6
\$\begingroup\$

I would like to know what is the typical rating of a polyfuse used on modern PCs (desktop or laptop) with USB 2.0 and 3.0. TE Connectivity apparently recommended in 2011 that a single USB 2.0 port (non-ganged) be protected by a polyfuse with hold current 750 mA and trip current 1.5 A, and double that for USB 3.0 that is not charging-enabled, and higher still for a charging-enabled port.

However, I have been unable to find the actual specification of the USB polyfuse ratings for desktop PCs and personal laptops. Many technical manuals I could find for various PC models mention that they do have over-current protection, but for whatever reason invariably neglect to give any further detail. (I recall some of them explicitly stated "polyfuse" but do not have them at hand right now.)

I can find Littelfuse specifications for the above-recommended polyfuses, and the Littelfuse 2016 selection guide recommends using one of its 0805L, 1206L, 1210L, 1812L or USBR polyfuses for USB, and all of these have hold current at most 2.6 A, but I have no idea what PC manufacturers actually use.

If "modern PC" is too broad, I am happy to know the answer just for the Dell Optiplex 780/790, the HP Pavilion p7-1030/1154 and the Lenovo Thinkpad X220/X230, which is around the time that integrated USB 3.0 ports seemed to start appearing (2011). I understand that later PCs can have BC (Battery Charging) 1.2 ports or PD (Power Delivery) ports, which are rated to provide up to 5 A, and so would have much higher overcurrent protection limits, which I am less interested in. My main motivation for asking this question is to know whether plugging a faulty device into an older USB port can ever result in that device drawing a much larger current (but not short-circuit) than the USB specifications provide for, and how much more (before the polyfuse trips). I chose these relatively popular PC models, but if you can only find information on other Dell/HP/Lenovo models, that would also be helpful.

\$\endgroup\$
9
  • 1
    \$\begingroup\$ I just found this Intel article about USB power delivery design issues, which recommends that the over-current protection have a preset value of 2.5 to 3 times the maximum allowable port current. Still no actual PC specifications though. \$\endgroup\$ – user21820 Jul 16 '20 at 15:46
  • \$\begingroup\$ I'm tempted to say; "Just put on a wrist band, pull the mother-board out, and see what protection they have put on the USB ports..". \$\endgroup\$ – Vinzent Jul 16 '20 at 18:56
  • \$\begingroup\$ Several people who actually design products with USB ports tell you that USB does not use polyfuses but you refuse to believe them. I guess the only way to actually answer this question is if a Dell or Lenovo or HP engineer who worked on those laptops actually write an answer to this question \$\endgroup\$ – slebetman Jul 17 '20 at 3:29
  • 1
    \$\begingroup\$ State of the art is to have a solid state circuit breaker for each USB port. This adds cost, of course. The system designers cut this corner not infrequently. \$\endgroup\$ – Nick Alexeev Jul 17 '20 at 3:33
  • \$\begingroup\$ @slebetman: Or someone opened one of those PCs up and figured out what they used. I'm sure there are many people out there who have done just that. I don't have the time and expertise to reliably trace such complicated circuitry. \$\endgroup\$ – user21820 Jul 17 '20 at 6:46
10
\$\begingroup\$

Sorry, but all answers are incorrect. Vast majority of PC mainboards do use polyfuses (aka resettable fuses) to protect USB ports from overcurrent and prevent damage from melting wires and connector's pins. Below are two examples, for ASUS Z170-A mainboard enter image description here And for ASUS P9X79LE enter image description here

USB specifications define that "high-powered hosts" must supply AT LEAST 500 mA for USB2, and 900 mA for USB3 ports (and more for Type-C). At least. General rule of design is consumer safety and to reduce OEM liability for setting fires, so the designers tend to limit the available current. High-side switches are expensive, and require even more expensive wire and extra pin on CPU or EC (or whatever). So simple polyfuses do their job quite economically.

Also, 95% (or (99?) of all usb hubs do use polyfuses as well.

As you rightfully found, maindoard use polyfuses at about double the minimum, some can supply up to the level of entire +5VSTBy limit (2.5 A and above).

