USB-C power supply protection

[golden_labels]

I'm adding the CH213K in case a bad power adapter comes across or something cause for around $0.07.... I think it is worth it. That's my thought. But I'll remove it if it's REALLY unnecessary.

Unlike barrel jacks, which can have either polarity and it’s likely that a wrong one is used due to either user error or bad markings, all USB connectors have only one, well defined polarity. There is no option for a mistake.

There is always a possibility that the USB-C power adapter is broken or a scam. But having polarity reversed is only one of the options. CH213K isn’t going to protect against voltages exceeding ±6.5 V. It’s not going to protect against the the power adapter being earth-referenced on positive rail.⁽¹⁾ Or it being connected to live. There is no way to protect against everything, and in this case the protection provided is only addressing only a very minor issue. There is an old saying: “any idiot can build a bridge that stands, but it takes an engineer to build a bridge that barely stands.” A part of a reasonable design is reducing costs where they don’t really improve much.

This is not a suggestion to skip that protection completely. But re-evaluate if you really believe you need it. And if you need CH213K instead of a generic schottky diode. Even at 1 A the difference between CH213K and SS34 is 4.6 V vs 4.8 V. Both are way above 3.3 V used by ESP32 and with a linear regulator you need to drop that voltage by converting power to heat. It’s a choice how it is distributed between the diode and the voltage regulator.

Again I stress it’s not my point to tell you to certainly not choose CH213K. It’s to trigger rethinking the design.

For the SMAJ0.5A I think yeah it makes more sense as you say, to be in front of everything. I read about how to place the CH213K and the SMAJ0.5A together and the suggestion was the diode to come after it, that's why I went that way.

SMAJ0.5A is a TVS meant to suppress very short “random” transients that may appear on the power lines. If we put it behind CH213K, it’s still going to protect the rest of the circuit. The problem is that CH213K remains unprotected. Unlike a true diode, which does not have connection to the GND and therefore never sees the full voltage spike, CH213K does.

In the CH213K datasheet it says that is equivalent to a schottky barrier diode plus a self recovery fuse, so I took that as it has a fuse...?

I still don’t understand how comes companies spending enormous money on putting a chip in production can’t hire a decent translator.⁽²⁾ But it’s very common among Chinese companies that their datasheets and manuals looks as if they went through poor man’s version of Google Translate. This one is not an exception and it’s plagued with errors suggesting the translator didn’t know English or electronics. The use of word “fuse” is probably some bad translation.

The purpose of a fuse is to:

• … reliably break the circuit in an overcurrent situation.
• … reliably increase resistance (and therefore reduce voltage and current seen by the circuit) if current rises too much. This is the a special case for polyfuses.

There is no option for it not doing so, because e.g. logic didn’t click or a defect in silicon disturbed an amplifier.

CH213K senses current and uses an amplifier to control the MOSFET. The function is primarily to not overheat the CH213K itself. The closest it gets to fuses is being an unreliable “egoistic polyfuse.”

⁽¹⁾ A wall adapter should not be earth-referenced at all. But out of the two possibilities, GND being bound to earth is relatively harmless. Compared to VBUS, which is going to be catastrophic the moment we connect this device to something else.
⁽²⁾ This is different from some smaller vendors, in particular consumer products manufacturers or eBay sellers, where I can completely understand that it’s just small companies that just try to live on an international market. My English is far from perfect, and I remember how Polish companies tried to appeal to western audiences in the past. But deploying a custom chip entails costs so huge that hiring a translator is 0.05% of the costs, and this particular company is not some small player. I expect the CEO makes monthly 100× more than the translator.

[schotkky]

Just FYI: CH213K is out of stock on LCSC and not available at all on digikey and mouser

Thanks for your thoughts on stock, jlcpcb seem to have 2000+ pieces

[schotkky]

Hi all
I am new to the forum

Are you an AI agent?

Haha, how does that make me AI agent, I am not

[schotkky]

SMAJ5.0A is for reverse polarity protection which seems fine

Using a TVS without a fuse it can blow is not exactly pointless but it's dangerous, and using a same or greater contact area Schottky diode as a fuse vs TVS diode in SMA package does not instill confidence to say the least (it's a fire starting trap).

In its datasheet it says it can act as fuse in a way so i am not sure if i need an actual fuse. This is what Google's AI gave me as comparison, not sure if it's correct and beneficial to the circuit:

"Traditional Fuse vs. CH213K:
Traditional Fuse: Permanently breaks the circuit ("blows") to stop a fault, requiring physical replacement.
CH213K: Intelligently restricts current flow and shuts off automatically during a fault, resuming normal operation after the fault is cleared."

[schotkky]

I'm adding the CH213K in case a bad power adapter comes across or something cause for around $0.07.... I think it is worth it. That's my thought. But I'll remove it if it's REALLY unnecessary.

Unlike barrel jacks, which can have either polarity and it’s likely that a wrong one is used due to either user error or bad markings, all USB connectors have only one, well defined polarity. There is no option for a mistake.

