USB 2.0 and below have a 2.5w (500mA) current limit

giantclam wrote on 2023-10-14, 02:04:

USB 2.0 and below have a 2.5w (500mA) current limit

VIA's design has one fuse for both ports. I'm assuming they did the combined 1A and then 2.5x to account for inrush. I'm thinking about breaking out the fuses and doing a dedicated fuse for each port.

Been a long time since I mucked with Eden/EPIA so I don't rightfully recall what they did ~ newer designs can certainly control power on a per port basis ...but overall USB 1.x ( OK for HID devices) is just too slow for me to worry about (ie; I have mainboards with onboard USB 1.x ...never use it, because a USB 2.0 PCI card shames it =) ...ofc, on a EPIA one can't entertain things like this.

I think the idea to re-create a specific rare board adapter is a great idea ! Hope you can come up with a working solution for those that may want it.
TBH I have never liked micro boards because of the need for such "optional" adapters to fully use what is on the board... just my opinion 😀

Horun wrote on 2023-10-14, 03:23:

I think the idea to re-create a specific rare board adapter is a great idea ! Hope you can come up with a working solution for those that may want it.
TBH I have never liked micro boards because of the need for such "optional" adapters to fully use what is on the board... just my opinion 😀

Probably nobody wants these but me 🤣. But I'm having fun, and I've been wanting to learn KiCad 😀

The overcurrent detection pins are most likely meant for an e-fuse IC (not to be confused with configuration e-fuses inside programmable chips). There are multiple terms for that type of chip ("load switch" seems to be another common one), so I don't know if there's one that I can recommend. It should be in a relatively hobbyist-friendly package for hand soldering (nothing with an exposed bottom pad - those will need hot air or a big hole under the chip for access with a regular iron). VIA took the simple method of using a resettable fuse (polyfuse), but the OC pin on the chipset usually allows the OS to pop up a warning message to tell you why your stuff stopped working.

For protection, the official spec is 500mA, but a little extra for non-compliant devices is nice. A 1.1A resettable fuse per port is the simple method. I don't like having multiple ports share one fuse (as VIA did) because one USB device failing makes multiple devices disconnect at once.

lti wrote on 2023-10-15, 01:51:

The overcurrent detection pins are most likely meant for an e-fuse IC (not to be confused with configuration e-fuses inside programmable chips). There are multiple terms for that type of chip ("load switch" seems to be another common one), so I don't know if there's one that I can recommend. It should be in a relatively hobbyist-friendly package for hand soldering (nothing with an exposed bottom pad - those will need hot air or a big hole under the chip for access with a regular iron). VIA took the simple method of using a resettable fuse (polyfuse), but the OC pin on the chipset usually allows the OS to pop up a warning message to tell you why your stuff stopped working.

For protection, the official spec is 500mA, but a little extra for non-compliant devices is nice. A 1.1A resettable fuse per port is the simple method. I don't like having multiple ports share one fuse (as VIA did) because one USB device failing makes multiple devices disconnect at once.

Oh nice I didn't know about the e-fuse ICs. I will do research on that!

I found a whitepaper from Littelfuse about USB protection. They claim hot plugging a USB device can result in 3x rated max current draw briefly. So they recommend 1.5A PTC for a 500mA USB 1.1 port. Yes, I agree about a fuse for each port. But I'm running into space constraints trying to do that. Probably why VIA took the easy way out.

In the same whitepaper they also recommend a varistor that clamps at 5.5V across VBUS and Ground. They claim, in testing, that they saw spikes up to 24V while hot plugging devices. So I'm going to incorporate that, as well. Conveniently, they happen to make one for this purpose 🤣.

Ordered some prototype boards!

This is the initial proof-of-concept board I will make up

I know the layout isn't so hot. But this is just a test board.

I added a 5.5V varistor across VBUS and ground. Several USB design docs I read suggested this as voltages can spike to 24V when hotplugging devices. I kept the 2.5A fuse. Again, design docs recommend this value for two USB 1.1 ports due to inrush currents when hotplugging.

If this concept seems to work, the final design will have dedicated protection for each port independently.

I've already ordered the parts and solder paste. The PCBs will come with a stencil. I plan to use my hot air station and microscope to make these. Those 0603 parts are tiny!

Found this... https://www.mini-box.com/s.nl/it.A/id.260/.f

giantclam wrote on 2023-10-20, 00:00:

Found this... https://www.mini-box.com/s.nl/it.A/id.260/.f

That's pretty cool. Unfortunately that is not the correct board. That's for EPIA boards that use a standard 2.54 header, already have the bus termination, transient voltage suppression, and EMI filtering, on the motherboard.

As you can see from my pics in the first post, the USB-CV1 is way more complicated.

I thought the problem with voltage spikes would be on the device end, not the motherboard end.

I wonder why VIA had a 1000uF cap on the 5V supply. I thought they were only supposed to have something like 22uF per port. They might be trying to have the cap supply the initial current when hot-plugging.