If someone still has a concern about newer mainboards, here is a snippet of Gigabyte X570 AORUS MASTER (released sometime in 2019), with polyfuses circled: enter image description here

\$\endgroup\$
8
  • \$\begingroup\$ Thanks! Can you state the date that these ASUS motherboards were introduced, so that it is easy to compare with other models' ages, without having to google them? And do you have any comment about the answer that claims that the X220 uses the TPS2069 power switch rather than a polyfuse? \$\endgroup\$ – user21820 Jul 17 '20 at 6:54
  • \$\begingroup\$ @user21820, yes, some high-end business-class (and enterprise-class) products do use more expensive high-side switches. Premium HP products use them more frequently. Laptops are more concerned with power limits of their battery, and tend to be more accurate in current limiting. \$\endgroup\$ – Ale..chenski Jul 17 '20 at 7:07
  • \$\begingroup\$ @user21820, regarding board introduction, are you banned from Google, or what? Initial BIOS release for Z170 is listed as 2015/08/10 \$\endgroup\$ – Ale..chenski Jul 17 '20 at 7:08
  • \$\begingroup\$ I wouldn't label other answers as "incorrect": polyfuses are a rare beast nowadays. Expensive stuff uses specialized ICs, and cheap stuff often uses no protection at all. I personally only own two USB host devices with polyfuses, about 10 with ICs (including Pi 4) and a whole drawer of garbage-grade USB hubs with nothing at all. \$\endgroup\$ – Dmitry Grigoryev Jul 17 '20 at 7:25
  • \$\begingroup\$ @DmitryGrigoryev: I don't know why I've to repeat so many times that my question is about the 3 most popular brands of Personal Computers, not hubs or chargers or RPIs. \$\endgroup\$ – user21820 Jul 17 '20 at 7:37
5
\$\begingroup\$

Not all computers necessarily use a polyfuse to provide overcurrent protection for USB. They may be using a dedicated IC known generally as a USB power switch.

If you know the specific USB power switch used, you can look up the datasheet and find the typical trip current. Finding this is tricky, as you'll need the schematic which is typically proprietary and unpublished. Otherwise, you can look at the PCB and figure out what power switch is used.

To give a specific example, I searched for a ThinkPad X220 schematic and found an alleged copy on a forum. From there, I found that the power switch used is the TI TPS2069. If you take a look at the electrical characteristics in the linked datasheet, you will find that the typical overcurrent trip threshold is 2.85A, with a minimum of 2.3A and a maximum of 3.4A.

\$\endgroup\$
3
  • \$\begingroup\$ Thank you very much for your post. Yours is the only one so far that attempts to answer my question. I'll go over the link and get back to you. =) \$\endgroup\$ – user21820 Jul 17 '20 at 5:03
  • \$\begingroup\$ I believe that you misinterpreted the linked PDF. The test you cite is actually for the "Overcurrent trip threshold", which is concerning the trip current. That is why they used a current ramp. They found that in the worst case it could reach 3.4 A before tripping. The hold current is the "Short-circuit output current". Page 1 also labels TPS2069 as "1.5 A" and says current limit is typically 2.1 A in agreement with the table. But thanks for finding the actual power switch used! Can you link the alleged X220 schematics copy so that I can have a look at it? \$\endgroup\$ – user21820 Jul 17 '20 at 6:44
  • 1
    \$\begingroup\$ @user21820 A google search found this. The USB is on page 100 of the X220 schematic, where U23 is the IC referenced. I also edited to be less assertive about polyfuses in light of the other more-informed answer. \$\endgroup\$ – user1599692 Jul 17 '20 at 13:43
4
\$\begingroup\$

There are no fuses in the general USB 3.0 controller (unless you count an electronic fuse circuit as a fuse). The fusing is provided by a mosfet that is built into an IC (host controller or external mosfet) with a current level detection circuit. If the current goes beyond the spec, then the circuit is shut down. With USB 2.0 and earlier the current limit is provided by the host controller and that limit is 500mA.

enter image description here Source: https://www.usb.org/sites/default/files/usb_20_20190524.zip (USB 2.0 ECN VBUS Max Limit.pdf)

A device cannot sink more current than is sourced, and it is my understanding (based on observation of shorted USB ports that I have shorted) that all ports have some kind of over current detection (and will flag the port in the OS when the current is overdrawn and shut down power to the port) .