There is always a possibility that the USB-C power adapter is broken or a scam. But having polarity reversed is only one of the options. CH213K isn’t going to protect against voltages exceeding ±6.5 V. It’s not going to protect against the the power adapter being earth-referenced on positive rail.⁽¹⁾ Or it being connected to live. There is no way to protect against everything, and in this case the protection provided is only addressing only a very minor issue. There is an old saying: “any idiot can build a bridge that stands, but it takes an engineer to build a bridge that barely stands.” A part of a reasonable design is reducing costs where they don’t really improve much.

This is not a suggestion to skip that protection completely. But re-evaluate if you really believe you need it. And if you need CH213K instead of a generic schottky diode. Even at 1 A the difference between CH213K and SS34 is 4.6 V vs 4.8 V. Both are way above 3.3 V used by ESP32 and with a linear regulator you need to drop that voltage by converting power to heat. It’s a choice how it is distributed between the diode and the voltage regulator.

Again I stress it’s not my point to tell you to certainly not choose CH213K. It’s to trigger rethinking the design.

For the SMAJ0.5A I think yeah it makes more sense as you say, to be in front of everything. I read about how to place the CH213K and the SMAJ0.5A together and the suggestion was the diode to come after it, that's why I went that way.

SMAJ0.5A is a TVS meant to suppress very short “random” transients that may appear on the power lines. If we put it behind CH213K, it’s still going to protect the rest of the circuit. The problem is that CH213K remains unprotected. Unlike a true diode, which does not have connection to the GND and therefore never sees the full voltage spike, CH213K does.

In the CH213K datasheet it says that is equivalent to a schottky barrier diode plus a self recovery fuse, so I took that as it has a fuse...?

I still don’t understand how comes companies spending enormous money on putting a chip in production can’t hire a decent translator.⁽²⁾ But it’s very common among Chinese companies that their datasheets and manuals looks as if they went through poor man’s version of Google Translate. This one is not an exception and it’s plagued with errors suggesting the translator didn’t know English or electronics. The use of word “fuse” is probably some bad translation.

The purpose of a fuse is to:

• … reliably break the circuit in an overcurrent situation.
• … reliably increase resistance (and therefore reduce voltage and current seen by the circuit) if current rises too much. This is the a special case for polyfuses.

There is no option for it not doing so, because e.g. logic didn’t click or a defect in silicon disturbed an amplifier.

CH213K senses current and uses an amplifier to control the MOSFET. The function is primarily to not overheat the CH213K itself. The closest it gets to fuses is being an unreliable “egoistic polyfuse.”

⁽¹⁾ A wall adapter should not be earth-referenced at all. But out of the two possibilities, GND being bound to earth is relatively harmless. Compared to VBUS, which is going to be catastrophic the moment we connect this device to something else.
⁽²⁾ This is different from some smaller vendors, in particular consumer products manufacturers or eBay sellers, where I can completely understand that it’s just small companies that just try to live on an international market. My English is far from perfect, and I remember how Polish companies tried to appeal to western audiences in the past. But deploying a custom chip entails costs so huge that hiring a translator is 0.05% of the costs, and this particular company is not some small player. I expect the CEO makes monthly 100× more than the translator.

Ok, i got what you're saying. If that's the case, how should i design the USB-C power section, considering it will only be a single power input, the circuit uses both 5V and 3.3V. For 3.3V the designer used AMS1117. The same USB-C is also used to program the ESP32, with a press of 2 internal buttons.

So if i may ask, what components would you use or eliminate for this that make sense for actual protection (that may be needed for USB-C including PD power source) and cost efficiency, how would you design the USB-C power section?

[golden_labels]

What is the tolerance on that 5 V? As Ian.M mentioned earlier you’re can’t expect to get full 5 V at all times.

If both around-5-V and 3.3 V are needed, I’d just connect it straight to the VBUS line. Done.

If there is a chance that the device itself fails⁽¹⁾ and I’d like to protect computer’s USB port, I would add two polyfuses after the power branches off to AMS1117: one for each branch. But I feel like this is already a kind of an extra protection. Usually we see nothing like that in devices. They get power straight from USB and that’s it.

The reason for using two separate fuses is that this way they can be selected to carry lower currents. Depending on the choices and circumstances that means they can either have lower current rating (and faster trip time) or have more headroom (and lower losses).

⁽¹⁾ Including something connected to the device downstream, if that’s the case here.

[Unixon]

Haha, how does that make me AI agent, I am not

Diodes obviously pointing backwards and a 5V bus connection bypassing protection circuit plus question phrasing makes me think this none of this is created by a human.

[golden_labels]

Diodes obviously pointing backwards and a 5V bus connection bypassing protection circuit plus question phrasing makes me this none of this is created by a human.

Did you read the original post, where it’s explained that’s being done by a person hired by OP? And that it has been suggested that person was relying on an LLM to generate the circuit?

