Variable Frequency Drive Using IR215x

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

DESIGN TIP International Rectifier • 233 Kansas Street El Segundo CA 90245 USA

DT 98-1

Variable Frequency Drive using IR215x Self-Oscillating IC’s


By John Parry
Purpose of this Design Tip

Applications such as high quality Electronic Ballast and Power Supply circuits often require continuous switch
frequency control over a specific range. Alternatively, the need may arise to dynamically select one of several
discrete drive frequencies using digital control signals. Both IR215x self-oscillating control IC’s and IR5xHxxx
Hybrid circuits are ideal for use in such areas and offer several advantages over traditional bridge drive methods.
This design tip describes operation of the integral oscillator and shows how simple techniques and inexpensive
peripheral circuitry may be utilized for variable frequency drive.

Topics Covered

1. Operation of Typical Fixed Frequency Circuit


2. Dynamic Frequency Selection by Switched Capacitor Method
3. Continuous Frequency Control by Offset Voltage Method

1 Operation of Typical Fixed Frequency Circuit

In perhaps the simplest implementation, self-oscillating control IC’s are configured with a single resistor and timing
capacitor in the circuit shown by figure 1a. This circuit exhibits 50% duty cycle and fixed frequency determined by
equation 1, in which the 75 Ohm term accounts for resistance of the oscillator output pin, Rt. In later sections of this
design tip, bootstrap components D1,Cb, power switches and gate resistors Ra and Rb are omitted for clarity. Figure
1b shows the operating waveforms to be expected using the IR2153 control IC. The Rt output pin behaves as a
voltage source switching between Vcc and Ground. Rt is pulled low as the Ct pin rises to a threshold of 2/3 Vcc and
switches high when Ct -falls to 1/3 Vcc. Note that Ct switching thresholds are set at a fixed proportion of Vcc and
derived from a ratiometric divider network within the IC.

Fig 1a. IR215x in Typical Fixed Frequency Application 1


f =
1.38( R1 + 75)C1 (1)
IC Bias Db +Vbus

C3 Cb
Vcc Vb Fig 1b. IR2153
R1 Ra Waveforms
Rt HO
Rt
Ct Vs
Ct
COM LO
Rb
C1
IC1 LO

Td

HO
2 Frequency Control using Series/Parallel Switched Capacitor Method

2.1 Operation

One of two or more running frequencies may be easily selected using either of circuits discussed in sections 2.2 and
2.3. In both cases, a small signal NPN transistor is used to add or bypass one of two capacitors in series or parallel.
This results in a change in frequency by altering the effective capacitance seen by the Ct node of the IC.

2.2 Parallel Capacitor Switch


Fig 2d. Parallel Capacitor Switch
The circuit of figure 2d shows how a small-signal IC Bias
transistor may be employed to connect an auxiliary IC1
capacitor to the oscillator on command. When the
transistor is in the off-state, diode D1 is blocking and Vcc Vb
C2 is out of circuit, so the oscillator frequency is high. R1
Rt HO
When Q1 is turned on, it carries charging current for
C3
C1 and the diode D1 provides the discharge path. This Ct Vs
effectively places C1 and C2 in parallel, increasing C1
the capacitance seen at the Ct node and reducing COM LO
D1
switch frequency. Resistors R2 and R3 should be
chosen such that Q1 is in saturation when the ‘f2’
C2
select input is high. Dividing the control signal in this
Q1 R3
manner is recommended over a single base resistor
because noise immunity of the control signal is f2 Select
improved. This is especially important in cases where
the control signal ground is distant from the COM pin R2
of IC1. Note the star point return to COM.

1 1
f1= f2≈
. ( R1 + 75)C1
138 1.38( R1 + 75)(C1 + C 2)

2.3 Series Capacitor Bypass Fig 2e. Series Capacitor Bypass

The circuit of figure 2e shows how a small signal IC Bias


transistor may be employed to bypass one of two IC1
capacitors connected in series. When the transistor is
in the off-state, diode D1 is blocking and timing Vcc Vb
R1
capacitors C1 and C2 are in series connection so the Rt HO
switch frequency is high. In the on-state, the transistor C3
carries charging current for C1 and the diode D1 Ct Vs
C2
provides the discharge path. This effectively removes COM LO
C2 from circuit, increasing the capacitance seen at the C1
D1
Ct node and reducing switch frequency. Resistors R2
and R3 should be chosen such that Q1 is in saturation
when the ‘f2’ select input is high. Note the star point Q1
R3
return to COM implied in the diagram.
f2 Select
C1. C 2 1
f1= f2≈
. ( R1 + 75)(C1 + C 2)
138 1.38( R1 + 75)C1 R2
2.4 Circuit Variation

