boost converter with low quiescent draw

[james_s]

I have a few different one-off projects in mind where I have a need to get 24 or 48V from a 12V input. I have a couple of those cheap modules based on the XL6009 or that's what they claim to be anyway and they work but the quiescent draw is around 40-50mA, not a lot but when something is powered 24/7 it adds up. Can I do better? Anyone got a part to suggest? Ideally something with an internal switch that is in a package I can deadbug, load is maybe 5 watts for maybe a minute at a time, low duty cycle.

Also looking for an efficient buck regulator to supply 3.3V at about 70mA from the same 12V input, that will have a continuous load on it though so I'm not as bothered by quiescent draw. I can probably salvage something from some random bit of scrap that will meed that need.

[CountChocula]

I use the SDB628 as my go-to boost regulator for small projects; it's cheap and works remarkably well, with quiescent current in the µAs, but it's not exactly easy to dead-bug and it won't go to 48V. (For my uses, I had a small PCB made where I can mount the chip and all ancillary components—happy to send one your way if you want to play around with it.)

Still, it looks like there's plenty of choice in this category, even if you want to go with THT mounting.

For a buck regulator, Mornsun makes these little 78xx compatible modules that are inexpensive and dead simple to use—you just plug them in, add a couple caps, and you get your 3.3V regulated output. They can easily be used to generate a negative rail as well.


—CC

[james_s]

Thanks, I wish quiescent draw was easier to find without digging into the datasheets, LT8364 looks promising, it can do 48V and is in a surface mount package that is large enough that I can easily throw something together.

I've used those 78xx substitute modules before and they work well for when there's an existing piece of hardware with a linear regulator in it but for this application I want either a single sided leadless module or something in a 8 pin SO package would be perfect. I have some modules somewhere I got surplus a while back that I'll try for this application, I actually forgot about those. 

[mariush]

Did you try MP3425 ?

Link : https://www.digikey.com/en/products/detail/monolithic-power-systems-inc/MP3425DL-LF-Z/5292359

Datasheet:  https://www.monolithicpower.com/en/documentview/productdocument/index/version/2/document_type/Datasheet/lang/en/sku/MP3425/document_id/1122

For buck regulators, pick any 100mA or higher buck regulator from Digikey/Mouser/Whatever ... lots of synchronous rectifier models that require very little extra components.

for example AP63203 (fixed 3.3v output) : https://www.digikey.com/en/products/detail/diodes-incorporated/AP63203WU-7/9858426
or AP62200 for the adjustable version : https://www.digikey.com/en/products/detail/diodes-incorporated/AP62200WU-7/10491524
or AP62301  https://www.digikey.com/en/products/detail/diodes-incorporated/AP62301Z6-7/12349219
or TPS562207 : https://www.digikey.com/en/products/detail/texas-instruments/TPS562207SDRLR/13563017

[james_s]

Glancing over the datasheet I don't see the quiescent draw, ie regulator operating with 48V on the output and no load applied. QFN is probably going to be tricky to prototype without a PCB though, not that I can't lay out a board but I was hoping to avoid it for this project.

[mariush]

There's an evaluation board available, if you feel it's too hard to make your own

Costs $25 + 5$ shipping : https://www.monolithicpower.com/en/catalog/product/view/id/1808?utm_source=octopart&utm_medium=component_search&utm_campaign=listing

You're right, it doesn't say the quiescent draw at 48v output, only

Supply Current (Shutdown) V EN = 0V 1 μA
Supply Current (Quiescent) V FB = 1.35V 650 900 μA

[james_s]

Ok I missed that, yes it looks like the quiescent on that part is quite reasonable.

[SiliconWizard]

I can only recommend LT parts for this, but they are a bit expensive. So depends on your application. Good thing is that LTSpice does simulate them pretty accurately too.

[james_s]

Well low price is always nice, but I'm only building 2-3 similar devices so really whether the part is $2 or $8 doesn't make *that* much difference. I was just annoyed when I realized the cheap boost modules I have while they work fine, would cumulatively consume much more power just sitting there than is ever used by the load. Half a watt doesn't sound like much but multiplied by ~8,000 hours a year and all the other small loads I have it adds up.

[SiliconWizard]

In that case, the LT8364 should be pretty good.

[james_s]

Do you have a suggestion for the inductor? I've designed buck and boost converters before but 500kHz was the highest operating frequency I've ever worked with and this LT part runs at 2MHz. The datasheet says "Choose a core material that has low losses at the programmed switching frequency, such as a ferrite core" but there are multiple types of ferrite and I don't see any specific examples and looking over inductor datasheets I haven't found one that even shows losses at that high of frequency. Seems like a fun challenge to get this right but I'm a bit out of my element and could use a bit of guidance.

