This is only a preview of the September 1994 issue of Silicon Chip. You can view 30 of the 96 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Articles in this series:
Items relevant to "Automatic Discharger For Nicad Battery Packs":
Items relevant to "Build The MiniVox Voice Operated Relay":
Items relevant to "An AM Radio For Aircraft Weather Beacons":
Articles in this series:
Items relevant to "Dual Diversity Tuner For FM Microphones; Pt.2":
Articles in this series:
Articles in this series:

How to use the
TEA1100 fast nicad
charger IC
The TEA1100 nicad charger IC, as used in
our recent Fast Nicad Charger project, has a
number of interesting features which put it out
in front. These include digital voltage sampling
and filtering as well as switchmode or linear
operation. We look at these in detail and go
through some design examples.
By DARREN YATES & LEO SIMPSON
The Philips TEA1100 Battery charger IC is a 16pin DIP package which
contains everything to produce a
simple yet highly integrated battery
charger for nickel cadmium (NiCd) and
the new nickel metal hydride (NiMH)
batteries which have a higher capacity
than nicads.
The TEA1100 has three methods of
guarding against overcharging: temperature detection, clock timeout and
an advanced form of voltage detection
referred to as “dV sensing”.
Supply voltage
Unfortunately, the chip has a fairly
awkward supply voltage range, which
is between 5.65V and 11.5VDC. This
makes it not quite suitable for car
operation without supply regulation
circuitry and not low enough to permit operation from a 5VDC regulator.
However, a 7808 or 7809 regulator
will be more than adequate and, in
fact, you can even get away with just
a standard zener diode/transistor
buffer voltage stabiliser such as that
used in the Fast Nicad Charger design
published in the May 1994 issue of
SILICON CHIP.
Linear or switchmode
As we mentioned in the introduction, the TEA1100 IC can run in linear
or switchmode operation. The benefit
of the switchmode option is the efficiency which can be gained by charging lower voltage cells from a higher
Fig.1: this circuit using the TEA1100 in switchmode was the basis for the Fast Nicad Charger published in
the May 1994 issue. The design example in the text will enable you to tailor the circuit to your application.
6 Silicon Chip
Fig.2: this linear version of the
TEA1100 circuit could be used
in RFsensitive applications. In
this case, the output of the chip
is taken from pin 2 rather than
the PWM output of pin 1.
voltage supply rail. This feature was
used in our Fast Nicad Charger project.
If you’re wanting to charge Nicads
in a noise sensitive application then
you can easily set the IC up to charge
in linear mode, greatly reducing the
circuit noise. The linear circuit uses
less components than the switchmode
circuit but has considerably more heat
dissipation, as you would expect.
Fig.1 shows a sample switchmode
circuit, very similar to that featured
in our May 1994 issue. Fig.2 shows a
linear charge circuit, powered from the
240VAC mains supply. And finally, to
give some idea of the chip complexity,
Fig.3 shows the block diagram of the
TEA1100.
dV sensing
Instead of comparing the voltage of
the battery being charged to a static
voltage reference, the TEA1100 uses a
dynamic process called “dV sensing”.
The “dV” term comes from calculus
and refers to the process of looking for
a very small change in battery voltage.
The TEA1100 compares the present
battery voltage to the previous sampled voltage and checks for a 1% drop.
The theory behind this is that when a
nicad is being charged, its voltage rises
very gradually towards full capacity
but once past this point, the battery
voltage begins to drop slightly.
If a battery charger circuit does not
look for this voltage drop, it will never
give an optimum charge – it will either
under or overcharge. Either way, the
battery life will ultimately be reduced.
Fig.4 shows the characteristic rise in
battery voltage during charge and the
slight droop as it reaches full charge.
The TEA1100 ends the charge cycle
upon sensing a 1% drop in the battery
voltage. That amounts to about 16mV
for a typical nicad cell. Now you might
be wondering how they manage to
reliably detect 16mV when the circuit
lines could be subject to all sorts of
noise and switching frequencies, if the
chip is being operated in switchmode.
The answer lies in the method of sampling the battery voltage.
