0

I am looking for an IC solution, ideally, but alternatively also could use a tiny circuit that would allow to accomplish the following feature:

Trigger an event by MCU that will fuse-out/disconnect data lines (D+ and D- in particular). This has to be permanent, irreversible and also data lines must be physically disconnected.

I am not sure if circuits like this exist as off-the-shelf IC solution. What I've found by searching different terms is mostly conventional circuit protection fuses which is _not_ what I'm looking for.

Use case for circuit like this would be when you have to use USB for programming, debugging, data/log transfer and etc. but you don't want to ship your device to end users with USB D+/D- lines connected to anything. This allows to have USB charging and data for debugging/development, but end user doesn't have to worry about "Do I trust this device and is it going to do any harm to my USB host".

I would use this in my future open-source, open-hardware project(s) so there will be a lot of people cracking open this thing and confirming that it's safe for use when you get it "sealed".

Calculon
  • 142
  • 12
  • Just cut the tracks on the PCB or remove links. – Andy aka Aug 04 '18 at 15:40
  • There are ways you can do this by just using lots of current and tiny traces (after all that is what a fuse is). But are you sure it is not easier to just have a internal programming port that you use to program, and use a second connector that only has power pins connected exposed to the user? – Joren Vaes Aug 04 '18 at 15:42
  • 1
    Take a look at the answer to this question. There may be smaller versions of the device more suitable to your application. – brhans Aug 04 '18 at 15:47
  • Keep in mind that for more than base rate charging you'll need the data pins anyway, though not connected to a traditional USB transceiver. It's not clear that average end users will be able to tell the difference until after plugging the device into a computer and seeing that it does not enumerate. – Chris Stratton Aug 04 '18 at 15:56
  • It is possible to choose a 20mA SMD fuse and bipolar transistors to burn the fuse but a jumper seems more prudent. – Tony Stewart EE75 Aug 04 '18 at 15:58
  • Another decent strategy is to make the populated USB connector power only, and provide the MCU's debug USB port on a 4-pin header not populated in production. For developer units you solder the header and make a custom USB cable to run through a notch cut where the case halves meet, for production you can hit the four though hole pads with pogo pins in a press jig. – Chris Stratton Aug 04 '18 at 16:00
  • "Do I trust this device and is it going to do any harm to my USB host" - If it's got a USB port, and I was feeling untrusting, I wouldn't plug it in, regardless of whether or not you told me the USB data was disconnected. If I wanted to power it, I'd use a USB charger. – Tom Carpenter Aug 04 '18 at 16:00

1 Answers1

2

The solution to "fuse off" the USB data lines sounds like an X-Y problem. You should focus instead just exactly what you need instead of trying to dream up something non-conventional that may in fact not work very well.

So the following should be what you address.

  1. Design the USB charging connector with just the connections to support the charging function.
  2. Design a different access method for your testing and programming connection. That could be a second buried USB connector, a set of pads that you use a pogo pin(s) fixture to access or a simple pin header of a different style.
Michael Karas
  • 57,328
  • 3
  • 71
  • 138
  • Yes, and pogo pins have the advantage of working again in the lab if you get a unit back from the field that you need to analyze. Having pads for 0 ohm (or in some cases another low value) resistors to connect the data lines for firmware developer's units can also be handy - being able to get work accomplished with no special fixtures but just a laptop and compact USB cable or two can be very convenient. – Chris Stratton Aug 04 '18 at 15:58