Now you're making me reconsider using 750mA fuses per port on a USB card. I never even tested it because I just have socket 7 motherboards and a single PCI slot running through a bridge on my modern computer.

lti wrote on 2023-10-21, 23:49:

I thought the problem with voltage spikes would be on the device end, not the motherboard end.

I wonder why VIA had a 1000uF cap on the 5V supply. I thought they were only supposed to have something like 22uF per port. They might be trying to have the cap supply the initial current when hot-plugging.

Now you're making me reconsider using 750mA fuses per port on a USB card. I never even tested it because I just have socket 7 motherboards and a single PCI slot running through a bridge on my modern computer.

My guess is they want a lot of local capacitance because it probably doesn't have a very low impedance path through the daughterboard, down the small pins, and then through the motherboard back to +5V.

The "resettable fuse" idea is good. Make sure you get one that is no more than about 6 volts because even though they are "amp" rated that is at stated voltage. In other words they blow at a certain "wattage",
not really exact but a 250v 1.5 amp will not protect 5v equipment at 1.5amps very well... just a thought

Horun wrote on 2023-10-22, 03:31:

The "resettable fuse" idea is good. Make sure you get one that is no more than about 6 volts because even though they are "amp" rated that is at stated voltage. In other words they blow at a certain "wattage",
not really exact but a 250v 1.5 amp will not protect 5v equipment at 1.5amps very well... just a thought

Good to know! I just checked and the ones I got from Mouser are 6V.

kingcake wrote on 2023-10-14, 03:31:

Horun wrote on 2023-10-14, 03:23:

I think the idea to re-create a specific rare board adapter is a great idea ! Hope you can come up with a working solution for those that may want it.
TBH I have never liked micro boards because of the need for such "optional" adapters to fully use what is on the board... just my opinion 😀

Probably nobody wants these but me 🤣. But I'm having fun, and I've been wanting to learn KiCad 😀

I'd be interested for when I get round to testing my EPIA boards. Didn't know a daughter board was needed to enable USB, and I'd like to be able to use USB keyboard and mouse. My biggest problem with these sorts of thread is that I no longer have the eyesight or steady hands to build them, and it's rare that somebody would want to build them to sell.

debs3759 wrote on 2023-10-22, 19:48:

kingcake wrote on 2023-10-14, 03:31:

Horun wrote on 2023-10-14, 03:23:

I think the idea to re-create a specific rare board adapter is a great idea ! Hope you can come up with a working solution for those that may want it.
TBH I have never liked micro boards because of the need for such "optional" adapters to fully use what is on the board... just my opinion 😀

Probably nobody wants these but me 🤣. But I'm having fun, and I've been wanting to learn KiCad 😀

I'd be interested for when I get round to testing my EPIA boards. Didn't know a daughter board was needed to enable USB, and I'd like to be able to use USB keyboard and mouse. My biggest problem with these sorts of thread is that I no longer have the eyesight or steady hands to build them, and it's rare that somebody would want to build them to sell.

Your EPIA boards might have the standard USB header. Some do, some don't.

I'll be building these up in the next week or two. If they seem to work ok, I'll gladly send you a prototype to test with. The broader the testing the better.

kingcake wrote on 2023-10-22, 20:22:

debs3759 wrote on 2023-10-22, 19:48:

kingcake wrote on 2023-10-14, 03:31:

Probably nobody wants these but me 🤣. But I'm having fun, and I've been wanting to learn KiCad 😀

I'd be interested for when I get round to testing my EPIA boards. Didn't know a daughter board was needed to enable USB, and I'd like to be able to use USB keyboard and mouse. My biggest problem with these sorts of thread is that I no longer have the eyesight or steady hands to build them, and it's rare that somebody would want to build them to sell.

Your EPIA boards might have the standard USB header. Some do, some don't.

I'll be building these up in the next week or two. If they seem to work ok, I'll gladly send you a prototype to test with. The broader the testing the better.

I have at least 7 (that's how many I've boxed up and labelled as Epia boards) different boards, so good chance at least one will need it. I'll happily help with testing, thanks.

debs3759 wrote on 2023-10-22, 20:37:

kingcake wrote on 2023-10-22, 20:22:

debs3759 wrote on 2023-10-22, 19:48:

I'd be interested for when I get round to testing my EPIA boards. Didn't know a daughter board was needed to enable USB, and I'd like to be able to use USB keyboard and mouse. My biggest problem with these sorts of thread is that I no longer have the eyesight or steady hands to build them, and it's rare that somebody would want to build them to sell.

Your EPIA boards might have the standard USB header. Some do, some don't.

I'll be building these up in the next week or two. If they seem to work ok, I'll gladly send you a prototype to test with. The broader the testing the better.

I have at least 7 (that's how many I've boxed up and labelled as Epia boards) different boards, so good chance at least one will need it. I'll happily help with testing, thanks.

A fellow EPIA hoarder! 🤣

Got the boards today! Holy crap, look how tiny the pads are for U1 and U2! I will definitely have to make these under a microscope...

Don't know if I mentioned it, but these tiny boards are actually 4-layer. Middle layers are VBUS and GND to reduce EMI.