If desired, both diode and bipolar transistor may be replaced with a single, N-channel MOSFET since the internal
body drain diode will serve the function of D1. In this case it may be necessary to account for output capacitance
(Coss) of the switch when in the off sate. MOSFET output capacitance is highest when the drain to source voltage is
low, as in this application and therefore it is preferable to select the smallest device available. If a small signal
bipolar transistor is used as shown in the circuits proposed here, output capacitance of the switch can usually be
neglected. Various combinations of series and parallel switch may be deployed as necessary to provide the required
number of selectable run frequencies.

2.5 Limitations of Switched Capacitor Circuits


Fig 2f. Effect of Q1 parametric
2.5.1 Repeatability spread on circuit of fig 2e.

Low Vbe(th)
Both series and parallel methods give good results if one of
High Vbe(th)
two distinct switch frequencies must be dynamically
selected. However, if continuous frequency control is f1
needed, some limitations appear. Continuous frequency
variation using either circuit requires that the transistor be
operated in the linear mode between limits. Unfortunately, f(osc) } Low Gain
oscillator frequency in this region is a function of both small- } High Gain
signal gain and switch threshold as portrayed in figure 5f .
Transistor selection or trimming may help increase
repeatability in volume manufacture, however in general,
f2
neither are appealing options. In such cases, continuous
frequency adjustment by offset voltage control may be a
preferable alternative. This topic is covered in section 3. 0.53 0.54 0.55 0.56 Vbe (Q1)
2.5.2 Duty Cycle Disturbance and Possible Effects

During the transition when Q1 enters the linear region and frequency starts to change, the duty cycle at the output
will temporarily deviate from its 50% nominal value. This statement applies to both circuits of figure 2d and figure
2e. When C2 starts to switch either in or out of circuit, the average potential across C1 must change to compensate
for the reduction or increase in the average voltage across C2. The charge required to do this can only be acquired by
inequality between Rt on and Rt off time, resulting in Ct asymmetry and subsequent duty cycle change at the output.
Fortunately, this effect is temporary and in the typical case,where timing capacitors C1 and C2 are either equal or of
a similar magnitude, duty cycle deviation will be minimal and remain for only a few switch cycles. If however, the
ratio between C1 and C2 is high, then many switch cycles may be required for the average voltage on the larger of
the two capacitors to be restored and for the duty cycle to return to its nominal value.

In a half bridge drive, duty cycle variation changes the average voltage at the output, in the same manner as would a
PWM circuit. The result is a low frequency transient superimposed on the much higher switch frequency. Certain
types of load, such as the series resonant LC circuit used in electronic ballasts may react unfavorably to the resulting
voltage transient if it is sufficiently extreme. If any type of resonant load is deployed, circuit parameters should be
verified in as the linear region starts. In particular, check current in magnetic components for saturation with the
equipment at expected operating temperature.
3 Continuous Frequency Control by Offset Voltage Method

If the oscillator frequency must be continuously variable between two limits and programmable by an external
voltage sourced signal, the method described here may be applicable. The concept is introduced below and circuits
with a more practical bias follow later in the document.

3.1 Explanation of Offset Voltage Control


Fig 3a. Offset Voltage Control
3.1.1 Test Circuit Test Circuit
IC Bias
For the purpose of this description, consider the ‘test IC1
circuit’ portrayed in figure 3a. This circuit serves only
to aid explanation and is not offered as a practical Vcc Vb
implementation. Note the lower plate of the timing R1
Rt HO
capacitor, C1 is attached to the output of a signal
generator which produces a square wave output. Ct Vs