Here is a good question on the current outputs of each spec: https://superuser.com/questions/1131515/usb-c-to-usb-a-cable-usb-3-0-with-high-current-at-2m-length If the manufacturer does not report which port belongs to each spec, then there isn't a good way to find out unless the circuit is reverse engineered. Many manufacturers produce USB ports that are out of spec.

\$\endgroup\$
20
  • \$\begingroup\$ Do you have evidence that the 3 biggest PC manufacturers that I listed produce USB ports that are out of spec? I do not care much about obscure manufacturers. \$\endgroup\$ – user21820 Jul 16 '20 at 15:48
  • \$\begingroup\$ Of course the device cannot draw more current than the host lets it. But even the USB specifications do not stipulate that the host must limit the current drawn from the USB ports, and I can find recommendations but not specifications. If those 3 PC manufacturers do follow prevailing recommendations, then I already have a good estimate on the limits. That is why I ask this question, explicitly to ask for actual specifications. Surely some people who work in this line will know what polyfuses are actually used with USB ports. \$\endgroup\$ – user21820 Jul 16 '20 at 15:54
  • \$\begingroup\$ By the way, I obviously have no capability to judge the accuracy of your answer, but your first point seems suspect, given that the Intel recommendations I just noted in a comment on my question explicitly show (section 2.2) their suggested topology for power distribution to USB ports, where there is in fact an over-current protection device directly in series with two ganged USB ports. I did not use the term "fuse" anywhere, because these are polyfuses, not fuses. \$\endgroup\$ – user21820 Jul 16 '20 at 15:58
  • 1
    \$\begingroup\$ @user21820. USB specifications are the only issue that matters here, as they are specific to USB standards, which has nothing to do with the product they are embedded in. PC's, tablets and cell phones are normally host devices able to charge or power-up another smaller device, yet all of them can request a charge from a device able to charge it. This is handled by USB protocols for charging. \$\endgroup\$ – user105652 Jul 17 '20 at 1:44
  • 1
    \$\begingroup\$ -1. Too many wrong statements. Most regular PC maniboards do use polyfuses. If you look carefully, the "function (out)" lists the current of 500 mA as MINIMUM. USB host controller does not provide any current limiting, it is a function of board, per designer choice. \$\endgroup\$ – Ale..chenski Jul 17 '20 at 6:13
3
\$\begingroup\$

I would like to know what is the typical rating of a polyfuse used on modern PCs (desktop or laptop) with USB 2.0 and 3.0.

My main motivation for asking this question is to know whether plugging a faulty device into an older USB port can ever result in that device drawing a much larger current (but not short-circuit) than the USB specifications provide for, and how much more (before the polyfuse trips)

These are conflicting requirements. 'Typical' ratings tell you nothing about the maximum possible. Even having information on a specific laptop model doesn't tell you, because you don't know that the manufacturer didn't change the circuit at some time (which is very common).

If you are designing a product then you surely don't want to limit its use to particular computer models and productions runs, and you don't even know that the user won't plug your device into something else with a USB port. So you should not assume that the host will provide any protection. One thing is for sure though. If the USB port is protected by a polyfuse (which I know some are) then the maximum possible current must be a lot higher than the hold current, which should be similarly higher than the USB port's rating.

If you need actual numbers then I suggest testing as many computers (and other host devices) as you can, using a USB load tester. Alternatively you could start an online survey, or look for existing surveys asking people to report the maximum current their USB ports can provide. This will provide more accurate data than polyfuse specs, since the actual trip current and time is dependent on its thermal characteristics when installed in the device.

