USB isolator project

Posted on Thursday, May 8th, 2014 in project logs by DP


Scasagrande shared his USB isolator project, that is available at Github:

I received inspiration for the project from other USB isolator projects on the internet. Mine features a few improvements to many of them:

  • USB-B and micro-USB connectors on the host (upstream) side
  • In addition to accepting power via the barrel connector (which goes through a basic 7805 low dropout linear regulator), there is also a micro-USB port for device-side power. This allows you to use the now common cell phone charger cube to power your isolated device
  • Push-button switch for easy USB disconnection events
  • Bypass jumper for linear regulator, allowing for 5V power from the barrel connector
  • Encased in a small box to protect the IC from the world

Via the project log forum.

Check out the video after the break.

This entry was posted on Thursday, May 8th, 2014 at 9:25 pm and is filed under project logs. You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.

14 Responses to “USB isolator project”

  1. jbeale says:

    Very nice! This can come in handy when interfacing to sensitive circuits (like debugging capacitive-touch systems)

  2. Bryan says:

    Is the power also isolated on this board? I have not had a chance to look over the schematics but the photo of the bottom side of the board only shows the Analog Devices digital isolator and the LDO regulator plus some other smaller parts.

    • The board is isolated, yes, so long as your power supply that you connect to the board is isolated from your PC. The board only creates isolation for the data lines; it is up to you to provide an isolated power source such as a 2-prong wall-wart adapter (eg: a cell phone charger cube).

      • Bryan says:

        Most power sources including most wal-warts are not going to be isolated. Providing power to devices on the isolated side of the ADuM4160 will provide a possible ground path destroying isolation unless the power source is also isolated e.g. a battery or an isolated DC-DC converter, or one of the ADUM540x family or ADUM5200 etc, right?

      • I might in the future change things up, but as of right now you need a power supply isolated from your PC.

      • I just checked out a random wal-wart on my table and it seems to be isolated.

  3. jbeale says:

    “Most power sources including most wal-warts are not going to be isolated.”
    Say what!? *All* low-voltage DC power sources should be fully isolated from the AC line. If it isn’t, it would be a safety hazard, and could not be approved for sale (UL, CE, TUV etc). Even the cheap ebay wall-wart outputs are isolated from the AC side, at least in my experience. If you have any wall-warts that are not isolated (show any measurable resistance from either AC plug to either polarity output) you should throw it away. Actually, first a safety complaint with the manufacturer.

  4. Oldtimer says:

    Isolation is in the “eye of the beholder”. For instance, medical isolation requires withstanding 4kV DC for 1 minute with less than 500uA leakage (implies high-integrity dielectric) and less leakage that 500uA at highest voltage, usually 240VAC at 60Hz (implies low capacitance). Standard isolation HiPot industrial limits are somewhere around 1 to 2.5kV, some perhaps less. In any case, ohmmeter using a low excitation voltage is not going to test for isolation integrity. See Wipro: “Medical Devices Isolation-“How safe is safe enough” ” white paper for a pretty comprehensive treatment of the subject.

    • Of course. It doesn’t feature any routed out slots, but it at least meets the requirements in the datasheet. I believe the closest the ground planes come together in my design is from one side of the IC to the other.

      • Matt Bennett says:

        For decent isolation that passes UL, take a look at supplies in major PC makers computers as an example. Slots, 5 mm spacing. I’ve also found that sharp edges on pads can make it easier for corona (and then a discharge) to form.

    • jbeale says:

      You are of course correct that there are degrees of isolation, and cheap consumer items are not medical-grade devices. I understood the original comment to mean connected as opposed to floating, something a basic low-voltage ohm-meter would register. In the other extreme, if you need (for example) less than 5 pF capacitance to ground, a standard medical-grade supply might not work either, and you’d probably need to use a battery.

Leave a Reply

Notify me of followup comments via e-mail. You can also subscribe without commenting.

Recent Comments

  • Edward Mallon: I've been oversampling those thermistors with the ADC and some pin-toggle noise for a while now, and that method is good. But lately I've been...
  • Marko: armature -> amateur
  • Crawford: Dibs,
  • bill: a sunny Sunday here
  • Iain NoSpam: A very random comment from a very random commenter