7 ms·

Speaking as someone in the industry who's worked on Type C-enabled gadgets, half the bloody problem is that the Type C products offered by the major semiconduct

Speaking as someone in the industry who's worked on Type C-enabled gadgets, half the bloody problem is that the Type C products offered by the major semiconductor companies are _terrible_. For sake of a simple example, consider designing a rechargeable flashlight. For whatever reason, you'd like it to be rechargeable over a USB-C port. You also want the flashlight to charge as fast as possible from whatever power source you connect it to, because you know people will need to charge it quickly. To do that, the ideal solution would be a single IC that sits on the USB port and monitors D+/D-, CC1/2, and the rail voltages and sets a few GPIOs to tell your battery charge circuit what is and isn't allowed on this port. Inexplicably, _that chip doesn't exist_. It does exist for USB 2.0 (that is, BC1.2), available from several vendors, and works perfectly. There are solutions marketed for Type-C... but they don't touch USB 2.0. That means someone who uses a Type-A to Type-C cable with a beefy older Type-A charger doesn't get fast charging. There are some bigger Type-C products that might work, but they're incredibly complex, physically large, expensive, often in BGA packages, and will blow out every line on your budget. Admittedly it's been about a year since I turned over every stone in the industry searching for my Magical Unicorn Charging IC (like the ones that exist for USB 2.0...), but I do pay attention to new product releases and I haven't seen anyone going in this direction. Probably you could get something customized (at medium volumes) or certainly full custom (at high volumes), but if volumes or margins are low enough to require COTS parts, there really aren't any workable solutions. I am confident that Type-C's market penetration will skyrocket as soon as the difficulty and cost of actually integrating it into a product reaches sane levels.

