digipot. This
𝑅𝑅( Fig.6,
is shown in
which is a modified
version of Fig.6. For a required maximum
effective value of R2 (the total parallel
resistance for the maximum output
𝑉𝑉!"#
and a digipot value
𝑅𝑅voltage
1+
' = 𝑅𝑅( )to be−used)
of RD the𝑉𝑉$%&
parallel resistor value can be
found using the parallel resistor formula
rearranged to give RP:
Fig.5. Digipot resistor current for the circuit in Fig.2.
the possible steps of the typical digipot.
Furthermore, a digipot with an end-to-end
resistance of 1.33kΩ is unlikely to exist;
however, devices with a resistance larger
than this can be used with the setting
restricted to the required values. Not
Practical Electronics | March | 2023
using the full range of values is also often
necessary in rheostat mode to account for
the device resistance tolerance – as we
discussed in Part 5.
An LTspice schematic for this circuit is
shown in Fig.2. This is similar to example
1
1
1
=
−
𝑅𝑅, 𝑅𝑅' 𝑅𝑅-
If we use a larger value of R1 (than in
Fig.1) of 240Ω then we need R2 = 2.64kΩ
for Vout = 15V.𝑅𝑅(For a 5kΩ digipot (RD)
𝑉𝑉!"#
3.75 #1
the=value
of +
RP𝑅𝑅is' 1/(1/2640 – 1/5000) =
'
5.56kΩ (as shown in Fig.6).
Simulation results are shown in Fig.7,
which confirm that the circuit is able to
𝑅𝑅'
𝑅𝑅./
𝑉𝑉!"# #
' = 𝑉𝑉$%& #
'
𝑅𝑅( + 𝑅𝑅'
𝑅𝑅./ + 𝑅𝑅0/
𝑅𝑅'
𝑅𝑅./
57
Fig.6. Modified version of the circuit in Fig.2 with reduced digipot current.
control the output voltage over a similar
range to the circuit in Fig.2. However,
notice that the relationship between the
resistor step value (which corresponds to
the digipot code) and the output voltage
is no longer linear (output voltage steps
are not equally spaced) due to the use
of the parallel resistor combination to
set the effective R2 value. The results in
Fig.8 confirm the reduction in current
through the digipot, which remains below
5mA over the range of input and output
voltages covered
Accuracy
As we discussed in previous articles,
digipots often have much worse
tolerance for their absolute resistance
value than is readily available for fixed
resistors. In potentiometer mode, where
the ratio of the two resistances from
the wiper to the two ends controls the
circuit, the excellent matching of the
on-chip resistors to each other can lead
to very good design tolerance. However,
in rheostat mode, as used here, that
does not apply.
Using a digipot in rheostat mode in
parallel with accurate fixed resistors
reduces errors due to digipot tolerance,
so this is another benefit of the circuit
in Fig.6 over that in Fig.2. Use of series
resistances with digipot in rheostat mode
also has the same effect. We previously
discussed use of series resistors to
reduce the control range to allow more
precise control of a circuit – it is often
worth considering use of digipots in
combination with fixed resistors to
improve or better tailor the control of
the circuit.
If we assume the fixed resistors
have negligible error in value and the
digipots have a 10% tolerance, then we
can calculate the variation in output
voltage for the two worst-case digipot
resistance values (at ±10%). This is just
a matter of applying the Vout equation
with the relevant highest and lowest
digipot tolerance values, and, in the
case of the circuit in Fig.5 applying the
parallel resistor formula to obtain the
effective R2 value. This gives a variation
of about 9.2% for the circuit in Fig.2
and worst-case of 5.1% for the circuit
in Fig.6 for the 15V output setting,
which is a significant improvement.
Alternative regulators
Fig.7. Simulation results from the circuit in Fig.6.
There is another major problem which
may make the circuits in Fig.2 and Fig.6
impractical – at least for the voltage ranges
used here – which is the maximum voltage
across the digipot. Many digipots have
maximum supply voltages of 5V (maybe
5.5V absolute maximum) and typically
the maximum voltage across the resistor
chain is equal to, or very similar to the
supply voltage. Devices with 15V are
available but tend to have larger resistor
values, which may be less suitable in
the type of regulator in these examples.
Not all adjustable regulators are
3-terminal devices. Often, regulators have
an adjustment input which is separate
from the regulator ground – a typical
circuit is shown in Fig.9 which features
the LT1121. The LT1121 is a micropower
LDO regulator capable of supplying
Vin
IN
LT1121
SHDN
GND
Fig.8. Digipot resistor current for the circuit in Fig.6.
58
Vout
OUT
R1
+
ADJ
Cout
R2
Fig.9. Typical adjustable regulator circuit
for a device with separate adjust and
ground pins.
Practical Electronics | March | 2023
𝑅𝑅(
𝑉𝑉!"#
− 1+
𝑉𝑉$%& is between
voltage (3.75V)
𝑅𝑅' = 𝑅𝑅( )
ground and the adjustment
pin, rather than the output
and the adjustment pin.
