Дипломная работа - Проектирование 10 кВт преобразователя

Подождите немного. Документ загружается.

Figure 27 : Scetch of chassis

The weight and volume for an enclosure of this solution is highly dependant on the

actual heat sink whose figures are estimated from the dissipated energy and the different

thermal resistance of the different components as illustrated in figure 26. However this is

only for the top-side transistors since the two sides will be mounted on separate heat sinks

along with one out of two rectifiers. The transistors used in the calculations for thermal

resistivity are the IRG4PSH71UD with a built-in diode, the rectifier is from Semikron

and the thermal resistance values are given in their datasheets.

Figure 28 : Thermal resistivity estimation

39.0//////

531

≈

sarectsasasaTOT

RRRRR

θθθθθ

(2.7.1.1.1)

“//” is equal to parallel coupling

The internal thermal resistances can be found in the datasheets while the thermal

resistance of the heat sink has to be calculated for each one of the components and then

joined using a parallel coupling as seen in 2.7.1.1.1. The temperatures given where the

ambient temperature T

= 70°C and the maximum junction temperature T

= 125°C. This

way of calculating is done because some of the components have different thermal

resistance and dissipated energy and then requires different cooling than the next one.

The amount of cooling and space on the heat sink for each component will be determined

the easiest way with their specific thermal conductivity

required from the heat sink, for

instance, the rectifier requires a thermal conductivity of about 0.59 which is about 23% of

the total 2.63 R

-1

. The layout of the heat sink used in the sketch of the chassis, placing of

the components and the actual distribution of the cooling area on the heat sink can be

viewed in figure 27.

Figure 29 : Heatsink layout

The heat sinks used in the model illustrated in figure 27 is from H S Marston, it measures

150x200x40mm and is optimized for natural convection having a thermal resistance of

0.4°C/W which is sufficient. However, this is at sea level only; an increase in altitude will

increase the thermal resistance significantly as explained before. The weight of 1.44 kg

for each side gives a rough estimation of the weight of the chassis since the steel sheets

forming the rest of the enclosure will weight much less.

Using the heat sinks as two sturdy walls and steel sheets of 250x150x1mm

as the other

walls an internal volume of 7.5dm

is reached, estimated to fit all the devices, circuitry,

filters and the capacitor-bank. The bottom and top plate can be made up from similar

steel sheets but the bottom should be slightly thicker to provide better support, thus

doubling the thickness. A 250x200x2mm and a 250x200x1mm along with the other walls

The conductivity is the inverse of the resistivity

Width x height x thickness,

made of aluminum weights about 610g which when added with the heat sink the whole

enclosure weights 3.5kg.

2.7.1.2. Interior heat sink

When mounting the heat sink inside the chassis the air circulation would be very limited

if not increased via an external source, the ambient temperature in the chassis along with

the junction temperature would probably rise to a critical level and the system would

eventually shut-down or burn. However even with a small fan airflow can be produced

through the interior raising the performance of the heat sink, often many times over as

illustrated in the figure below.

Figure 30: Thermal resistance factor vs. airflow

Where F is a factor of which the thermal resistance of a heat sink is reduced depending on

the airflow across the heat sink. The curve resulting in a horizontal line is mostly due to

the fact that airflow does not increase the radiated heat.

This would allow a reduction in weight due to removal of excess heat sink at the cost of a

few watts of lost effect for the fan and the possibility of fan-breakdown, causing a fatal

temperature build-up. The fan would at best push in air at the temperature up to 70°C and

the slightly warmer air coming out would blend with the ambient air and if the

compartment is not too small it would not raise the environment temperature

significantly. The increased internal temperature has to be included in the calculations

where size of the heat sink is determined. This solution would also suffer in cooling

capacity due to a decreasing air density but it can at least be reduced by serial-coupling

many fans which can create a strong pressure in the chassis by having a smaller outlet.

However it would not be nearly enough to compensate for the loss in pressure at higher

altitudes since even if the convection could be doubled or tripled it initially represents

such a small share of the total the reduction in thermal resistivity would be insignificant.

Most aircrafts and particularly JAS 39 GRIPEN have equipment for providing

pressurized air at a few degrees above 0°C and this can strongly reduce the need for a

large heat sink at the cost of a less stand-alone device. Since the air can contain moist and

particles the air cannot be sprayed directly on a heat sink or the component but is meant

to be ran through a cooling block similar to the one of conventional fluid cooling;

although not nearly as effective due to the difference in specific heat capacity. Here a

failure in the cooling system would inevitable lead to a failure in the inverter as well as

all other equipment depending on the same cooling system. It has to be considered here

whether a loss or a gain in weight is achieved depending on the efficiency of the cooler, if

it requires more weight and power to produce this very cooled pressurized air than

required by a passive heat sink a total gain in weight is achieved which was not the final

task.