extralego 8y ago
Wouldn't there always be companies trying to sell the cheapest qualifier? Wouldn't the expensive products you mention that "might work" become more affordable if they were manufactured at a higher rate? (But maybe they aren't because they are harder to sell) >I am confident that Type-C's market penetration will skyrocket as soon as the difficulty and cost of actually integrating it into a product reaches sane levels. Isn't the current market of so many low-cost Type-C devices evidence that the cost is already low?
- Tloewald 8y ago
  The cost of crappy things that don't solve the problem described which leads to the mess we're complaining about is low, yes.
vertexFarm 8y ago
Forgive me for sounding extremely ignorant. I know this is not my area of specialty and I admit I'm about to sound stupid. I want to admit this right off the bat. Can't you just hook up the VCC and ground from any USB charging IC to a type-c socket? It's all +5v right? I don't understand why the type of connector dictates how fast a cable should be useful for charging. That seems like it should be up to the simple gauge and length of the wire and thus how much current it can handle. From what I've heard re: that chart in OP's link that shows the different fast charge rates for different manufacturer's wall warts and cables, most of that is caused by proprietary ICs doing sneaky things like throttling the charge rate when it discovers it isn't connected to an official Huawei cable or whatever through some resistor value going to ground from one of the data lines, or some similar method. Couldn't the IC in the device read the resistance / inductance of the cable and determine through that how much current it could handle? I feel like this is all bad faith shenanigans through proprietary cable DRM. When charging is concerned, a conductor is a conductor assuming there's enough thickness to prevent overheating, right?
- theatrus2 8y ago
  Type-C offers charge rates up to 100W, which means the voltage is ramped to 20V. There is an entire protocol to negotiate power delivery (including who charges who) http://www.usb.org/developers/powerdelivery/ http://www.usb.org/developers/powerdelivery/
  - digi_owl 8y ago
    And the latter part have caught some people unaware, where a battery pack and a phone will ping pong who is charging who until both are dead.
    - chimeracoder 8y ago
      > And the latter part have caught some people unaware, where a battery pack and a phone will ping pong who is charging who until both are dead. Or, if you're really unlucky, it'll attempt to charge your wall. No, I'm not joking: https://twitter.com/chimeracoder/status/941820122406064128 https://twitter.com/chimeracoder/status/941820122406064128
kjeetgill 8y ago
If an equivalent exists for USB 2.0, I'm surprised a company hasn't come it to sweep up the low hanging fruit. Why do you think that is? Is there something in the spec that's holding anyone back? or is it a chicken and egg problem where no one invests in Type C because no one has yet?
- tjoff 8y ago
  According to the spec a USB 2.0 to USB C cable must have a resistor that tells the USB C end to max deliver 0.5 A. If another resistor is used to signal higher currents using that cable could fry peripherals that follow the USB 2.0 spec. Also, I don't know much about USB 2.0 but I thought we had a whole slew of proprietary solutions for fast charging and all were incompatible with each other. So it was quite messy even before USB C came and made everything even worse. For charging though you generally wan't to slow charge as it greatly reduces wear on batteries anyway.
  - AngryData 8y ago
    You only really need to slow charge when you get near to top or bottom end of the battery's charge. From around 20-80% charge you can charge the fuck out of you shit with no drawbacks, but if you try and push the those limits then you start seeing significant wear.
    - tjoff 8y ago
      Source? I thought heat was the primary source of wear. And fast charge produces lots of heat regardless if it's in the middle of the range or not.
      - lpmay 8y ago
        I'm not an electrochemist, but I have been told one of the major wear mechanisms is lithium plating, which is a function of anode voltage and charge current (and probably 100 other things). As I understand, the standard CI/CV charge cycle is very conservative to avoid this complexity. I believe more sophisticated parts like the TI Max charge series get around this by running estimators of the cells anode voltage and can scale charge current appropriately. Sorry no good references to offer off hand, this is all just stuff I've been told by engineers and chemists. Take with a grain of salt.
chx 8y ago
There's no innovation in this sector. It needed Apple to kick Ti's sorry ass to get us to Ti 83 even instead of the Ti 82. It says quite a lot every Thunderbolt enclosure uses this chip: there are no other USB PD controllers (USB PD is used to negotiate Alternate Modes). Another example: you can buy a small and cheap USB C to DisplayPort adapter with power passthrough (< 20 USD). You can buy a (relatively) small and cheap USB C to dual DisplayPort MST hub (<50 USD). You can not buy a USB C to dual DisplayPort MST hub with power passthrough unless it's a full dock, huge in size and expensive (200 USD at least).
- mmagin 8y ago
  Is this because of patents, or is this because nobody wants to do the R&D?
  - chx 8y ago
    The latter. The entire area of vaguely PC peripheral chips died in 2014 when MegaChips took over ST's DP chips. Consider how much a 10 GbE card still costs -- this is the same speed! Apparently, it's not cheap making chips working with 10-20-40 gigabit/s line speeds but the market demands low cost. USB C wouldn't get far if a plain USB C to DP adapter would cost a hundred bucks.
    - civilitty 8y ago
      > Consider how much a 10 GbE card still costs -- this is the same speed! Apparently, it's not cheap making chips working with 10-20-40 gigabit/s line speeds but the market demands low cost. USB C wouldn't get far if a plain USB C to DP adapter would cost a hundred bucks. You can thank the laws of physics for that because at 10+ gbit data rates you start needing crazy internal clock speeds (which can be significantly higher than the input clocks). Since the distance a signal can travel before it loses too much energy to register logic HIGH is proportional to the wavelength, you have to pack everything close enough while still doing what your specs call for. In the case of 10GigE and USB Type C, this means having to move to smaller fabrication processes than most companies are prepared for or even have access to. Most protocols designed to be that fast are either massively parallelized to avoid this problem like PCIe/DDR2+ or very expensive to implement as peripherals/silicon like MIPI3/10GigE/USB Type C.

extralego 8y ago

Wouldn't there always be companies trying to sell the cheapest qualifier? Wouldn't the expensive products you mention that "might work" become more affordable if they were manufactured at a higher rate? (But maybe they aren't because they are harder to sell) >I am confident that Type-C's market penetration will skyrocket as soon as the difficulty and cost of actually integrating it into a product reaches sane levels. Isn't the current market of so many low-cost Type-C devices evidence that the cost is already low?

Tloewald 8y ago
The cost of crappy things that don't solve the problem described which leads to the mess we're complaining about is low, yes.

vertexFarm 8y ago

Forgive me for sounding extremely ignorant. I know this is not my area of specialty and I admit I'm about to sound stupid. I want to admit this right off the bat. Can't you just hook up the VCC and ground from any USB charging IC to a type-c socket? It's all +5v right? I don't understand why the type of connector dictates how fast a cable should be useful for charging. That seems like it should be up to the simple gauge and length of the wire and thus how much current it can handle. From what I've heard re: that chart in OP's link that shows the different fast charge rates for different manufacturer's wall warts and cables, most of that is caused by proprietary ICs doing sneaky things like throttling the charge rate when it discovers it isn't connected to an official Huawei cable or whatever through some resistor value going to ground from one of the data lines, or some similar method. Couldn't the IC in the device read the resistance / inductance of the cable and determine through that how much current it could handle? I feel like this is all bad faith shenanigans through proprietary cable DRM. When charging is concerned, a conductor is a conductor assuming there's enough thickness to prevent overheating, right?