The PWM (pulse width modulation)
is disabled for 10 clock cycles, after
which the sample and hold amplifier
takes a meas
urement of the battery
voltage. This way, the noise generated
by a “ringing” or decaying supply rail
is removed and a much greater degree
of accuracy maintained.
The 10cycle delay gives sufficient
time for the inductor to stop ringing
but it does mean that the inductance
must lie within a particular range – it
must be high enough in value so that
it will perform its job as an inductor
in a switchmode circuit but it must
be small enough in value so that the
supply rail is quite stable by the time
10 clock periods have passed. We’ll
talk about this more a little later.
The dV sensing comes under the
block entitled “battery full detection”
in the diagram of Fig.3.
As already noted, the TEA1100
does not compare the battery voltage
to a static reference. Because it is
a dynamic process, the monitored
input voltage need only be between
0.385V and 3.85V. This is fed to pin 7
which is labelled “VAC” for Voltage
ACcumulator.
The way it works is like this: The
VAC voltage is sampled at a rate equal
to the clock frequency divided by 216.
Each VAC voltage sample is digitised
and stored in a register with a quoted
resolution of 12.5 bits. At the time of
the next sample, the stored value is
converted back to an analog voltage
and compared with the voltage on
the VAC pin.
If the VAC voltage is higher than
the stored value, then this new value
is digitised and stored in the register,
overwriting the previous value. If not,
the previous value remains in the
register. The circuit then checks for a
1% drop as we mentioned before and
if found, switches the circuit to trickle
mode and flashes the LED (connected
to pin 15) to indicate that the batteries
are fully charged.
This clever mix of analog and digital
circuitry results in a dynamic process
which takes the battery’s physical
characteris
tics into consideration.
Since no two nicads charge up to
exactly the same voltage, this relative
method provides accurate “full” detection for all cells, regardless of their
final voltage. Incidentally, much the
August 1994 7
VP
12
Vref
10
VS
6
NTC
3
Rn
11
IB
5
CP
9
Vr1
SUPPLY
GND
16
Vhigh
PROTECTION
Vr3
MAINS
ON
RESET
V
In
LSP
Iref
>
t
AO
2
A2
Vr2
Vlow PROTECTION
Vr4
LS
4
A1
>
>
OSC
DISABLE TIME OUT
>
R
s+h
BATTERY
FULL
DETECTION
VAC
7
PWM
1
PWM
&
R
TIME
OUT
PROTECT
>
LED
15
R
1/10
OSC
TO
PWM
:1:2:4
PRESCALER
COUNTER
CONTROL
CURRENTLESS SENSING
AUX PULSES
13
OSC
8
PR
14
SYNC
Fig.3: the block diagram of the TEA1100. This complex chip senses the small
drop in voltage which occurs at the end of charge for nicad & NiMH batteries, so
that the charger can be automatically switched off.
same monitoring method was used in
the “Fast Charger for Nicad Batteries”
featured in the January and February
1991 issues of SILICON CHIP.
The beauty of the dV sensing system is that the VAC input (pin 7) can
be anywhere between +0.385V and
+3.85V. This means that the VAC
resistor divider network can be the
same whether you wish to charge two
or 10 cells, or any number of cells in
between. To satisfy this condition
in the circuit of Fig.1, R14 should
be 47kΩ while R15 should be 10kΩ.
C8, the input filter capacitor, can be
10µF 16VW. Note that to satisfactorily
charge 10 cells, you will need an input
voltage of at least 22V DC when in
switchmode because the maximum
pulse duty cycle is 78%.
The above is based on an overvoltage level of 1.7V/cell and a nominal
battery voltage of 1.2V/cell.
The VAC input has four voltage
thresholds which determine the chip’s
behaviour. Firstly, below 0.3V, the IC
assumes a short circuit (crook) battery
and switches to trickle charge mode;
above 0.385V and below 3.85V, the IC
uses the dV voltage detection method
8 Silicon Chip
to determine the charge state; and
finally, above 4.25V, the IC assumes
open circuit or no batteries present
and switches off. The impedance of
this input is greater than 200MΩ.
Note too that for charging just one
or two cells, the VAC input (pin 7) can
be connected directly to the cell(s).