2.7.1.3. Internal fluid cooling

A third way of cooling in an aircraft such as the JAS 39 GRIPEN is to use its fuel as a

coolant before it is combusted. The fuel is pumped around and reaches a temperature of

maximum 100°C keeping a cooling bed at the same temperature which is equivalent to a

very large heat sink with the same temperature.

This system is also very dependant on the external cooling system and will fail or become

able to only supply as much smaller load in case of a failure in the external cooling

system. The main advantage here is the reduction in weight but also no mechanical parts.

Designing the chassis also require less consideration since no wind-tunnels has to be

made. In the case of using this system each heat sink in the sketch is replaced with a

cooling bed of similar size although very likely to significantly remove weight of the unit.

2.8. Transistor drives

The design of the transistor drive circuit is another subsystem which deserves attention as

a thoroughly designed driver along with snubber circuits can reduce switching losses

extensively. However, not only the losses but all switching behavior can be altered and

this is wanted as there is no universal perfect switching. All applications require

individual design, for example a highly inductive load such as a motor will cause a large

voltage over-shoot if the switching is too fast. This over-shoot is very unwanted as it may

bring the IGBT outside its safe-operating-area (SOA).

Figure 31 : Descriptive driver internal circuit (IGBT within dashed line)

The main factions to alter for a satisfying behavior are the positive bias voltage, the

negative bias voltage, or the gate resistor. However, the gate resistor can be in parallel

with a diode and a resistor in series as this will change the turn-on and turn-off behavior

independently.

The +V

is often set on a level as close to the maximum gate voltage as possible with

regard to tolerance which can be as much as 10% in some cases. This because a high gate

voltage during the ON-state minimizes the ON-resistance (or ON-state V

saturation

voltage) while a too high gate voltage may permanently damage the device. Respectively

a low -V

will increase the OFF-resistance causing less loss during the OFF-state but

will increase chance of shoot through currents if too small [7].

Setting +V

voltages high (and -V

low) while keeping the gate resistor value constant

will result in a low switching time and thus low switching loss. However, a faster

behavior will as stated before cause unwanted surge currents and voltage spikes over the

devices possibly destroying the device.

A rule of thumb is to set the +V

and –V

values to 15 and -5 respectively while

mainly altering the gate resistor to achieve a satisfactory switching behavior. Below is a

table describing common transistor behavior when changing driver parameters.

Main characteristics +V

rise -V

rise RG rise

CE(sat)

Fall - -

Fall - Rise

off

- Fall Rise

Turn-ON surge voltage Rise - Fall

Turn-OFF surge voltage - Rise Fall

dV/dT malfunction Rise Fall Fall

Current limit value Rise - Fall

Short circuit withstand

capability

Fall - Rise

Radiational EMI noise Rise - Fall

Table 7 : switching characteristics vs. driver dimensions

Using the values recommended by the rule of thumb the maximum current through the

gate resistor can be estimated and then the proper driver circuit can be chosen as well as a

sufficient power supply. The gate resistor value has been chosen as the default value

given in the datasheet for the transistor irg4psh71u from International Rectifier, with this

value the switching loss has also been estimated.

515

−+

−++

GEGE

peakG

= 4 A (2.8.1)

The R

is an internal gate resistor common in larger transistor modules but for the models

represented in this thesis this resistance is negligible.

As well as the peak current a sufficient continuous current will be required from the

driver and the power supply which can be calculated as in equation 2.8.2.

(

)

(

)

=×+×=−×+×= 5620057010000 pFnCVCQfI

GEiesgswG

0.006 A (2.8.2)

Where f

is the carrier frequency, Q

is the gate charge from 0V to +V

and C

ies

is the

input capacitance. As these currents are per driver and each transistor has its own driver

the average current supplied by the power supply has to be at least six times the above

stated value. The high peak current can be managed by a buffer capacitor but then the

power supply has to be slightly bigger. Without the buffer capacitor the power supply has

to be able to provide the peak current which will give a vast loss and a heavy device.

If all the power losses are completely consumed by the gate resistor then the power

required from the driver is shown in equation 2.8.3.

⎟

⎠

⎞

⎜

⎝

⎛

×=

⎟

⎠

⎞

⎜

⎝

⎛

−×++×=

256200

15570

10000

)(_

pFnC

VCVQfP

GEiesGEgswonlossg

=0.0435 W (2.8.3)

Since the gate turn-ON turn is as large as the turn-OFF loss the total loss is simply the

double turn-ON loss as in equation 2.8.4.

