Login
You're viewing the functional.cafe public feed.
  • Sep 6, 2026, 9:30 AM

    Granted I'm an amateur analog designer so take the following with a grain of salt ...

    I've been trying to implement a USB 2.0 High Speed PHY. Not one that just works in most cases, but one that would actually pass compliance (and yes, that's a big difference) and meet specs ( also different, not all specs are tested during compliance tests )

    And damn, this turned out to be way harder than I thought. It's 26 year old tech and yet present quite a few challenge 😅

    💬 1🔄 0⭐ 4

Replies

  • 💬 1🔄 0⭐ 0
  • Sep 6, 2026, 10:40 AM

    @whitequark Yeah, at this point I'm pretty sure a 5G serdes is simpler "conceptually". But doing that in GF180mcu ( which despite its name actually only has 350nm transistors ) I'm not even sure if it's doable 😅

    💬 1🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • Sep 6, 2026, 11:29 AM

    @whitequark Yes, that would be the absolute only way to do it.

    But you'll still some switching elements going at full rate, some kind of pre-amp and then the comparator will need to settle in a fraction of the UI even with staggered clock phases.

    I don't have much experience as said before but I did hear second hand that Bunnie did design a 5G serdes on 180nm (with actual 180nm devices) and it wasn't easy and didn't wish it to anyone else ...

    💬 0🔄 0⭐ 0
  • Sep 6, 2026, 6:56 PM

    @tnt ~25 yrs ago I worked for Intersil and we had a meeting with a little analog design house who'd been part of Intel's original product design effort about buying a USB 2.0 HS PHY for integration into some of our WiFi chipsets. Spent all day going over their IP design. Those things are not simple. IIRC that project ended up a dead end, but the little design house got bought up by SMSC and then later by Microchip. Now that stuff is all commodity and everyone has forgotten the details.

    💬 0🔄 0⭐ 0