0

this hub was for a friend of mine, he asked me if I could repair it for him, but when I opened the case, I was shocked! behold the fried chip and PCB! enter image description here

enter image description here

he said he want the chip to be replaced, because it was working so good for him and he never encountered any problem during copying multiple files to different flash drives (unlike other usb hubs that stop file copying when a new flash drive is added). the chip is hs8836a. there is a small hole between pin# 12(DCP) and 13(VCC). is there any way to improve this design so it won't happen again? I was searching for some good usb hub ICs and I have found gl850g and FE1.1S. which one is better?

M A K
  • 79
  • 5
  • 3
    That looks like an example how an USB hub should not be done. We also don't know what happened when the chip fried so we don't know how to prevent it. Buy a better USB hub, it is a miracle that this even works, so the other hubs must be much worse. – Justme Aug 28 '23 at 20:08
  • 3
    That's not a USB 3 hub. It's a USB 2 hub with blue connectors for better looks. – TooTea Aug 29 '23 at 11:47

2 Answers2

8

This is the most bad-quality modern computer electronics PCB I have ever seen. And I've seen some fecal matter. This is the fecaliest matter of all.

This board should have been rejected at the factory, and shredded. See how the copper layer throws bubbles? That means it's broken, because not only does the copper layer become unreliable, it also wildly changes the characteristic impedance of the lines, and that's important for USB, and especially for SuperSpeed USB.

Speaking of which: even if this board wasn't damaged, it would be so bad that I'm surprised it worked for your friend so often. I guess that is a testiment to the robustness of USB transceivers, and of the protocols atop of USB's physical layer that can recover when errors happen.

Because this board does everything wrong.

  • USB mandates protection diodes on the inputs. None.
  • USB DP and DN lines are pseudo-differential and controlled-impedance, but relatively relaxed. The lines on the board are layed out as if these were very slow signal lines (a couple kHz), and not lines that carry up to 480 Mb/s. You cannot just make one wider than the other, because it saves you a dollar on a million devices produced due to less etchant usage. That is just a bad idea, it makes the signal quality worse, without real reason. This is like if someone who built least-quality toys in the 1980s was allowed to design electronics post-2012. Like. What does that person do for a living? Certainly not lay out high-speed PCBs.
  • While really bad, this is limited in effect, because that trace is short relative to the signal bandwidth. Much worse, the differential SuperSpeed lines: You cannot reasonable lay them out on a single-sided PCB. It just doesn't work well – your traces become bad to produce and route, because there's no ground plane with which they would form a differential coupled microstrip line with ground, and still need to have the USB-specific characteristic impedance, and your traces need to become very large. And these traces are nowhere close to what a controlled-impedance line should look like: varying in thickness, varying in distance to each other, laid across worst-quality substrate, and deliminating copper… This thing definitely reflects about much signal power back at the transmitter as it lets through.
  • There's no current limiting at all. That's plain illegal. Your USB devices should be protected against short.
  • Mechanically, these USB connectors aren't even soldered in on both sides. That's just sloppy. And bad.
  • The connectors are blue. And according to the USB standard, that means "USB 3 or newer", including SuperSpeed. Now will you look at that: all but the bottom connector only have the USB2 lines connected. This is fraud. Your friend bought a hub thinking it did USB3, whose connectors explicitly state through color "we're USB 3", but only the bottom one is USB3.

In other words, if you gave me that hub, new, for free, I would make sure it gets brought to the correct electronics disposal agencies.

There's nothing to save about it. It was totally broken when it still worked, and I'm happy it finally gave up the ghost, so that the amount of data loss your friend experienced while using it to connect to storage devices was limited.

Marcus Müller
  • 88,280
  • 5
  • 131
  • 237
  • 1
    My initial thought was that "bubbles" happened when the USB got fried. Is it possible the copper layer could have passed visual inspection at time of manufacturing but de-laminated over time? – Chester Gillon Aug 28 '23 at 20:15
  • Well said! I believe USB "mandates ESD protection diodes" is not entirey true. If there is a certain ESD level that must be handled, the chip may handle it internally and if it doesn't then some external components may be required. But a spec mandating that diodes must be applied means they must be applied even if they are not needed :) – Justme Aug 28 '23 at 20:36
  • 1
    @ChesterGillon Unless I'm missing something, that ground routing is abysmal and has plenty of narrow weak spots - I can't imagine that frying the board could have caused such severe damage in the nice, wide parts of the planes while sparing the really narrow parts. – nanofarad Aug 28 '23 at 20:38
  • 3
    One minor point, based on [this](https://electronics.stackexchange.com/questions/107669/usb3-with-fewer-wires/266990#266990) and other sources, the USB 2 and USB 3 parts can operate independently, so there is no protocol reason why the bottom port would not work as a USB 3.0 device despite the USB2 lines going through the hub. – Tom Carpenter Aug 28 '23 at 21:08
  • 3
    @ChesterGillon these bubbles look like "air or water got trapped under the copper laminate while gluing it to the base substrate". My suspicion is that the board material itself is already rejects and was bought by the PCB-manufacturing company for the simple fact that it was by far the cheapest thing. – Marcus Müller Aug 29 '23 at 00:33
  • @TomCarpenter need to correct that. USB spec seems to say that USB2 and USB3 hubbing are just parallel, independent things in the same case. – Marcus Müller Aug 29 '23 at 10:15
  • 1
    Amusingly: [this answer](https://electronics.stackexchange.com/a/634414/53368) from last year mentions this very type of hub - one port USB3, others USB2. – Tom Carpenter Aug 30 '23 at 12:31
0

The IC was likely blown due to ESD or transient voltage spike. As others point out there is no protection outside of the IC's own ESD rating.

At the very least I'd add diode protection or consider opto isolation in a better design.

  • 5
    Why would you add opto-isolation to a garbage consumer product? It might be simpler to buy a product that is better to begin with, than to spend money on polishing garbage. – Justme Aug 28 '23 at 21:25
  • 3
    USB2 opto-isolation costs a multiple of this hub and is technically not without challenge. – Marcus Müller Aug 29 '23 at 00:31