(

)

(

)

=×+××=−×++×= 2562001557010000

pFnCVCVQfP

GEiesGEgswlossg

0.087 W (2.8.4)

Selecting the Half Bridge Gate Driver IR2114SSPbF from International Rectifier easily

match the driving needs for each half bridge with the least amount of external

components.

The internal maximum loss of this driver is about 1.5 W each which adds up to the

dynamic loss as explained earlier. To provide the best performance both positive and

negative power supply is recommended, although it’s not crucial.

Driver

component

Value Manufacturer and

model

Mass(g) Dimensions

(mm)

Loss(w)

Driver - IR2114SSPbF 3*4 3*(8x9x2) 3*1,5

Buffer

capacitor

43 nF Jamicon 3*2 3*(5x11) -

Resistor 5 Phoenix 6*0.1 6*(3.2x2) 6*0.087

Table 8 : Driver components

2.9. Power supply

To have a versatile solution being able to provide the controlling system with power

regardless of whether the AC-source or DC-source is used a high voltage step-down

converter is necessary. It has to be connected to the DC-bus and output a double voltage

capable of supplying both the controller and the driver. A fly-back solution from

Fairchild has proven to be an excellent choice since with a high voltage IGBT transistor

instead of the common MOSFET it is capable of handling the voltage spikes common in

fly-back circuits [17]. This along with a transformer with double secondary windings two

different voltages with different polarity is achieved.

Input voltage: < 707V

Efficiency: ~80%

Output voltage:

-5V,+15V±5%

Maximum output power: <25W

Weight: ~70g

Size: ~90x50x30mm

Table 9 : Power supply specifications and dimensions

The physical figures are estimated from the application note [17] using the same

transformer core and the same transistor as used in the example in the same application

note. However a few components such as rectifier and smoothing capacitor can be

removed since it is fed with a direct current.

2.10. Controller

The controller is logic device responsible for handling all signals and take appropriate

action to ensure operation. It handles all fault signals and causes a halt if a serious

problem arises. It also observes the actual voltage levels and steers it towards the proper

level. The controller is of the less crucial components when it comes to power losses or

weight and thus it is not examined in detail. Still, a suitable off-the-shelf component has

to chosen to have some guidelines. The IRMCK203 from International Rectifier is a fully

capable controller requiring minimal external circuits. However since it requires a 3.3V

power supply a fixed regulator has to be used to decrease the voltage from the power

supply.

Since the lost power due to the fixed regulator is dependant on the voltage drop and the

current through it the lowest voltage from the power supply should be used.

The above chosen controller dissipates 1.2 W @3.3V typically according to the data sheet

which gives a current of 400 mA. The combined loss of the controller and the regular are

thus 2 W

(

)()

VVmA 3.37.1400 +× . The space and weight consumed by the controller is

not of vast proportions and could easily be shared with the power supply and the driver

on the same circuit board.

Controller

compoenents

Value Manufacturer

and model

Mass

(g)

Dimensions

(mm)

Loss

(w)

Controller - International

Rectifier

IRMCK203

7 15x15x5 1.2

Voltage

regulator

3.3 V National

Semiconductor

LP3852

9 10x21x5 0.8

Buffer

capacitor

25 V

10 μF

Jamicon 2x2 5x3 -

Table 10 : Controller components

3. Losses

In this part mostly the components responsible for the main part of the loss are

considered, those are the main supply rectifier and the transistor-bridge. To some extent

the input filter and the capacitor bank will also add up to the losses. Mainly two types of

transistors have been considered, the IGBT and the MOSFET. The switching loss in the

rectifier is in this case ignored due to the low frequency and the low switching energies in

the diodes, which leave only the forward conduction loss remaining.

However in the IGBT-bridge the largest part of the loss is due to the high switching

frequency in the slow IGBT-transistors, a minor loss also arise due to forward voltage

drop. In the transistor-bridge the free wheeling diode is usually included further

contributing to power loss but in the cases of missing a built-in diode the IGBT-bridge

and a suitable diode will be evaluated individually. There will also be some loss in the

filter capacitances and the snubber circuit depending on which type chosen and the

severity of the compensation. Some minor power losses will occur in a switched voltage

supply and the control circuit but it is almost insignificant in amount and hard to

optimize.

