# High power (up to 400W) LED DC/DC power supply design considerations

I am designing a power supply for a high power COB LED requiring 360W @ ~75V from a variable ~60-80V DC input. I have noticed that it is difficult to find switching controller ICs that advertise being able to operate at such high power outputs, and those that do tend to be relatively complex with 40+ pins and a large external BOM (e.g. LT8210).

Previously I designed a much simpler buck LED driver circuit using the AL9910, albeit at a lower power output and single topology, that was as close as it gets to the barebones buck converter circuit, but it worked in that application without any issues. I am now trying to understand what the inherent differences are between the high and low power applications, and whether I could simply use a "basic" buck/boost switching IC with "upgraded" components or if there would be some major problems with this approach.

Some possible explanations for the differences between the ICs that have occurred to me are:

1. Efficiency - The AL9910's peak efficiency is just above 90%, while the LT8210 appears to be closer to 98%, which is certainly much more important at higher power outputs, though not necessarily a major concern in my application.
2. Heat - The higher power output generates more heat in the components involved, which at <90% efficiency is very significant, however I would have assumed that this does not have a major effect on the switching controller IC itself if the external components are capable of dealing with it. Although wasteful, I might prefer a larger heatsink and fan than a more complex circuit overall.
3. Regulation - Tight voltage regulation and noise reduction.
4. Inductive Coupling - The higher currents involved have a greater effect on signal traces.
5. More Complex Applications - It makes sense that such high power systems would usually be found in more complex applications, where the additional circuitry for precision, safety, etc, is worth it given the higher cost, so the simple designs are uncommon.

To summarize my question:

What practical limitations or unexpected difficulties exist when increasing the power output of basic external power switch controller ICs beyond that of their recommended "typical applications", and are these limitations (in)directly addressed by more complex controller ICs?

A concrete example for this question:

The LT3757 provides a typical application with a 8V-16V input and a 24V 2A output. Vin is the only pin tied to the power portion of the circuit, the switching MOSFET is external. Is there any reason why a circuit with a 100V 5A output could not be made by adjusting the MOSFET, inductor and feedback voltage dividers, assuming the input voltage is scaled up by an equal amount?

As this is for a hobby DIY project I am not overly concerned about efficiency, safety and cost of components, but I would like to avoid designing and ordering a PCB that ends up not working at all.

• I think you've done a yeoman's job at describing the situation well. I'll give you a +1 for that, as it's not often done this well. That said, I also agree with the one person (at this moment) who voted to close the question because of a lack of focus. You are asking for way too much and should be able to narrow this down a bit, saving less related questions for a different moment. Look through your summary and see if you can organize them into "tightly related groups" and then pick one highly related group of them to ask, setting aside the rest for later.
– jonk
Aug 29, 2020 at 20:56
• This is the first time I've ever wanted to both +1 a question and at the same time also wanted to vote to close it, too.
– jonk
Aug 29, 2020 at 20:57
• Thank you for the feedback @jonk, I hope my edit has made the question more direct. It seems my original selection of questions was as wide as the topic of power supplies itself! Unfortunately the combination of great curiosity and a vast selection of possible solutions always leads to many of them from me. Aug 29, 2020 at 22:21
• It's still very broad -- "please contrast high power systems with low power systems" -- and doesn't really address what I'd hoped would be "narrow and concrete." But I retracted my vote to close, anyway. (I give up the chance to reverse that, in doing so, as well.) The rest is good and perhaps someone will tackle a facet of your question. In my opinion, everything gets more complicated when moving from lower power towards higher power. Nothing is left unscathed. Everything is harder, takes more thought, requires the addition of still more and more protections, caveats, etc.
– jonk
Aug 29, 2020 at 22:28
• Once more edited this time with a concrete example. That is indeed the general concern I had as what is an obvious part of the answer to my question is that there is most likely a good reason for the "complexity" - nobody is doing it for fun. My hope still in asking is that it might be the case that eliminating a part of those protections and the expectations of certain characteristics could allow for a basic circuit that works, but I do fully anticipate a variety of reasons why this is not possible. Aug 29, 2020 at 23:07

## 3 Answers

Your question appears to be along the lines of ..... this is just a hobby DIY application, so does it make sense to use a 'larger' fairly simple design I know works rather than start anew with something more complex using components better suited to the specific application.

In the limited application you mention, I would give a guarded YES. You'll particularly need to pay attention to your controller chip's drive capacity and buffer it if necessary.

Unfortunately with this much power to regulate in Boost/buck mode, the stability for all requirements of load step-up or down to tight error limits requires complex control of the input and output energy with slope compensation, absorption with pre-bias, soft start, OVP, OCP, OTP/

Another example is the TI LM5036

This also adds to the cost of the source supply which is an unregulated supply that requires a boost.

My suggestion is to choose the appropriate power supply design from AC to DC so that the architecture is optimized rather than compromised. This results in only a step-down or Buck mode which is far more stable and less complex. Power sources of this size commercially also need to have active PFC.

So there is no simple solution to add-on to your supply, yet the volume BOM cost can be, low as a purchased product.

Make vs Buy => BUY

Why not think of your job as a 5 A current source. This would give the LEDs about 75 VDC. The current does not need to be stunningly accurate. If the current had a small negative temperature coefficient this would make the design more robust. If you can accept this for your job you can just use discrete parts and things are very simple. It will be just a buck current limiter which will be as efficient as a buck converter without turn on losses.