Component C57 beneath the "blank" port doesn't look right to me.

The traces in that area appear to go toward the non-working ports as well.

Start checking F1 ~ F4, likely they are old school micro fuses, which burns if a short or overload happens.

Newer cards come with PTC or "fuses" that restore continuity when they cool down

P.S. Note that apparently there are five ports but only four fuses, so two ports must be connected to the same line, which is consistent with your problem

Lylat1an wrote on Yesterday, 22:07:

Component C57 beneath the "blank" port doesn't look right to me.

no unpopulated ports, its a crystal, cap is crooked but soldered ok

The Solutor wrote on Yesterday, 23:43:

Note that apparently there are five ports but only four fuses, so two ports must be connected to the same line, which is consistent with your problem

concurred, F4 is smoked
vico measure resistance of F4, or just solder a wire below it

Indeed, fuse F4 is open; thanks to everyone for your help

Those aren't fuses. They're zero-ohm resistors, as shown by the marking "000". Manufacturer cheaped-out and didn't actually install fuses. So this board has no current limiting or circuit-breaking for attached USB devices.

F4 resistor might have failed open-circuit because it burned out due to excess current draw on the associated port. Likely to happen because the manufacturer didn't even use an appropriately-sized part - those things are probably only good for <500mA.

HwAoRrDk wrote on Today, 07:23:

open-circuit because it burned out due to excess current draw on the associated port.

Do you realize that this is, more or less the description of a fuse? In our case a fuse that proved itself on the field.

If they tested those bridges and realized that the break around the intended current, kudos to them to have used a bit of lateral thinking to spare some little money.

Let me tell a story, few years ago I used to buy a commercial liquid accelerant to light my stove. It used to cost 2€ /for a 1 liter bottle. Not a gift but acceptable.

Then they stopped to sell that brand and started to sell a different one that did cost almost 5 euros for a 3/4 of liter bottle.

I didn't mind to spend that for something which is kerosene, no matter what the label tells.

So I looked around and on the same shelf few meter apart ther were

1) A "lamp petroleum" can (which is kerosene) less than 2 euro for a liter.

2) "The barbecue / fireplace lighter" (which is kerosene), almost 5 euro for 3/4 l.

3) A bottle of "liquid destainer" (which is kerosene), 9 euro for 1/2 liter.

Aside the obvious laugh about how modern consumerism works and how definition matters, guess what I bought (and still buy this day) 😁

Here we are almost in the same situation

HwAoRrDk wrote on Today, 07:23:

F4 resistor might have failed open-circuit because it burned out due to excess current draw on the associated port. Likely to happen because the manufacturer didn't even use an appropriately-sized part - those things are probably only good for <500mA.

They basically relied on the current capacity of the 0 ohm resistor to be the fuse, it's a dodgy way to do it but it does approximate a fuse. Just not one that is predictable or safe.

The Solutor wrote on Today, 08:43:

Do you realize that this is, more or less the description of a fuse? In our case a fuse that proved itself on the field.

If they tested those bridges and realized that the break around the intended current, kudos to them to have used a bit of lateral thinking to spare some little money.

Yes, but also no. A fuse has defined characteristics on how much current it will hold, and when and how fast it will break current, typically specified by the manufacturer with an I^2t curve. A zero-ohm resistor has none of that. Hell, manufacturers don't even guarantee they're actually 0Ω - typically such a part will have a resistance tolerance specification, which can be as much as ±10%. Yes, it sounds nonsensical that you can have a 10% deviation from zero, but nothing is ever zero resistance (if it is, congrats, you've created a perfect superconductor and a Nobel prize will surely be arriving at your doorstep any day now 😜).

That the part failed in a way that approximates a fuse is just happenstance. This is purely a case of the card manufacturer penny-pinching and installing a $0.002 zero-ohm resistor in place of a $0.10 fuse. When you sell the entire card wholesale for perhaps $15, and your profit margin is perhaps $5, 4x $0.098 saving boosts your profit margin by 39 cents.

HwAoRrDk wrote on Today, 09:55:

Yes, but also no. A fuse has defined characteristics on how much current it will hold, and when and how fast it will break current, typically specified by the manufacturer with an I^2t curve. A zero-ohm resistor has none of that. Hell, manufacturers don't even guarantee they're actually 0Ω - typically such a part will have a resistance tolerance specification, which can be as much as ±10%. Yes, it sounds nonsensical that you can have a 10% deviation from zero, but nothing is ever zero resistance (if it is, congrats, you've created a perfect superconductor and a Nobel prize will surely be arriving at your doorstep any day now 😜).

Yes but also, no, we all know the theory, but I also know the practice.

