DC DC Boost Converter Implementation

[skyflyer54]

Hi,

I have currently built a DC DC Boost Converter that can boost from 20Vdc to 325Vdc. But the PWM signal that I use to drive the gate is about 94% duty cycle which is very high.

Can anyone advise me that is it ok to have such a high duty cycle or should I cascade 2 boost converter to reduce the duty cycle? Or is there a better way to implement the design?

Please Advice...

 

[dknguyen]

Hi,

I have currently built a DC DC Boost Converter that can boost from 20Vdc to 325Vdc. But the PWM signal that I use to drive the gate is about 94% duty cycle which is very high.

Can anyone advise me that is it ok to have such a high duty cycle or should I cascade 2 boost converter to reduce the duty cycle? Or is there a better way to implement the design?

Please Advice...

It is my understanding that as the duty cycle gets higher, small variations in the duty cycle (like jitter) cause exponentially higher changes in the output. if your boost converter is working just fine and you can tolerate whatever ripple it's putting out, I don't see why you would need to double the complexity of the converter.

 

[crutschow]

I agree with dknguyen that if it's working ok you probably don't need to modify it, but it sounds like you're operating at the converter's limits.

The high duty-cycle may be due to the inductor being used near it's limit. You didn't say how much current you're drawing at the output, but the boost inductor average current will be about 325/20 times that value. A larger inductance/higher current rated inductor should reduce the duty-cycle.

 

[Ubergeek63]

add 10% more windings and charge the boost inductor through the added windings.

 

[OutToLunch]

94% duty cycle is correct for the amount off boost you are creating.

D = 1 - Vin/Vout

 

[crutschow]

94% duty cycle is correct for the amount off boost you are creating.

D = 1 - Vin/Vout

That is correct. My thinking was off.

 

[Diver300]

You might be better to make a transformer with a second winding on the the coil.

The inductor will still store energy from the 20V supply for one part of the cycle and dump it into the 325V output on the other part of the cycle.

With a 10:1 transformer the duty cycle becomes much nearer to 50%.

 

[Ubergeek63]

You might be better to make a transformer with a second winding on the the coil.

The inductor will still store energy from the 20V supply for one part of the cycle and dump it into the 325V output on the other part of the cycle.

With a 10:1 transformer the duty cycle becomes much nearer to 50%.

that is what i was suggesting ... perhaps I was not clear enough

 

[skyflyer54]

Hi all,

Thanks for all the suggestion...

I have already used a 600mH inductor, that should be a very high inductance value.. Or should I wind some more to have it even larger?

my output works fine but i'm curious, what will happened if i continue to tune my duty cycle even high said like almost 100%.. What anything in the circuit breakdown?

I agree to the formula stated above as that was what I have learnt during school time and went the value what inputted, the duty cycle indeed is almost 94%.

 

[OutToLunch]

100% duty cycle means that the fet will be on all the time, the current in the inductor, and therefore in the fet, will continue to increase and your fet will blow.

nominal duty cycles near the limits (close 100% for a boost and close to 0% for a buck) will work fine for a system that has a consistent load that does not fluctuate. problems will arise when the load increases - either slowly or quickly. more load requires larger duty cycle. a fast load step will require a sudden increase in duty cycle to keep the output from dropping out too much. parasitic board and component losses can also contribute to an increase in duty cycle (I*R drops will force higher duty cycles and as the load increases those I*R drops increase as well).

if the system does not have the headroom to accommodate for more duty cycle then the output will fall out, the duty cycle will go to 100% and unless there are protection features such as overcurrent or undervoltage that will shut things down, then the fet and possibly other components will fail.

 

[Ubergeek63]

600mH?!? are you sure it isn't 600uH?

Come to think of it, you never said how much power you expected to get out. for a simple boost inductor you will need 40x the output current for the peak input current, assuming critical conduction mode. Other wise you will be in continuous conduction and will need to watch the dissipation on the diode and FET.

 

[skyflyer54]

600mH?!? are you sure it isn't 600uH?

Come to think of it, you never said how much power you expected to get out. for a simple boost inductor you will need 40x the output current for the peak input current, assuming critical conduction mode. Other wise you will be in continuous conduction and will need to watch the dissipation on the diode and FET.

Ya... I used machine to measure the inductance and it shows 600mH... which i think it is super huge.. I hope that my converter can come out 325V @ current of about 2.5A. Is it possible?