1
1
1
The voltage
=
−across R2 sets
𝑅𝑅
𝑅𝑅
𝑅𝑅
,
'
its current (and -the current
in R1) to 3.75/R2 and so the
output voltage is:
𝑉𝑉!"# = 3.75 #1 +
Fig.10. LTspice schematic for an LT1121 regulator with digipot control.
voltage to obtain 3.75V at the adjustment
150mA of output current with a dropout
pin. Like the LT1085, some current flows
voltage of 0.4V. The output voltage is
into the adjustment pin. This is about
adjustable from 3.75V to 30V (fixed output
150nA for the LT1121, much smaller than
versions are also available). The device
the LT1085 and similar devices, which
has a 5V logic, active-low shutdown pin
facilitates use of larger resistors in the
(output turned off if shutdown is driven
feedback network (R1 and R2 in Fig.9).
below about 0.6V). The shutdown pin
can be floating if not used.
Internally, the adjustment pin is
Circuit operation
connected to the control circuit input.
The operation of this circuit is similar
The control circuit adjusts the output
to the circuit in Fig.1 except the reference
𝑅𝑅(
'
𝑅𝑅'
Note that the resistor
ratio in the formula is
the
round
𝑅𝑅' opposite way𝑅𝑅./
𝑉𝑉!"# # compared
' = 𝑉𝑉$%&
'
to #the
𝑅𝑅( + 𝑅𝑅'
𝑅𝑅./formula
+ 𝑅𝑅0/
for the circuit in Fig.1, with
the resistors
numbered
𝑅𝑅'
𝑅𝑅./
𝑉𝑉!"# = 𝑉𝑉$%& #1 + ' #
'
similarly.
To avoid confusion,
you
𝑅𝑅( 𝑅𝑅./ +if𝑅𝑅0/
consult the datasheet, also note that this
uses a different resistor numbering.
Like the previous circuit, an error
voltage of IadjR𝑉𝑉1 can
be added 𝑅𝑅
to' 'Vout to
!"# = 3.75𝑁𝑁 #1 +
𝑅𝑅(current
account for the adjustment pin
if required. The datasheet states that the
value of R2 should be less than 400kΩ,
so a 20kΩ is easily suitable and may be
available for 15V device. A 50kΩ or 100kΩ
digipot could also be used.
Fig.10 shows an LTspice schematic of
the LT1121 circuit with digipot control.
The digipot is in potentiometer mode
and has a resistance of RAB. There is a
resistor (R3) between the digipot and
ground – this sets the maximum output
voltage when the digipot wiper is fully
at the B end, where the effective value
of R1/R2 in the above formula is RAB/R3.
If we want a maximum output voltage
of 15V then we need R1/R2 = 3 (because
3.75 × (1 + 3) = 15), so with RAB = 20kΩ
we get R3 = RAB/3 = 6.67kΩ. When the
digipot is fully at the A end R1 (= RWA)
is zero (ideally) so Vout = 3.75V.
Simulation models
Fig.11. Simulation results from the circuit in Fig.10.
The digipot in Fig.10 is modelled using
two parameters. The total resistance
(RAB) is parameter Rdigipot_AB. The
resistance between the wiper and B
terminal (RWB) is parameter Rdigipot_
WB. One of the digipot resistors is simply
equal to Rdigipot_WB, the other (wiper
to A terminal, RWA) is calculated using a
behavioural resistor with value:
R = {Rdigipot_AB} - {Rdigipot_WB}
Fig.12. Digipot resistor current for the circuit in Fig.10.
Practical Electronics | March | 2023
The simulation steps Rdigipot_WB through
a set of values to give a simulation similar
to the other examples above. As discussed
when we used behavioural resistors in
Part 5, it is necessary to ensure that the
calculated value resistor is never zero, so
the first stepped value of Rdigipot_WB is
1mΩ rather than zero. Also, Rdigipot_WB
is a little larger than 20kΩ to prevent the
subtraction returning zero.
59
𝑉𝑉!"# = 𝑉𝑉$%& #1 +
𝑉𝑉!"# = 𝑉𝑉$%& #1 +
Electronically controlled resistance – Part 7
𝑅𝑅'
'
𝑅𝑅(
𝑅𝑅'
' + 𝐼𝐼)*+ 𝑅𝑅'
𝑅𝑅(
𝑉𝑉!"#
𝑅𝑅' = 𝑅𝑅( )
− 1+
𝑅𝑅𝑉𝑉'$%&
𝑉𝑉!"# = 𝑉𝑉$%& #1 + '
𝑅𝑅(
1
1
1
=
−
𝑅𝑅,𝑅𝑅' 𝑅𝑅' 𝑅𝑅𝑉𝑉!"# = 𝑉𝑉$%& #1 + ' + 𝐼𝐼)*+ 𝑅𝑅'
𝑅𝑅(
𝑅𝑅
𝑅𝑅'
(
Fig.13. LTspice schematic for digipot regulator control with
lower
voltage requirements.
𝑉𝑉!"# =
3.75digipot
#1 + supply
'
Simulation results are shown in Fig.11.
The basic pattern is similar to the previous
examples, but the steps are non-linear.