[SiliconWizard]

A good starting point is to have a look at the evaluation board, the reference is on AD's page for the LT8364, and the schematic is provided.
For this evaluation board, they have used a WE 74437324022, 2.2µH inductor. If you have determined a different value for the inductor, going for the same series is a safe bet.

This series of inductors: https://www.we-online.com/en/components/products/WE-LHMI

[tooki]

I just did my final school project, and my task involved boosting 5V to 15V with under 500μA quiescent current. Turns out there’s a lot of stuff. Browse the anti parametric search — Iq is a column you can add, if it’s not there by default — and then simulate the designs with WEBENCH power designer. (My little boost converter draws about 40μA quiescent in the assembled boards. I should pull out the electronic load and see how much current it can deliver. It should manage at least 100mA IIRC. My own load is only around 3μA so I didn’t need to stress test it as such.)

[james_s]

Interesting that the one on the eval board is apparently powdered iron rather than ferrite as suggested in the datasheet. I'm going to have to read up again on the tradeoffs of powdered iron vs ferrite, it's been a while since I dove into that.

[Siwastaja]

Do you have a suggestion for the inductor? I've designed buck and boost converters before but 500kHz was the highest operating frequency I've ever worked with and this LT part runs at 2MHz. The datasheet says "Choose a core material that has low losses at the programmed switching frequency, such as a ferrite core" but there are multiple types of ferrite and I don't see any specific examples and looking over inductor datasheets I haven't found one that even shows losses at that high of frequency. Seems like a fun challenge to get this right but I'm a bit out of my element and could use a bit of guidance.

This is not a datasheet parameter, but some manufacturers have modeled AC loss of their parts and offer online tools where you simply enter your Idc, ripple%, duty cycle and f_sw and they output you DC loss, AC loss and recommendation of part numbers:

e.g. Coilcraft: https://www.coilcraft.com/en-us/tools/power-inductor-finder/#/search
Also Wurth Elektronik RED Expert tool.

There are more; look it up at manufacturer websites instead of just datasheet.

Without such data, the only way is to try your luck and test.

[Alti]

I have a couple of those cheap modules based on the XL6009 or that's what they claim to be anyway and they work but the quiescent draw is around 40-50mA, not a lot but when something is powered 24/7 it adds up. Can I do better?

That draw does not come from the chip:

Quiescent Supply Current: 2.5mA
Shutdown Supply Current: 70uA

[tooki]

I have a couple of those cheap modules based on the XL6009 or that's what they claim to be anyway and they work but the quiescent draw is around 40-50mA, not a lot but when something is powered 24/7 it adds up. Can I do better?

That draw does not come from the chip:

Quiescent Supply Current: 2.5mA
Shutdown Supply Current: 70uA

As I learned the hard way, the datasheet Iq unfortunately does NOT mean what you probably think it means (the current draw with the power supply running, but with no load). It actually means the current in the chip when it is enabled, but not switching the MOSFET. But even with no load, in reality your MOSFET will be switching, in order to maintain the output voltage. Note that in the datasheet, the test condition for Iq is Ven=2V (i.e. enabled, running) and Vfb=Vin (and above the table, it says Vin=12V). Meanwhile, the nominal Vfb voltage is 1.25V. So basically, they’re applying a really high feedback voltage, ensuring the thing doesn’t switch, because from its point of view, the output voltage is already much too high, so it doesn’t need any pulses.

Most other data sheets I have looked at simply say that the condition is “non-switching” for Iq.

Remember in my comment above about my boost converter with a quiescent current of about 40μA? That is roughly what I expected from simulations. But the datasheet Iq is 1μA (0.07μA in shutdown).

[james_s]

Without such data, the only way is to try your luck and test.

That's been my approach in the past, with mixed results. Sometimes it turns out great, sometimes less so. Magnetic engineering is an area of interest for me but it's something that still feels practically like voodoo. Not as bad as RF though, which is also involved in a 2MHz converter. I'll try to tweak my layout such that a few different inductors will fit.

[SiliconWizard]

Interesting that the one on the eval board is apparently powdered iron rather than ferrite as suggested in the datasheet. I'm going to have to read up again on the tradeoffs of powdered iron vs ferrite, it's been a while since I dove into that.

Wurth claims that this is a special "mixture", whatever that means. All I know is that I've used this series (and similar ones) for DC/DC switching bucks and boosts up to 2MHz with no issue, and they are made for this. I'm not advanced in physics enough nor have internal info about Wurth's designs to fully understand how they can get as good as ferrite with those powdered irons, but apparently it does the job.

Very interesting topic indeed but you may be digging a bit too deep here in this case.