Using LED drivers as DC/DC converters?

[aanders1]

I was told to use **broken link removed** as a DC/DC converter to make 2.7V - 3.6V (at up to 9A) into 24V-36V at 1 A. (Yes, I know that problem 1 is that the input voltage I have is too small for this chip, but pretending I can double it for now, let's get into the other issues.)

Like most LED drivers I'm seeing, there is a PWM associated with it. Can these be run at 100% duty cycle to provide constant DC output? Let's just say the device I'm powering would not be happy with fluctuating voltage (mini-CPU) so if not, that's bad.

Ideally my converter (for the Vo and Vi above) is 90% efficient or greater and fits into a cubic inch. I laughed when I heard the requirements, but then again what do I know.

In the meantime, I'm trying to get as close as possible. Could it be that an LED driver is the answer? I have trouble believing a chip this small can do this for me with constant current output without massive heat problems at the very least...what am I missing?

Any suggestions on anything that come close to my requirements?

 

[Ubergeek63]

No way, that thing only has a 1A switch.

36V*1A=36W ... 36W/2.7V=13A ... with a transformer that allows a 50% duty cycle: 13A*2=26A peak (normally you would be looking at a DC of 2.7/36=7.5% for that voltage ratio or a peak current of 13/.0075=170A).

Dude, you need a second lithium, at least.

 

[ronsimpson]

Look at the FAN5336 or FAN5330
There are many LED PWMs that work down to 2.7 volt.
Some work to 36 volts out.
None I know of work to 9A in (1A out)
You could parallel up 9 of the FAN5330

 

[aanders1]

Right. I realize I may need to just line up a bunch of components in parallel to handle the current, though fewer parts is always nice...

(And yeah, I know my voltage/current ratios don't make V1*I1=V2*I2 work if you take the min in and max out...believe me, I know. That's one of the reasons I need as much efficiency as possible.)

Does anyone know though if these chips can be driven at 100% duty cycle (PWM) to provide something that looks like DC out?

 

[blueroomelectronics]

100% duty is DC What are you trying to build?

 

[ronsimpson]

100%=smoke!

There are PWM ICs that drive external transistors where 9 amps is possable.

Look at ZXSC300.

 

[Ubergeek63]

Right. I realize I may need to just line up a bunch of components in parallel to handle the current, though fewer parts is always nice...

(And yeah, I know my voltage/current ratios don't make V1*I1=V2*I2 work if you take the min in and max out...believe me, I know. That's one of the reasons I need as much efficiency as possible.)

Does anyone know though if these chips can be driven at 100% duty cycle (PWM) to provide something that looks like DC out?

Fine... max in min out:

24V*1A=24W ... 24W/3.6V=6.7A continuous ... you are still talking about 14A peak with a 50 % duty cycle worthy transformer and 45A peak with just and inductor. Still far out of the capabilities of single chips and you still have no chance of getting it in one cubic inch...

Remember, not only do you have to handle the DC current levels the inductor can not saturate at 50A even assuming max in and min out. Not to mention the fact that on a single lithium cell, of say 4AHr, you will get all of 17 minutes of run time, that is assuming it can put out that kind of current continuously without turning into a torch! (that is the voltage of the chemistry in which there are some that will handle it)

 

[Willbe]

You know, the link implies that the output of this thing acts like a current source, and not a voltage source. . .is that OK for your app?

 

[Ubergeek63]

You know, the link implies that the output of this thing acts like a current source, and not a voltage source. . .is that OK for your app?

Actually that depends more on what you use for feedback in many cases, but controllers meant to be current sources always have lower sense voltages.

Be that as it may, that part has the tiniest fraction of the capacity he is asking for.