[xvr]

Just FYI: CH213K is out of stock on LCSC and not available at all on digikey and mouser

Thanks for your thoughts on stock, jlcpcb seem to have 2000+ pieces

JLCPCB ran out of CH213K stock a week ago. They currently have 708 units remaining. It appears demand for this product is very high, and it's selling out quickly.

[schotkky]

What is the tolerance on that 5 V? As Ian.M mentioned earlier you’re can’t expect to get full 5 V at all times.

If both around-5-V and 3.3 V are needed, I’d just connect it straight to the VBUS line. Done.

If there is a chance that the device itself fails⁽¹⁾ and I’d like to protect computer’s USB port, I would add two polyfuses after the power branches off to AMS1117: one for each branch. But I feel like this is already a kind of an extra protection. Usually we see nothing like that in devices. They get power straight from USB and that’s it.

The reason for using two separate fuses is that this way they can be selected to carry lower currents. Depending on the choices and circumstances that means they can either have lower current rating (and faster trip time) or have more headroom (and lower losses).

⁽¹⁾ Including something connected to the device downstream, if that’s the case here.

I think the absolute lowest is 4.5V. All chips are 3.3V and the DAC and ADCs use both.

So just 2 polyfuses, ok. I understand this take on it, USB tends to be pretty stable and accurate. My main concern to use at least something is cause of the PD on the USB-C power adapters in case something happens. Will the polyfuses also be useful for that too or am I overthinking the whole thing...

[schotkky]

Just FYI: CH213K is out of stock on LCSC and not available at all on digikey and mouser

Thanks for your thoughts on stock, jlcpcb seem to have 2000+ pieces

JLCPCB ran out of CH213K stock a week ago. They currently have 708 units remaining. It appears demand for this product is very high, and it's selling out quickly.

Need to keep an eye on it for future stock if I am about to use that component, atm still figuring out the best approach for power section, even though it seems simple

[xvr]

Need to keep an eye on it for future stock if I am about to use that component

You can pre-order from JLCPCB and your components will be reserved for you.

[schotkky]

Need to keep an eye on it for future stock if I am about to use that component

You can pre-order from JLCPCB and your components will be reserved for you.

That's good to know. First need to sort out the circuit with final components

[Unixon]

Did you read the original post, where it’s explained that’s being done by a person hired by OP? And that it has been suggested that person was relying on an LLM to generate the circuit?

Yes I did, but I suppose OP should have immediately recognized the absurdity of that schematic.

[schotkky]

Did you read the original post, where it’s explained that’s being done by a person hired by OP? And that it has been suggested that person was relying on an LLM to generate the circuit?

Yes I did, but I suppose OP should have immediately recognized the absurdity of that schematic.

That's the thing, I already recognized a few but didn't expect an "engineer" could make such basic mistakes and I was a bit lenient when going through the schematics cause of that reason. Some of the mistakes I caught by accident even

[unikeyname]

JLCPCB ran out of CH213K stock a week ago. They currently have 708 units remaining. It appears demand for this product is very high, and it's selling out quickly.

Yeah the demand for that is very high since generally WCH's reference design and available repo in the wild kinda love using that chip. For mass production (>= 2 reels)you will need to wait like 2+ week if you do it through agent, speaking from first hand experience.

That chip is certainly not the direct replacement for bi-unidirectional TVS, so you are supposed to have those diodes if you expose them for the noisy (SMPS for example) voltage source input, or actually, any voltage input in general.

CH213K senses current and uses an amplifier to control the MOSFET. The function is primarily to not overheat the CH213K itself. The closest it gets to fuses is being an unreliable “egoistic polyfuse.”

It's 100% not replacable for any kind of (resettable) fuse. Although now look back at all those contents, I feel a bit at fault for suggesting such chip for this level of begineer, albeit this is very much a junior friendly chip. OP lacks a bit of reading comprehension ability for the datasheet, maybe you want to hone your basic electronic skill further..

Or you could practice more through trial and errors! Any kind of effort to move forward is welcomed.

[schotkky]

It's 100% not replacable for any kind of (resettable) fuse. Although now look back at all those contents, I feel a bit at fault for suggesting such chip for this level of begineer, albeit this is very much a junior friendly chip. OP lacks a bit of reading comprehension ability for the datasheet, maybe you want to hone your basic electronic skill further..

Or you could practice more through trial and errors! Any kind of effort to move forward is welcomed.

I have some basic knowledge, but I hired someone else to design for me and came here for a reason. So any other straightforward input or suggestion is welcome.

BTW, you shouldn't feel bad for suggesting that chip.

[unikeyname]

idk but people are a bit mean in here. They are quite ready to flame for superiority feeling and some of the reply has a bit negative connotation somehow you know.

[Unixon]

idk but people are a bit mean in here. They are quite ready to flame for superiority feeling and some of the reply has a bit negative connotation somehow you know.

No, it's just a little bit of bad luck

Well, you saw the basics: https://www.eevblog.com/forum/beginners/usb-c-power-supply-protection/msg6262288/#msg6262288