Output voltage
This brings us to an important
feature of the Fast Nicad Charger published in May 1994 and one which has
caused confusion to many constructors of this circuit. Since the circuit
relies on dV sensing to end the fast
charging mode, it goes without saying
that it will not work unless it is actually charging cells. If you attempt to
test the circuit without a nicad battery
load, it will switch off.
Our testing instructions for the
above circuit would have added to
this confusion by referring to an open
circuit output voltage test. The point is
that you cannot test the charger’s output voltage unless cells are connected.
If you attempt to simulate the presence
of cells with a large electrolytic capaci
tor, the output voltage will rise until
pin 7 reaches +4.25V whereupon the
circuit will switch off.
In fact, the circuit of May 1994 does
not even need the switch to select between two and four cells. The switch
setting for two cells can be omitted and
then circuit will happily charge two,
three or four cells in series without
further modifications.
In admitting this mistake, we can
only plead that it only become obvious after close reading of the copious application information which
Philips has made available on the
TEA1100.
0.5% detection
In some cases, such as “fastcharge”
nicads and NiMH cells, a dV of 0.5%
is more appropriate due to the higher
level of input charge current they can
tolerate. This IC can provide charge
rates up to an incredible five times the
battery capacity or “5C”. An example
of this would be charging a racing pack
in about 15 minutes.
This increased sensitivity can be
easily achieved by inserting a zener
diode of about half the battery voltage
into the sensing resistor string. An example of this can be seen in Fig.5. The
zener diode is selected to be about half
of the fully charged battery voltage,
based on a level of 1.7V/cell.
Protection
Apart from the active protection features already mentioned, the TEA1100
features undervoltage shutdown and
temperature sensing with a thermistor
input circuit.
The first of these, the undervoltage
shutdown, activates when the supply
voltage falls below 5.25V. In this case,
the IC goes into a “power down” mode
in which it becomes nonactive and
draws around 35µA (45µA maximum).
The second form of protection
involves a negative temperature coefficient (NTC) thermistor to monitor
the temperature of the battery during
charging. This feature wasn’t included in our May 1994 project to keep
the construction simple. In practice,
where this feature is used, the therm
istor is incorporated into the battery
pack and is automatically connected
when the battery is put on charge.
The temperature monitoring feature
is recommended for batteries which
need to be recharged as soon as they
have been removed from their load.
The classic example of this is
1200mA.h racing packs for electric
model aircraft and cars. The drain on
these batteries is very high  often tens
of amps or more – and so they will be
quite hot (or even stinking hot!) when
they are removed from the load.
The danger is that if you fastcharge
a hot nicad battery, you can damage it.
The temperature protection provided
by the TEA1100 prevents fast charging
from occurring while the battery temperature is outside the specified range.
The NTC thermistor is featured on the
circuit of Fig.1 and is connected to pin
3. If the thermistor is not required, it
can be omitted from the circuit, together with R11.
Fig.4: the
voltage
characteristic
of a 2cell
nicad battery
back during
charge. If
charging
continues
beyond the
droop in
voltage, cell
damage can
occur.
VOLTAGE (V)
For example, for a 6 cell pack, the
maximum voltage is 6 x 1.7V = 10.2V,
so a zener diode of 5.1V would be
suitable. The maximum voltage level
the VAC input will now see is 5.1V, so
the input resistor divider must now be
recalculated accordingly. R14 on Fig.1
could then be reduced to 22kΩ.
CHARGE TIME (MINS)
capacitor connected to pin 13.
This timeout period is usually set to
about 125% to 150% of the expected
fast charge time but in critical high
charge rate applications, you can set
it to the expected charge time (100%).
In practice, the timeout period
should only be set by adjusting the
capacitor (C at pin 13), as varying the
reference resistor will change other
circuit parameters.
Design example
The easiest way to understand how
to use this IC is to go through a design
example, using the circuit of Fig.1.
This way, you’ll get an idea of what
has to be done and the order in which
you have to do it.