This may make it difficult to set specific
values at the higher voltage end (depending
on the number of wiper positions). The
situation can be improved with series and
parallel resistances and was discussed in
the context of amplifier gain last month.
The results in Fig.12 show a much lower
digipot current than in the previous
examples – under 600µA at all points.
𝑉𝑉!"#
𝑅𝑅' = 𝑅𝑅voltage
− 1+
()
divider
𝑉𝑉$%& for the digipot (the two
inputs to op amp U2), thus:
𝑅𝑅'
𝑅𝑅WB
./
𝑉𝑉!"# #
' = 𝑉𝑉$%& #
'
𝑅𝑅
+
𝑅𝑅
𝑅𝑅
./ + 𝑅𝑅0/
WB
WA
1 ( 1' 1
=
−
𝑅𝑅, 𝑅𝑅' 𝑅𝑅-𝑅𝑅
𝑅𝑅./
'
𝑉𝑉!"# = 𝑉𝑉$%& #1 + ' #
'
𝑅𝑅( 𝑅𝑅./
0/
WB + 𝑅𝑅WA
At the same time, the regulator’s internal
𝑅𝑅( the adjustment pin
circuitry
ensures
𝑉𝑉!"# = 3.75
#1 + that
'
'
is at the reference𝑅𝑅voltage
𝑅𝑅'(Vref, which is
3.75𝑁𝑁
+ ' The digipot
!"# = for
equal to 𝑉𝑉
3.75V
the#1
LT1121).
𝑅𝑅(
is connected between the adjustment pin
and ground so it has a constant 3.75V
Lower digipot supply
𝑅𝑅'
𝑅𝑅./
it,
which
The circuit in Fig.10 has a low digipot𝑉𝑉!"# #across
'=
𝑉𝑉$%& #should be no
' problem for
𝑅𝑅( + 𝑅𝑅'
𝑅𝑅./ + 𝑅𝑅0/
devices running from 5V supplies.
current but does not allow the digipot
As discussed
last month,
if we define the
to control the full output range of the
𝑅𝑅'
𝑅𝑅./
= 𝑉𝑉$%& #1
+ ' #as a value N,
' where N = 0
position
regulator (up to 30V) because the full𝑉𝑉!"# wiper
𝑅𝑅( 𝑅𝑅./ + 𝑅𝑅0/
with the wiper
at B and N = 1 when the
output voltage occurs across the digipot.
wiper is A, then the digipot potential divider
Most digipots have maximum supply
function becomes NVref, so we have:
voltages below 15V. A possible solution
to this is provided in an article by Robert
𝑅𝑅'
𝑉𝑉!"# = 3.75𝑁𝑁 #1 + '
Swartz in Electronic Design (January
𝑅𝑅(
2016): Extend Low-Voltage Digipot
Resolution to Control an Adjustable
Design calculations
Regulator. A version of this technique
The ratio of R1 and R2 sets the maximum
applied to the LT1121 is shown in Fig.13.
output voltage for the wiper position N =
The circuit in Fig.13 contains an additional
1. For an output of 30V (the maximum for
op amp in the regulation control loop, which
the LT1121) we need R1/R2 = 7 (because
ensures that the potential divider voltage
3.75 × 1 × (1 + 7) = 30). The circuit in
from R1 and R2 is the same as the potential
Fig.13 uses 70kΩ and 10kΩ to achieve
this. The output voltage is linearly
dependent on the wiper position, which
is an advantage over the circuit in Fig.10.
The circuit can operate with values of N
which give Vout at 3.75V or more. With R1/
R2 = 7 the minimum value of N is RAB/8.
For a 20kΩ digipot this is 2.5kΩ – the RWB
resistance (Rdigipot_AB parameter) in the
simulation is stepped from this minimum
value to account for this. In a real system
this would have to be ensured by the
software. Alternatively, a suitable resistor
can be placed in series with the digipot, in
a similar way to R3 in Fig.10, which allows
the full range of the digipot to be used.
Simulation results are shown in Fig.14.
Again, this is similar to previous results,
but shows the larger output range of 3.75V
to 30V and the linear relationship between
the digipot setting at the output voltage
(evenly spaced output voltage traces).
This circuit requires more components
(particularly the op amp) but provides
significant advantages over the others
discussed above. The fixed voltage across
the digipot (once the input voltage is
sufficient for the regulator to operate)
leads to a constant current through the
digipot of Vref/RAB = 3.75/20k = 188µA.
The discussion in this article has
focused on the use of digipots to control
voltage regulators. We have not discussed
all aspects of the circuits, for example
selection of suitable values for the
capacitors. Some regulator circuits may
require additional components which
have not been included in the schematics
here – for example, protection diodes.
In all cases, device datasheets and other
relevant technical document from device
manufacturers should be consulted when
developing designs.
Simulation files
Fig.14. Simulation results from the circuit in Fig.13.
60
Most, but not every month, LTSpice
is used to support descriptions and
analysis in Circuit Surgery.
The examples and files are available
for download from the PE website.
Practical Electronics | March | 2023