The signal generator output (Va) is synchronized in C1 COM LO


both frequency and phase to the Rt pin of IC1 and set
for positive peak of V1 volts and a negative peak of
V2 volts with respect to COM. The signal generator 0 Va
output sums with the lower plate of C1, therefore we Sync
shall refer to the peak-to peak value of Va as the offset
voltage Vos, where Vos=V1+(-V2). This
configuration will yield the waveforms of fig 3b. Square wave signal generator.
1. Va frequency & phase locked to Rt out
2. Va output peak to peak = Vos Volts
3. Signal generator must be able to source and sink
current through C1 without distortion

3.1.2 Operation of Test Circuit

Assuming the Rt output has negligible rise and fall


times, as Rt switches to Vcc, the signal generator Fig 3b. Results from circuit of Fig3a.
output immediately ‘adds’ 1xVos to the lower plate of Vcc
C1 and consequently the potential seen at the Ct pin
also increases by 1xVos. This reduces the total charge
(and time) required to raise Ct from 1/3*Vcc to the
2/3*Vcc switch threshold set internally by IC1. - Rt
When Ct reaches the upper threshold, Rt immediately
switches state to 0V and the signal generator output
Va switches from V1 (+) to V2 (-). The difference 0V
(Vos) is now ‘subtracted’ from C1 lower plate and
consequently, the Ct pin. Again, the total charge Ct 2/3*Vcc
required to take Ct between thresholds is reduced. Vos

Vos
1/3*Vcc
The oscillator frequency is therefore higher than
would be the case in the typical RC configuration of Vos
Va V1
figure 1a. Furthermore, since charge and discharge
0V
times are affected equally, the oscillator duty cycle V2
remains fixed at 50%.
Vos=V1+(-V2)
Equation 3 below shows that the free run frequency is a function of both Vcc and the peak-to-peak offset voltage,
Vos. Although in this example, V1 and V2 are equal, this is not a requirement for the circuit to function correctly as
the run frequency is independent of Va displacement relative to ground.

1
3.2 Implementation of Offset Voltage Control f =
 Vcc 
−2 C1( R1 + 75). Ln   (3)
In figure 3a, when the Rt output is high, the signal
 2Vcc − 3Vos 
generator sinks charging current for C1 and holds
Va=V1. Conversely, when Rt is low, the signal
generator sources current and holds Va=V2. In both
cases the signal generator plays a passive role and Fig 3c. Offset Voltage Control
serves only to clamp the lower plate of C1. Equivalent Circuit to fig 3a.
IC Bias
IC1
This means that the signal generator shown in figure
3a can be reduced to the simple bipolar voltage clamp Vcc Vb
depicted in the equivalent circuit of figure 3c. R1
Running frequency is dependent only on Vcc and Vos Rt HO
and both can be controlled, therefore several
Ct Vs
derivatives of this simple circuit are possible.
COM LO
C1
3.3 Frequency Control by Vcc Adjustment

In the circuit of figure 3d, the offset voltage is fixed


using a small signal zener diode ZD1, where; +
Va waveform
Vos=Vz-(-0.6). Fixing the offset voltage in this way -V2 V1 Va
+ V1
engineers a relationship between oscillator frequency 0
V2
Vos

and Vcc so the IC bias supply doubles as the control Diodes D1 / D2 are assumed ideal. i.e.;
Vf=0, C=0, Trr & Trf = 0
signal itself.

Substituting arbitrary values for C1,R1 reveals the oscillator response to Vcc changes as shown in figure 3e. Note
that switching starts at a high frequency as Vcc passes the positive going under-voltage lockout threshold
(UVLO+). As Vcc continues to rise, the run frequency falls until the internal zener clamp of the IR2153
avalanches preventing further change. An external zener diode can be used to modify the Vcc limit and fix the final
run frequency as required. If the Vcc start-up waveform rises in a predictable manner or can be made to do so, this
technique can be very useful as the basis for frequency sweep in a warm-start fluorescent electronic ballast.