theatrus2 8y ago
Type-C offers charge rates up to 100W, which means the voltage is ramped to 20V. There is an entire protocol to negotiate power delivery (including who charges who) http://www.usb.org/developers/powerdelivery/ http://www.usb.org/developers/powerdelivery/
- digi_owl 8y ago
  And the latter part have caught some people unaware, where a battery pack and a phone will ping pong who is charging who until both are dead.
  - chimeracoder 8y ago
    > And the latter part have caught some people unaware, where a battery pack and a phone will ping pong who is charging who until both are dead. Or, if you're really unlucky, it'll attempt to charge your wall. No, I'm not joking: https://twitter.com/chimeracoder/status/941820122406064128 https://twitter.com/chimeracoder/status/941820122406064128

kjeetgill 8y ago

If an equivalent exists for USB 2.0, I'm surprised a company hasn't come it to sweep up the low hanging fruit. Why do you think that is? Is there something in the spec that's holding anyone back? or is it a chicken and egg problem where no one invests in Type C because no one has yet?

tjoff 8y ago
According to the spec a USB 2.0 to USB C cable must have a resistor that tells the USB C end to max deliver 0.5 A. If another resistor is used to signal higher currents using that cable could fry peripherals that follow the USB 2.0 spec. Also, I don't know much about USB 2.0 but I thought we had a whole slew of proprietary solutions for fast charging and all were incompatible with each other. So it was quite messy even before USB C came and made everything even worse. For charging though you generally wan't to slow charge as it greatly reduces wear on batteries anyway.
- AngryData 8y ago
  You only really need to slow charge when you get near to top or bottom end of the battery's charge. From around 20-80% charge you can charge the fuck out of you shit with no drawbacks, but if you try and push the those limits then you start seeing significant wear.
  - tjoff 8y ago
    Source? I thought heat was the primary source of wear. And fast charge produces lots of heat regardless if it's in the middle of the range or not.
    - lpmay 8y ago
      I'm not an electrochemist, but I have been told one of the major wear mechanisms is lithium plating, which is a function of anode voltage and charge current (and probably 100 other things). As I understand, the standard CI/CV charge cycle is very conservative to avoid this complexity. I believe more sophisticated parts like the TI Max charge series get around this by running estimators of the cells anode voltage and can scale charge current appropriately. Sorry no good references to offer off hand, this is all just stuff I've been told by engineers and chemists. Take with a grain of salt.

chx 8y ago

There's no innovation in this sector. It needed Apple to kick Ti's sorry ass to get us to Ti 83 even instead of the Ti 82. It says quite a lot every Thunderbolt enclosure uses this chip: there are no other USB PD controllers (USB PD is used to negotiate Alternate Modes). Another example: you can buy a small and cheap USB C to DisplayPort adapter with power passthrough (< 20 USD). You can buy a (relatively) small and cheap USB C to dual DisplayPort MST hub (<50 USD). You can not buy a USB C to dual DisplayPort MST hub with power passthrough unless it's a full dock, huge in size and expensive (200 USD at least).

mmagin 8y ago
Is this because of patents, or is this because nobody wants to do the R&D?
- chx 8y ago
  The latter. The entire area of vaguely PC peripheral chips died in 2014 when MegaChips took over ST's DP chips. Consider how much a 10 GbE card still costs -- this is the same speed! Apparently, it's not cheap making chips working with 10-20-40 gigabit/s line speeds but the market demands low cost. USB C wouldn't get far if a plain USB C to DP adapter would cost a hundred bucks.
  - civilitty 8y ago
    > Consider how much a 10 GbE card still costs -- this is the same speed! Apparently, it's not cheap making chips working with 10-20-40 gigabit/s line speeds but the market demands low cost. USB C wouldn't get far if a plain USB C to DP adapter would cost a hundred bucks. You can thank the laws of physics for that because at 10+ gbit data rates you start needing crazy internal clock speeds (which can be significantly higher than the input clocks). Since the distance a signal can travel before it loses too much energy to register logic HIGH is proportional to the wavelength, you have to pack everything close enough while still doing what your specs call for. In the case of 10GigE and USB Type C, this means having to move to smaller fabrication processes than most companies are prepared for or even have access to. Most protocols designed to be that fast are either massively parallelized to avoid this problem like PCIe/DDR2+ or very expensive to implement as peripherals/silicon like MIPI3/10GigE/USB Type C.