In appendix D a spreadsheet has been formed over suitable transistors to ease the

choosing of the

most suitable transistor, different technologies such as IGBT and

MOSFET has been evaluated as have different types of semi conducting materials. In the

spreadsheet there is a majority of IGBT before MOSFET, this is due to the difficulty

finding a MOSFET with a V

rating of about 1200V, this is the recommended voltage

when having a line voltage of 400V due to large voltage peaks. We were able to find one

MOSFET with a V

of 800V, however voltage peaks are likely to destroy this device

and the fact that is has a very large R

DSon

that generates a vast amount of dissipated

energy makes this device not suitable.

3.1. Switching losses

3.1.1. Transistor

The switching losses can be derived from the switching energies found in datasheets but

usually requires some conversion since the rated currents and bus voltages might differ

from the application specific currents and voltages. The conversion can be done for the

current using equation 3.1.1.1 and 3.1.1.2 [7]:

)/(

' aveaveonon

ratedIIEE = (3.1.1.1)

)/(

' aveaveoffoff

ratedIIEE =

(3.1.1.2)

Where α and β are multipliers depending on the device and E

on’

and E

off’

are the energies

at the rated average current which is calculated according to equation 3.1.1.3.

ave

×=

(3.1.1.3)

The same should be done analogous with the voltage since it usually differs as well.

•

Turn-On loss (P

)

)(

aveoncon

IEfP

(3.1.1.4)

Where f

is the switching frequency and E

ave

)

is the Turn-On energy at the average

current I

ave

[7].

•

Turn-Off loss (P

off

)

)(

aveoffcoff

IEfP

(3.1.1.5)

Where fc is the switching frequency and E

off

ave

)

is the Turn-Off energy at the

average current I

ave

[7].

3.1.2. Diode

)(FWDrr

averr

IE ×= (3.1.2.1)

)/(

' aveaverrrr

ratedIIEE = (3.1.2.2)

ave

×=

(3.1.2.3)

is the reverse recovery energy per switch which with the switching frequency f

gives

the switching loss of the diode. Usually all characteristic values for the diode can be

found in the datasheet; however it may require some interpolation between values in

graphs to determine the appropriate value for the specific application, such as estimated

junction temperature.

It is very common for an IGBT or a power MOSFET to be co-packed with a silicon FWD

which gives some benefits such as shared cooling and less wiring reducing capacitive and

inductive noise but it also makes it slightly more difficult to experiment with other types

of diodes. A few IGBTs suitable for this application are available with and without a

FWD making interesting experiments possible, especially to use a SiC FWD to optimize

switching loss.

3.2. Conduction losses

3.2.1. Transistor

The losses in the IGBT can be approximated using the information supplied in the

datasheet and some knowledge of the application specifics such as switching frequency

and the power factor of the load [13].

•

The ON-state power loss in an IGBT transistor can be calculated as in equation

54:

⎭

⎬

⎫

⎩

⎨

⎧

⎟

⎠

⎞

⎜

⎝

⎛

××

⎟

⎠

⎞

⎜

⎝

⎛

+××

⎟

⎠

⎞

⎜

⎝

⎛

+×××

⎟

⎠

⎞

⎜

⎝

⎛

⎪

⎭

⎪

⎬

⎫

⎪

⎩

⎪

⎨

⎧

0000

cos

VIRV

sat

ππ

(3.2.1.1)

Where cosφ is the power factor of the load. I

is the root mean square value (RMS) of the

phase output current and both V

and R can be derived from the following picture

illustrating the output characteristics of an IGBT [13].

Figure 32 : Approximate output characteristics

•

Total power loss (P

losstot

)

The total power dissipation can be calculated as in equation 3.2.1.2 [13]:

satoffonlosstot

PPPP +

= (3.2.1.2)

3.2.2. Diode

The total loss and dissipated heat in the diode is dependent on many factors such as

switching frequency, characteristics of the diode and how inductive the load is. An easy

approximate model for power loss calculations is described in equation 3.2.2.1 [13]:

⎭

⎬

⎫

⎩

⎨

⎧

⎟

⎠

⎞

⎜

⎝

⎛

××

⎟

⎠

⎞

⎜

⎝

⎛

+××

⎟

⎠

⎞

⎜

⎝

⎛

−×××

⎟

⎠

⎞

⎜

⎝

⎛

⎪

⎭

⎪

⎬

⎫

⎪

⎩

⎪

⎨

⎧

0000

cos

FWDsat I

VIRV

ππ

(3.2.2.1)

Where cosφ is the power factor, I

is the forward RMS current, R and V

form up the

voltage drop of the diode in the same manner as for the IGBT but probably with slightly

different values. Usually all characteristic values for the diode can be found in the