 

[skyflyer54]

100% duty cycle means that the fet will be on all the time, the current in the inductor, and therefore in the fet, will continue to increase and your fet will blow.

nominal duty cycles near the limits (close 100% for a boost and close to 0% for a buck) will work fine for a system that has a consistent load that does not fluctuate. problems will arise when the load increases - either slowly or quickly. more load requires larger duty cycle. a fast load step will require a sudden increase in duty cycle to keep the output from dropping out too much. parasitic board and component losses can also contribute to an increase in duty cycle (I*R drops will force higher duty cycles and as the load increases those I*R drops increase as well).

if the system does not have the headroom to accommodate for more duty cycle then the output will fall out, the duty cycle will go to 100% and unless there are protection features such as overcurrent or undervoltage that will shut things down, then the fet and possibly other components will fail.

Omg.. u mean my circuit components will blow if the duty cycle reached 100% due to always on of FET and current build up in the inductor?

I'm thinking of using a micro controller to do the protection features but I'm not sure will it be a good approach or not. May I know what is the usual or normal way to design overcurrent or undervoltage protection?

 

[Ubergeek63]

Yes... What kind of inductor is it? do you know the number of turns? As I said before, start with an inductor rated for your output current and at 5% of the turns number in series.

If you have 1A out of a 60mH you will have 40A into 150uH.

it looks like about 63% at 2100Hz...you REALLY want to knock that down by another order of magnitude to 6mH and 15uH for 63% @ 21KHz... or you could run continuous mode and deal with less input current ripple and more dissipation in the switching elements

 

[skyflyer54]

Yes... What kind of inductor is it? do you know the number of turns? As I said before, start with an inductor rated for your output current and at 5% of the turns number in series.

If you have 1A out of a 60mH you will have 40A into 150uH.

it looks like about 63% at 2100Hz...you REALLY want to knock that down by another order of magnitude to 6mH and 15uH for 63% @ 21KHz... or you could run continuous mode and deal with less input current ripple and more dissipation in the switching elements

Actually I don't really know how many turn but I wind till max and I use the machine to measure the value. Is there a good way to calculate the inductance value?

What u mean by 40A into 150uH and what is the 21KHz?

Is there any good website that teach us how to design and do all the calculation Or do u have a better design to guide me along?

I think i'm really starting to be lost inside jungle now.

 

[Ubergeek63]

what are you using for a controller? And don't say a pathetic PIC because what you are attempting to do is way out of the capabilities of most here, never mind a noob such as yourself.

What are you using for a inductor core? For windings? For a power switch? For a switching diode?

Oh fine...

using a UC3842 targeting 40% duty cycle for 325V@2.5A from 20V:

"generic" duty cycle is

1-Vin/Vout = 1-20/325 = 94%

100%-94% = 6% on the output but we want 60% so
6/0.6=10, 10-6 = 4, and (94+4)/4 = 24.5
or a turns ratio of 25 to get a 40% duty cycle.

The output current needs to be increased by the duty cycle to compensate for the switch on time that the current is not delivered to the load.

Ityp = Iout/DUTYout = 2.5/0.6 = 4.2A

Iin = Ityp*RATIO = 4.2*25=105A 40% of the time. Your FETs need to handle at least 200A continuous and preferably 400A.

Primary inductance assuming the FET is on 20uS and 30% ripple current:

L = dt*Iin*0.3/V = 20*105*0.3/20 = 30uH

Operating frequency should be F=1/(20uS/0.4)=20KHz

So make a 30uH primary and a secondary with 25 times the primary turns.

 

[Ubergeek63]

well that is what i get for rushing through this sort of thing during work hours.

I will fix it over the weekend

 

[Ubergeek63]

well what do you know...I was right

the other way is pick a frequency and duty cycle target.

calculate the primary inductance from 40% of 50KHz (8uS), 20V and the "inductor ripple" of 105A*0.3 = 31.5A:

Lp = TpVp/Ip = 8*20/31.5 = 5uH

now we have three other equations to work with:

Ls = TsVs/Is, Ls = LpT^2, and Is=Ip/T

If we plug the second two into the first and solve for T we get:

T = (TsVs)/(IpLp) = (12uS*325V)/(31.5A*5uH) = 24.76 = 25:1

 

[dknguyen]

This line of PICs is designed for SMPS applications and has things like built-in comparator that control interrupts along with a special interrupt that works with the special PWM module (that can itself output a bunch of weird PWM schemes used for SMPS) to provide cycle-by-cycle current limiting. It's internal oscillator also has built in clock dithering which is neat.

SMPS & Digital Power Conversion - Application Features - 16-bit Microcontrollers and Digital Signal Controllers

 

[Ubergeek63]

looks like that would work