Let’s say we wanted to design the
timeout circuit to run a charger which
will charge up a set of four nicad cells
in one hour. If we use the 150% rule,
then our timeout period, tTO, will be
Timeout counter
Finally, there is the backup protection of a timeout counter, which
automatically shuts down the charger
after a time equal to 226 times the clock
period, has expired. The clock period
is determined by the reference resistor
connected to pin 10 and the timing
Fig.5: a zener diode equal to
half the fully charged battery
voltage can be added to the
circuit to enhance the dV sensing
capability so that it will detect a
drop of 0.5%.
1.5 x 60 mins = 90 mins.
The timeout period is determined
by the following formula:
tTO = 226 x Tosc x p
where Tosc is the clock period and
p is a prescaling factor which you
can program to be either 1, 2 or 4,
depending on how you connect pin
8. By leaving pin 8 open, you set the
prescaling factor to 2. Connecting it
to pin 6 sets it to 1 and pulling pin
8 to ground sets it at 4. The beauty
of this system is that it allows you to
have three different charge periods
without having to change the timing
components.
For our example, let’s connect pin
8 to pin 6 to set the prescaling factor
(p) to 1. The oscillator frequency (1/
Tosc) now needs to be 12.4kHz (ie, Tosc
= (90 x 60)seconds/226). As mentioned
be
fore, this frequency is set by the
time constant formed by the reference
resistor Rref (R13) and the oscillator
capacitor Cosc (C7) based on the following equation:
Tosc = 0.93(Rref x Cosc)
Now Rref is chosen to be within the
range of 12.5kΩ and 125kΩ based on
the necessary charge current. In our
example, let’s assume that the resistor
is 27kΩ. Plugging this value into the
above equation gives a value for Cosc
of .0032µF which we can quite happily
round to .0033µF.
Charge current settings
OK. Let’s say that we wish to charge
our batteries at a fast rate of 700mA.
R4 and R8 are used to set the current.
R4 should be a 5W type. You have
some leeway in picking the value
of this resistor, so long as its value
August 1994 9
Using the TEA1100 fast nicad charger IC
results in a voltage drop of between
50mV and 200mV when the circuit
is in fast charge mode. You can work
out a suitable value for R4 from the
following equation:
Vcs = Ifast x Rcs
where Ifast is the fast charge current
and Rcs is R4. In our design example,
0.1Ω will give us 70mV which is within the desired range. R8 is referred to
as the fast charge current set resistor
Rfc and it can be calculated from the
following equation:
Rfc = (Ifc x Rref x Rcs)/1.25
where Ifc is the fast charge current rate,
Rref is the 27kΩ reference resistor R13,
and Rcs is the 0.1Ω current sensing
resistor R4.
By using this equation, we get a value for Rfc of 1.512kΩ, so a 1.5kΩ 1%
resistor will be perfect for R8.
determined by the worst case ripple
current at the trickle current setting
and follows this equation:
Lmin = Vo’max(1delta)Tosc/2Iav
where Vo’max is the maximum battery
voltage plus the forward diode voltage drop. For four cells, this works
out to be 6.8V + 0.7V = 7.6V. This is
based on the fact, that the maximum
voltage per cell will be 1.7V; “delta”
refers to a charge current duty cycle
of 50%.
So, using the above equation, we
get a minimum inductance value of:
Lmin = 7.6 x (10.5) x 80 x 106/2 x
0.35 = 434µH.
Hence the inductor can be anywhere
between 5mH and 434µH. Why not go
for the perfect compromise and settle
upon 2mH?
What inductor?
Winding an inductor presents many
constructors with a problem since they
don’t have access to the necessary
information involving readily available toroids. Indeed, a comprehensive
article on this subject alone could
take many pages. However, to keep it
simple, we’ll just deal with the three
readily available iron powder toroids
made by Neosid and available from
Altronic Distributors and Jaycar Electronics.
The general formula for inductance
using these toroids is:
n = 1000 √(L/AL)
where n is the number of turns, L is
the inductance in millihenries (mH)
and AL is the inductance factor of the
particular core. For the smallest core,
Neosid 1773222, 14.8mm OD, AL is
44; for the medium core, Neosid 1774222, 33mm OD, AL is 59; and for
the largest core, Neosid 1774522,
44mm OD, AL is 116.
