Buck LED driver from 24V input

Gordon,

Buck regulators are designed to take an input voltage of 24V, for example, and regulate it down to 7V, for example.  Do you need a power supply output of 7V at 2A?  Is your input source +24V?  If so, we have a power quick search at power.ti.com that will provides a wide variety of ICs from which to pick.  For 24V to 7V conversion at 2A, the TPS5420 is a good simple fit.

Hi Chris,

No, I have a different problem from that.

My load is between +24V and +17V, so I have to regulate the +17V wrt the +24V.

You could look at it another way by transposing everything downwards, so it looks like I have a -24V supply and I have to regulate a -7V supply from that -24 for my -7V load. I hope that makes sense.

It's also complicated some by this being a constant current load an I need to adjut the `7V' up or down a little to control dissipation.

I need multiple such supplies in the unit, probably eight.

I can not regulate on the positive side. as the multiple loads have a common positive connection.

I did say it was atypical :-)

I could use +24V, put the common connection at 0V, and use inverting regulators to get -7V, though that seems rather obtuse for several reasons.

Gordon.

If I understand you correctly, you are trying to regulate 7V across an element whose common connection is the higher voltage, of 24V or ground (for a -7V output)?

An inverting buck boost is an option and we have a lot of collateral to help you design one.  The TPS5430 should be able to deliver 2A with your Vin and Vout.  Here are some app notes describing the design of such a converter:

http://focus.ti.com/lit/an/slva257a/slva257a.pdf

http://focus.ti.com/lit/an/slyt286/slyt286.pdf

http://focus.ti.com/lit/an/slva317a/slva317a.pdf

This application sounds similar to an LED driver, so I just wanted to make sure that you can't connect your loads in series or some other configuration similar to this app note: http://focus.ti.com/lit/an/slva374/slva374.pdf

Hi Chris,

Yes it's an LED driver but I can't do that configuration in this product.

In this case the LEDs are paralleled and the anodes are common, so all the control electronics have to be on the cathode side. As we're doing high powers in fairly small spaces, I want the supplies to track the LED voltage drops, i.e., I deliver constant current but can't afford too much dissipation in the current controllers.

I can certainly see the TPS5430 being useful, maybe in this context, maybe not.

Gordon.

Gordon,

How many LEDs do you have in parallel and how much current do you want to drive through each of them?  We have a wide variety of multi-channel LED drivers, but most are boost converter based: TPS61195 as an example: http://focus.ti.com/lit/ds/symlink/tps61195.pdf 

With a multi-channel converter, the channels can usually be paralleled to get higher current through just one LED string.

Here is a linear LED driver to look at: http://focus.ti.com/lit/ds/symlink/tl4242.pdf  You could use a regular buck converter to step down the 24V to around 8V first.  Would this IC work if it could handle 2A?

Hi Chris,

I can't answer the question about how many in parallel as it depends on the particular system and particular die. We do our own die and wire bonding and our LEDs are not typical.

My aim for this particular task is 2A `per LED', though other models have 10A `per LED'.

The LED modules can contain mixed die, so I have to manage voltage drops that differ significantly (say 2V ... 5V) and all the anodes are commoned, so I have no opportunity to control that side of the LEDs. FWIW, they're common because they're all bonded to the same metal block and heatsink.

Regards,

        Gordon.

The huge difference in voltage drops makes this more difficult, as something must dissipate the extra voltage in the lower voltage LED.  At 10A, this could be 30W.

It sounds like the TLC5960 LED driver controller could do what you want: http://focus.ti.com/lit/ds/symlink/tlc5960.pdf

It contains many, many dimming features but you don't need to use those of course.  Again, the problem will be how to dissipate 30W in the paralleled FETs that would be required.

Hi Chris,

That has similarities with what I do at present, except that my sinks have to be linear, I can't tolerate PWM there. But as you note, with significant differences in voltage drop, the dissipation in the sink circuits is significant, with all the obvious consequences.

What I'm trying to achieve is switching regulation on the low side, preferably smooth enough that I don't need the linear current sinks (I'm sceptical that's feasible), or if I still need the linear sinks, a low voltage across them after adjusting for differences between LED voltage drops.

Again, though, thanks for drawing my attention to that part. I think it may well be useful, though in product that doesn't demand the low ripple.

Regards,

        Gordon.

Hi Chris,

Perhaps I should reitterate my original question.

I have the means to control the LEDs as I need to. What I don't have is the means I'd like to control the cathode end power supplies to handle the differences in voltage drops on the LEDs.

The reason I posted in the digital supplies area is that I think the quad-channel controller has the potential to do that .. It's software controlled, synchronous and has the possibility to bi-phase, which may allow me to elliminate the linear controls, improve the efficiency and reduce my heat management issues. I say may, because I have a number of control issues that can be difficult for digital supplies to handle .. precise current control at the percentage level, quick turn-on and turn-off, and so on.

At present, I already use a CPU to manage the existing single supply via a digital potentiometer in the feedback, but the digital psu already within it, already have the means to do that., I think.

The catch with the digital supply is that TI's tools for it seem again to be locked into the view `all supplies are referenced to zero', which may mean I have to use inverting supplies evn though I don't reall want to (it puts all my other stuff below zero as well, making me isolate or level-shift all over the place, etc.), or encouraging a kind of fake 0V at around +7 below which I work, but above which the PSUs work. All sounds messy to me.

Regards,

        Gordon.