Fig 3d. Frequency Control by Vcc Adjustment


Fig 3e. Oscillator Vcc dependance
IC Bias C1=1nF,R1=27K,Vz=1.8V using cct. 3d.
IC1
160000
Vcc Vb
R1
Rt HO 120000
f(osc) Hz

Ct Vs
80000
COM LO
C1

40000

Va waveform
ZD1 Va 0
8
9
10
11
12
13
14
15
16

Vz
0 Vos
0.6

Vcc V
Zener diode ZD1 must be low power type
3.4 Frequency Control by Vos Adjustment - A Simple Non-Linear VCO

Fixing Vcc and adjusting the offset voltage will also cause frequency change. This allows a simple, non-linear
voltage controlled oscillator to be easily constructed using the circuit of figure 3f. In this example, the negative
clamp voltage, V2 is set by the forward voltage drop of D1. The positive clamp voltage, V1 is set by a control
voltage applied to the base of Q1. This small signal PNP transistor behaves as an emitter follower and provides
sufficient current gain to allow the control voltage to be driven from a comparatively weak source. Capacitor C2
is presented between base and collector of Q1 and helps to provide a solid V1 clamp and filter noise voltage on
the control input. C2 should be of the order of 100pF or less. If a low-impedance control voltage source is
available, then Q1 may be replaced with a single low voltage silicon diode of the same type as D1.

Fig 3f. Frequency Control by Vos Adjust Fig 3g. Frequency Control by Vos Adjustment
Vcc=15V,Rt=24K,Ct=1nF Circuit of fig.3f
IC Bias
IC1
100000
Vcc Vb
R1 80000
Rt HO

F osc (Hz)
Ct Vs 60000

COM LO
C1 Va 40000

20000
D1
Va waveform
Q1 0
2.5

1.5

0.5
2

0
Vctr+0.6
C2 0
0.6
Vos

Vctr (V)
Q1 = small-signal switching
+
Vctr Vctr = Control signal input

3.5 Limitations of Offset Voltage based Circuits

3.5.1 Frequency Error due to Clamp Non-Linearity.

Ideally, the bi-directional voltage clamp should source/sink current at precisely at V1/V2 clamp irrespective of the
current passed and should exhibit infinite impedance between these limits, however in practice, this is not the case.
Although similar limitations will exist to some degree in any implementation of offset voltage control, for the purpose of
example, imperfections in the zener diode clamp of figure 3d are highlighted as follows;

a. The diode will pass a reverse leakage current prior to the zener limit.
b. There is a recovery time from both forward and reverse conduction (inequality will also change duty cycle)
c. The diode has a small equivalent parallel capacitance.

Limitations a,b and c cause distortion of the Va waveform in the form of “rounding” and reduced dv/dt during Rt
transitions. The overall effect is to introduce small errors in equation 3. This presents as a reduction in frequency change
with a given change in Vcc, however, some steps can be taken to improve the results. Choosing a small signal zener
diode with a power rating of 250mW or less is highly recommended and will generally reduce errors in the
frequency/Vcc performance to an acceptable level. Alternatively, the zener diode may be replaced with a number of
small signal silicon diodes in series configuration to produce the required clamp threshold by summing Vf drops.
3.5.2 Frequency Error due to Clamp Threshold Changes.

In both implementations of offset voltage control presented here, the clamp thresholds V1 and V2 fix the offset voltage
Vos, which in conjunction with Vcc determines the run frequency. In practice, both clamp thresholds will vary slightly.
For this example, we will again consider limitations of the zener-diode based circuit of figure 3d, though the underlying
principles are equally applicable to any implementation of offset voltage control;

d. Both Vz and forward conduction threshold Vf are dependent on temperature.


e. Equivalent series resistance is not zero and so V1 and V2 will vary slightly with charge/discharge current.

Limitations d and e introduce small errors into equation 3, however, since temperature coefficients of Vz and Vf are
reasonably predictable, the overall frequency error can be easily calculated. In general, since the charge/discharge
current is very low, series resistance of the clamp does not cause significant error. Variation in leakage current discussed
in 3.5.1 represents a more significant source of error in this regard.

3.5.3 Noise Considerations and PCB Layout Requirements.

The oscillator trigger thresholds are a function of Vcc referenced to the COM pin. Since the Ct pin is noise sensitive, all
circuits connected directly or indirectly to the Ct pin MUST be returned either the Vcc or COM node at the IC. Such
PCB tracks used for this purpose should not be allowed to carry high currents from the load circuit or current from other
noisy sources. This is especially important with offset voltage control is being used since the noise margin is reduced as
Vos becomes an increasing proportion of Vcc. Tracks returning to COM should be as short and direct as possible.

You might also like