Having calculated the number of
turns to obtain the required inductance
on the core of your choice, you then
must check whether it is likely to be
saturated at your proposed operating
current. To do this, we calculate the
core energy with the following formula:
E = LI2
where E is measured in joules, L is
the inductance in henries and I is the
current in amps. For the three cores
Earlier on, we mentioned that with
switchmode operation, you have to
be careful in selecting the value of the
inductor – too low a value will result
in the circuit not working efficiently
and too high a value will result in
the dV sensing circuitry picking up
remnants of the switching voltage due
to the “ringing” effect of the inductor.
For this dV sensing to work, the induc
tance current should have decayed to
zero within nine clock cycles. So the
maximum inductance is set by the
following equation:
Lmax = 9 x Tosc x (Vo + Vf)/Io
where Tosc is the period of the clock
frequency, Vo = the flat battery voltage
(around 1V per cell) plus the voltage
drop across the fast recovery diode
D2 (usually taken as 0.8V) plus the
voltage across the current sensing
resistor. Io is the average current
through the inductor which is a fast
charge current.
In the example we’ve been working
through, this would give us a maximum inductance of:
Lmax = (9 x 80µs x 4.8V)/700mA
= 5mH.
This assumes four cells with a flat
voltage of 1V each, plus the 0.8V
drop for the fast recovery diode, D2.
The 80µs figure is the clock period at
12.4kHz.
The minimum inductance value is
10 Silicon Chip
Winding an inductor
under discussion, the maximum
stored energy levels are 0.71mJ for the
14.8mm OD core; 5.1mJ for the 33mm
OD core; and 16mJ for the 44mm OD
core (OD stands for outside diameter).
If the core you have chosen will
saturate at the required current and
inductance, then you will have to use
a bigger core.
One final point must be covered
here before we leave the subject of
inductors and that is that the actual
current flowing in the inductor referred to in the formulas above is the
pulse current; it is not the charging
current. Typically, the pulse current
will be twice the average charging
current.
Trickle charge
When in trickle charge mode, the
TEA1100 continues to pulse the
battery with the fast charge current
but at a much lower duty cycle. As it
seems with just about everything else
on this IC, you have a choice of one
of two ways to set the trickle current,
depending on how you connect pin
11, designated the “Rn” input.
The first method is to leave pin
11 unconnected. In this case, the
repetition and duration of the trickle
current pulses is determined by the
chip itself.
The repetition rate is set as 214 x
tTO = 330ms in our example. The duration time is set to 0.75 x 29 x Tosc,
where Tosc is the clock period. In our
example, this works out to be 31ms.
This also gives us a duty cycle for the
trickle current of 9.4%.
The average trickle charge current
based on this duty cycle is set by the
following equation:
Itrickle = Ifc/2 x duty cycle = 30mA.
The second method is to set the
average trickle current yourself by
connecting a resistor Rn to pin 11. The
rule for this resistor is that it must be
within the range of 25kΩ to 250kΩ
and must be greater than the reference
resistor Rref.
The new trickle current equation
looks like this:
Itrickle = Ifc x (Rref/Rn) x duty cycle
With Rn equal to Rref (27kΩ), the
trickle current is 60mA and 7mA with
Rn equal to 250kΩ.
Linear design example
Let’s say that we want to charge
three “AA” cells in one hour, using
the circuit of Fig.2. The required
TABLE 1
Number of cells to be
charged
Transformer
secondary voltage
(V RMS, full load)
Capacitor value
(µF/A)
Capacitor voltage
rating (VDC)
2
7
4000
16
3
9
3000
25
4
11
2400
25
5
13
2000
35
6
15
1700
35
7
17
1500
40
8
19
1300
40
9
21
1200
50
10
23
1100
50
current is based on the following
equation:
Iout = (A.h x 60 x 1.4)/charge time
(mins)
So for a 600mA.h battery, the current
would need to be:
Iout = (600mA.h x 60 x 1.4)/60
= 840mA
In case you’re wondering why it just
isn’t 600mA, the reason is that there
are substantial losses in the battery
when charging takes place, so you
need to increase the charge current
by 40% to make up for these losses
(ie, heat etc.)
At this current, the main pass diode
D5 can still be a 1N4004 but the transistor will have to be something like
a TIP32C, a device which can handle
the current and the power dissipation.
And it will need a heatsink.