\$\endgroup\$
8
  • \$\begingroup\$ I am not designing any product. I am just an ordinary user of devices. Some of them are cheap and I am legitimately worried that they may for whatever reason (shoddy design or malfunction) draw more current than they should. If plugged into an older PC's USB port, this ought to trip the over-current protection, so I want to know what ratings the polyfuses have. And I am not so rich as to have many computers to experiment on by trying to overload their USB ports... \$\endgroup\$ – user21820 Jul 16 '20 at 16:28
  • \$\begingroup\$ Then you only have to know what current your computers can deliver. Buy a USB load tester (very cheap on eBay) and you can find out. Or just don't worry about it. If your PCs are well designed then they should not be damaged by a device that draws more current than it should. \$\endgroup\$ – Bruce Abbott Jul 16 '20 at 18:26
  • \$\begingroup\$ The power supply of most PCs according to specifications can provide up to roughly 15 A on the 5 V rail. That's way bigger than what I think the over-current protection for the USB ports are. Otherwise I wouldn't have asked this question. Your last sentence is what I hope is true, but I am driven by curiosity since I am unable to find a single bit of information to verify that! \$\endgroup\$ – user21820 Jul 16 '20 at 18:30
  • \$\begingroup\$ Without poring over the schematics for your computers or having the exact same model ourselves we can't tell you either. The only way to find how much current they can really deliver is to test them. \$\endgroup\$ – Bruce Abbott Jul 16 '20 at 18:36
  • 1
    \$\begingroup\$ Sorry, I don't have any PCs that modern (The DELL Dimension 5150 I am using now is 15 years old). I looked at a couple of spare motherboards. One had 1.3A polyfuses, the other 1.6A. I have also removed a lot of polyfuses from old motherboards - most are between 1.3A and 2A. But that information is completely useless to you. You might do better googling for user tests on machines similar to yours. \$\endgroup\$ – Bruce Abbott Jul 16 '20 at 19:31
1
\$\begingroup\$

No modern PC would use a polyfuse. They use a high-side switch with built-in current limiting. The limiter serves three functions: power domain control, overcurrent protection, and signalling the host when the overcurrent condition occurs (OC trip.) A polyfuse can only do one of those things, and not very accurately or cost-effectively.

The PC port must deliver at least what is required by version of the USB spec it claims to support. For USB2 and earlier, this was 500mA. For USB3.0 this is at least 900mA, and can be up to 2.1A for a battery charging port.

USB3 Type C introduces a more sophisticated approach with options for higher voltage.

The takeaway is, no polyfuse in a PC. And even for non-PC applications, based on my direct experience (set-top boxes) USB power switches are very, very cheap (cheaper than a polyfuse) and more predictable.

\$\endgroup\$
5
  • \$\begingroup\$ I know what you stated in your 2nd and 3rd paragraph, but that is not what I asked about. As for your claim "no modern PC would use a polyfuse", can you please back it up with evidence regarding any of the 6 specific PC models I mentioned? Or are you referring to much more recent models? Older modern PCs definitely did have polyfuses, so you have to qualify your answer anyway otherwise it's misleading or wrong. \$\endgroup\$ – user21820 Jul 16 '20 at 16:57
  • \$\begingroup\$ For example, this thread has people claiming that the Lenovo Z51 (which came no earlier than 2015) has a polyfuse, though that thread is completely unrelated to my question. \$\endgroup\$ – user21820 Jul 16 '20 at 17:08
  • 2
    \$\begingroup\$ I looked at the thread. The poster making the claim about polyfuses in regards to that laptop clearly doesn't know what they're talking about. For example, their claim "Normally a polyfuse takes 10-15 minutes to reset" is absolutely, utterly wrong. The fuse just needs to cool down, then it can reset - a process takes only a few seconds, not 15 minutes. You base your thinking on a flawed premise in some random forum posting, why would you (let alone me) waste time pursuing it further? Use a load switch and get on with life. \$\endgroup\$ – hacktastical Jul 16 '20 at 18:46
  • \$\begingroup\$ I did not base my thinking on that forum post. Your assumption shows that it is your thinking that is flawed, not mine. I explicitly stated that that thread has "claims", not evidence, not facts. I merely asked you for evidence for your claim. And your last sentence is rude. Don't be rude just because you don't have an answer. \$\endgroup\$ – user21820 Jul 17 '20 at 5:06
  • 1
    \$\begingroup\$ "polyfuse takes 10-15 minutes to reset" is absolutely, utterly wrong." Sorry, you are misinformed. Cooling down is not sufficient, the material must restore its original crystalline structure, which takes a lot of time. Check Wikipedia, "This cooling usually takes a few seconds, .... The device may not return to its original resistance value; it will most likely stabilize at a significantly higher resistance (up to 4 times initial value). It could take hours, days, weeks or even years" \$\endgroup\$ – Ale..chenski Jul 17 '20 at 8:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.