Unlike what the average Joe thinks, fuses aren't there to protect the device, they are there to protect what's upstream, linees, cables, generators, batteries... from burning, exploding, failing catastrophically. So, usually, they aren't critical (I mean theri specs, not their function).

In that specific case a 600mA~1A fuse would be likely the right choice, but is not like if you used a 2 or 3 Amp fuse your PC would instantly catch fire, in case of a short, they would burn before the circuit's trace just like a 600mA one.

As for the minimal price difference, it's true that cheaping on that doesn't make an huge difference, but if you look the matter as any industry do, anything that cost more money than a different solution is considered wasted money, it's matter of principles rather than quantity from that POV.

That said maybe the project had self restoring fuses rather than normal ones, in that case they spared more consistent money, while providing an inferior product, in that case the behavior is not condonable.

I get the impression you're a fan of Muntzing. 😉

The Solutor wrote on Today, 10:19:

Unlike what the average Joe thinks, fuses aren't there to protect the device, they are there to protect what's upstream, linees, cables, generators, batteries... from burning, exploding, failing catastrophically. So, usually, they aren't critical (I mean theri specs, not their function).

Indeed. But, their spec can be critical. A fuse that takes longer to blow than the components it's meant to protect take to burn up/overload/etc. is pointless. If a 1A-rated fuse takes 10 seconds to blow at a sustained 4A overload, but some upstream components release the magic smoke after 5 seconds at that load, it's an incorrectly-specified fuse.

The Solutor wrote on Today, 10:19:

As for the minimal price difference, it's true that cheaping on that doesn't make an huge difference, but if you look the matter as any industry do, anything that cost more money than a different solution is considered wasted money, it's matter of principles rather than quantity from that POV.

Err, I was trying to make the point that a minimal price difference does make a huge cost saving. 😜 A production run of 100,000 units at a cost-saving of $0.39 each amounts to $39,000 extra profit.

The Solutor wrote on Today, 10:19:

That said maybe the project had self restoring fuses rather than normal ones, in that case they spared more consistent money, while providing an inferior product, in that case the behavior is not condonable.

Yeah, it's common that USB ports are protected by resettable PTC fuses. I suspect that's what should have been used on that card. Also often these days thermally-protected current-limiting load switching ICs that can raise a fault signal. In fact, that VT6212 chip has 4 input pins dedicated to such an overload-flagging function. But obviously not used here. 😁

HwAoRrDk wrote on Today, 12:57:

I get the impression you're a fan of Muntzing. 😉

First time I hear of him

Anyway when I was at school, I had an exam in designing a simple PCB it was the very common beginner's circuit to test the status of the 12V cars battery.

Just three leds that tuned on if the battery was too low, normal or charging, well our professor gave us a schematics that was way more complex by the same function circuit I already knew.

In short I designed the PCB using "my" circuit not the one provided, "my" circuit had half the components, so my PCB was the cleanest and most compact of the class, no one noticed, they just tested that it was working as intended, and it was.

Technically I cheated, the exam was about PCB design not project, but I don't feel guilty at all.

Not sure if the above is a Muntzing thing 😀

Indeed. But, their spec can be critical. A fuse that takes longer to blow than the components it's meant to protect take to burn up/overload/etc. is pointless. If a 1A-rated fuse takes 10 seconds to blow at a sustained 4A overload, but some upstream components release the magic smoke after 5 seconds at that load, it's an incorrectly-specified fuse

That's hardly what happens, a short will either burn the fuse (even if it is rated twice the intended current), but assuming you use an exaggerated one or just a wire, you either blow the trace on the card or you trip the protection of the main PSU depending how thick the trace is

A short is a short, there is a brief but strong transient, an overload may be different, but even old machines specced for 500mA usually can provide way more current w/o any problem.

It's just that practice w/o theory may be bad, but theory w/o experience may be even worse.

Err, I was trying to make the point that a minimal price difference does make a huge cost saving. 😜 A production run of 100,000 units at a cost-saving of $0.39 each amounts to $39,000 extra profit.

Sorry I got that wrong, so we were agreeing here.

Yeah, it's common that USB ports are protected by resettable PTC fuses. I suspect that's what should have been used on that card. Also often these days thermally-protected current-limiting load switching ICs that can raise a fault signal. In fact, that VT6212 chip has 4 input pins dedicated to such an overload-flagging function. But obviously not used here.

Yeah, perhaps who designed the PCB was reasonably long sighted, there is a through the hole alternate pads next to the smd ones, but the board could be a reference design provided by VIA, then made by dozen of companies with different take on balancing quality and cheapness.