Power dissipation
Table 1 gives the required transformer secondary voltage and the
suggested capacitance per amp of
required current and voltage rating of
the filter capacitor. Now for our design
example, to charge up three cells, we
need a transformer secondary voltage
of 9.1V. The power dissipation can be
found from the following equation:
Pdiss = 1.3 x Iout x (Vsec  2.0)
= 1.3 x 0.84A x (9.1  2.0)
= 7.8W
Basing this on a maximum temperature rise of 55°C above ambient, the
required heatsink will have to be better
than 55°C/7.8W or 7°C/W.
Now obviously, this is quite a bit of
power being wasted so you will have
to decide whether the need for a linear
charger outweighs the benefits of the
switchmode alternative.
OK, so we’ve determined the cur
rent we require and now we have
to tell the TEA1100 what we want.
To do this, we again start with a
reference resistor of 27kΩ, just as
for the switchmode version. Next,
we have to choose the main current
sensing resistor (R1) and again, for
our charge current of 840mA, a 0.1Ω
5W resistor will give us 84mV which
is good enough. Remember that this
resistor doesn’t set the current on its
own. This is done by resistor R3 on
the circuit.
This resistor is determined by the
following equation:
Rfc = (Rref x Rcs x Ifc)/1.25
and in our design example, R3 becomes:
R3 = (27kΩ x 0.1Ω x 0.84A)/1.25
= 1.814kΩ
A value of 1.8kΩ will be close
enough.
Trickle charge
As with the switchmode version,
the trickle charge current can be set
to just about anything you want. By
connecting the prescaling pin (pin 8)
to pin 6 and leaving resistor R7 open
circuit, the TEA1100 will automatically set the trickle charge current to
1/20th of the fast charge rate. In our
example, this would work out to be
42mA.
Now this may be too high, in which
case, you can change the trickle current by connecting resistor R7 from
pin 11 to ground. The relationship
between this resistor and the trickle
charge current is set by the following
equation:
R7 = (1.25 x Rfc x 0.094)/(Itrickle x
Rcs x p)
Let’s say we wanted the trickle current to be 15mA instead of 42mA. By
working through the above equation,
resistor R7 would need to be:
R7 = (1.25 x 1.8kΩ x 0.094)/(15mA x
0.1Ω x 4) = 35.2kΩ.
A 36kΩ 1% resistor will get you
fairly close to the mark.
You should note a couple of things
here. Firstly, we’ve had to change the
prescaling factor to four. Now the
reason for this is that the prescaling
factor not only works on the timing
circuitry but also on the charge current ratio; that is, the ratio of the fast
charge current to the trickle charge
current.
With a prescaling factor of one (pin
8 to pin 6), the maximum ratio is 20:1.
For a prescaling factor of two (pin 8
open circuit), it is 40:1 and for four (pin
8 to ground), it’s 80:1. Now for our design we want a ratio of 840mA/15mA =
56:1. Setting the prescale to either one
or two won’t get us this value so we
have to go to a prescale factor of four.
The reason for the change is that
if resistor R7 is greater than twice
the reference resistor R6, then the IC
automatically selects half of the fast
charge reference current. This gives
us our maximum 20:1 with a prescale
of one, 40:1 with p set to two and 80:1
with p set to four.
In most situations, resistor R7
should not be less than the reference
resistor. If by working through the
equations, you find that R7 is less than
R6, either change the prescaling factor
or remove the resistor from the circuit
altogether.
Timeout counter settings
The last thing to do is to set the
timeout period and since we have
already set the reference resistor R6
to 27kΩ, the only component value
which affects the time is capacitor
C4 and this can be determined by the
following equation:
C4 = (60 x timeout)/0.93 x Rref
x p x 226
Getting back to our design example, let’s say that we’re happy with a
trickle current of 42mA and we want
the timeout period to be 60 minutes.
Capacitor C4 then works out to be:
C4 = (60 x 60)/(0.93 x 27kΩ x 1 x 226)
= .00213µF
(.0022µF will be close enough).
Note too that this capacitor value
will change if you change the pre
scaling factor as in the above example
where we looked at a trickle current
SC
of 15mA.
August 1994 11
