Electronics > Projects, Designs, and Technical Stuff
High-side load switches for USB power rated 5A?
HwAoRrDk:
I'm wondering if anybody knows of a high-side load switch of the kind typically aimed at switching 5V USB devices that can handle 5A?
Currently I have a design that uses a 3A switch like the Diodes AP22815 or Richtek RT9742, but I'm interested in something that can handle higher loads up to 5A. However, I'm having trouble finding anything that fits the bill, and I'm not even sure that something like that exists. By the way, I'm aware that such a thing might not exist, because what kind of USB device even draws that much current? But I'm not actually switching a USB device, just something else that runs on 5V, and USB load switches fit the bill perfectly for my use case.
The features I'm primarily interested in are:
- Over-current and short-circuit protection
- Reverse current/voltage blocking
- Fault flag output
Most general-purpose high-side load switches I've found seem to be aimed at automotive use, but aren't suitable as their minimum operating voltages are 5.5-6V, and they don't have reverse blocking.
I've found one or two load switches aimed at USB PD (e.g. NXP NX20P5090), but they would be an extreme pain to use, as they have tiny, tiny WLCSP BGA packages. :(
Any suggestions for parts that fit the bill?
T3sl4co1l:
Dumb(?) idea: use two in parallel?
Might be load switch controllers (external MOSFET), not sure offhand, also probably overkill for the low voltage and simple application.
Tim
HwAoRrDk:
How would that work with the over-current protection? How do you balance the current across the pair?
As I understand it, the aforementioned devices that I'm currently using perform OCP by current-limiting when they detect the threshold has been exceeded, which then subsequently triggers thermal protection. But the thresholds are kind of approximate, and not necessarily consistent from individual device to device, so wouldn't that lead to some undesirable behaviour?
For example, say I'm using a pair of 2.5A devices, but they have slightly differing RDS(on). The device with the lower on-resistance will take more current, and thus potentially trip too soon because it has too large a share of the current. For example, its share is 2.6A versus 2.4A on the other, so it trips OCP, even though the total current hasn't exceeded 5A.
Would I just need to AND the two fault output signals? So that we don't consider an actual fault until both devices have tripped?
HwAoRrDk:
After a lot more searching, I have actually found a few suitable devices:
- Diodes Inc. DPS1135
- ON Semi NCP45780
- Microchip/Micrel MIC2044/2045
It's a bit frustrating trying to find these things, because it varies from manufacturer to manufacturer whether they categorise them under "USB" or "Power Management". I had only been looking at the latter. ::)
The DPS1135 looks pretty good - seems like it would do the job, and it's fairly inexpensive (~£1 ea.). It does have a weird 3-sided 0.5mm pitch QFN package, but that's probably manageable. The NCP45780 also looks like a good option, but is twice the price of the DPS1135. Also another unusual QFN package with VIN and VOUT on separate central pads, but workable. The Micrel part, while nice on paper and in an easy-to-use TSSOP package, is stupidly expensive (~£3.50!).
I have questions about a couple of features of the DPS1135:
What is Fast Role Swap (FRS)? I gather it's part of USB PD, to do with a USB device supporting the ability to either supply power or receive power. It's not clear to me what value I should give the FRS pin for usage where PD isn't involved and I'm just supplying power. Plus the description on page 12 of the datasheet indicates it works off a pulse rather than level? :-//
The output discharge control (DISC2) seems to be completely independent of everything else (such that you could even enable it when output is powered! :scared:) and not automatic. If I wanted to make it automatic, such that I didn't have to add another MCU GPIO (of which I don't really have any spare) to control it, could I derive control of it from the EN signal so that the output discharge is active whenever EN is off? Maybe invert the EN signal with a small transistor and include a short RC delay?
T3sl4co1l:
--- Quote from: HwAoRrDk on October 03, 2023, 11:25:21 am ---How would that work with the over-current protection? How do you balance the current across the pair?
As I understand it, the aforementioned devices that I'm currently using perform OCP by current-limiting when they detect the threshold has been exceeded, which then subsequently triggers thermal protection. But the thresholds are kind of approximate, and not necessarily consistent from individual device to device, so wouldn't that lead to some undesirable behaviour?
For example, say I'm using a pair of 2.5A devices, but they have slightly differing RDS(on). The device with the lower on-resistance will take more current, and thus potentially trip too soon because it has too large a share of the current. For example, its share is 2.6A versus 2.4A on the other, so it trips OCP, even though the total current hasn't exceeded 5A.
Would I just need to AND the two fault output signals? So that we don't consider an actual fault until both devices have tripped?
--- End quote ---
Well, so what? If one drops, suddenly there's 5A on the other, it immediately goes into current limiting, it opens thermally a moment (100s us?) later, and both are faulted out. Both stay off until they cool down, which won't track, but this might be as much a feature as a bug: consider it a time-delay 50% foldback. If under MCU control, then reading the fault signals (AND or OR would be acceptable, depending on which characteristic you want; use OR to avoid "undesirable" behavior) and turning them off for a safe margin before turning on again (in which both are expected to be ready again) avoids the "foldback" behavior. :)
Tim
Navigation
[0] Message Index
[#] Next page
Go to full version