Safety of USB "Y" connectors for power?

[Mike S.]

I've always been leery of storage devices that claim to be 100% USB
powered. Most of them having moving parts usually come with a "Y" cable
that draws power from USB ports in the expectation of getting 1 amp total,
rather than the expense (and possible inconvenience) of providing a
separate plug-in power supply.

Sometimes I'm in a situation where there is only one free USB port. Other
times, I'm still distrustful - like the portable DVD burner whose
manufacturer's drive specs tell me that it may very well draw more than 1
ampere at a time.

For that reason I've taken to connecting a wall-wart that provides 5V USB
power to the dummy plug on the other end of that "Y" connector. But then,
I'm worried again. Does the back current from an external PS pose any danger
to the PC's USB ports? Then again, what about the back current from one
USB port to the next, when connected to both limbs of that "Y"? Anyone
with knowledge of motherboard design care to comment (besides saying that
I worry too much...)

 

[Rod Speed]

I've always been leery of storage devices that claim to be 100% USB powered.

It works fine for the lower powered devices like mice etc.

Most of them having moving parts usually come with a "Y" cable
that draws power from USB ports in the expectation of getting
1 amp total, rather than the expense (and possible inconvenience)
of providing a separate plug-in power supply.

Only the high power devices. You dont get that with mice etc.

Sometimes I'm in a situation where there is only one free USB port.

You can always add a powered hub.

Other times, I'm still distrustful - like the portable DVD
burner whose manufacturer's drive specs tell me that
it may very well draw more than 1 ampere at a time.

Plenty of powered hubs handle that fine.

For that reason I've taken to connecting a wall-wart that provides 5V
USB power to the dummy plug on the other end of that "Y" connector.

Makes more sense to get one with a USB socket.

But then, I'm worried again. Does the back current from
an external PS pose any danger to the PC's USB ports?

Not if its properly designed.

Then again, what about the back current from one USB port
to the next, when connected to both limbs of that "Y"?

Not if the device being powered is properly designed.

Anyone with knowledge of motherboard design care
to comment (besides saying that I worry too much...)

They're just in parallel at the motherboard end for the power
lines, usually with some form of over current protection.

 

[Arno]

I've always been leery of storage devices that claim to be 100% USB
powered. Most of them having moving parts usually come with a "Y" cable
that draws power from USB ports in the expectation of getting 1 amp total,
rather than the expense (and possible inconvenience) of providing a
separate plug-in power supply.

Sometimes I'm in a situation where there is only one free USB port. Other
times, I'm still distrustful - like the portable DVD burner whose
manufacturer's drive specs tell me that it may very well draw more than 1
ampere at a time.

For that reason I've taken to connecting a wall-wart that provides 5V USB
power to the dummy plug on the other end of that "Y" connector. But then,
I'm worried again. Does the back current from an external PS pose any danger
to the PC's USB ports? Then again, what about the back current from one
USB port to the next, when connected to both limbs of that "Y"? Anyone
with knowledge of motherboard design care to comment (besides saying that
I worry too much...)

If it is a current controlles port, I think there may
be some protection circuitry. However, many mainboards only
have 2A fuses in the USB power lines per port pair. A
"wall-wart" 5V PSU could blow these (very hard to replace)
if it had a bit higher voltage that the computers's PSU.

Incidentially, the external PSU will only supply power in the
first place, if it has higher volatge. This means that for
fused USB ports it is entriely useless. These would also be
able to supply something like 1.5A per port-pair anyways.
For current controlled ports, the situation is a bit better,
but still problematic. You may well damage hardware there as
well.

Bottom line: These Y-Cables are only ever safe to use if
the used USB ports are on the same 5V supply. I would strongly
advise you toi stop using the external PSU in addition.

Arno

 

[John Weks]

If it is a current controlles port, I think there may be some protection circuitry.

There normally is.

However, many mainboards only have 2A
fuses in the USB power lines per port pair.

It isnt that many now that have nothing but that.

A "wall-wart" 5V PSU could blow these (very hard to replace)
if it had a bit higher voltage that the computers's PSU.

Nope, there should be a diode at least in the USB powered device that stops that from happening.

The USB ports arent electrically paralleled.

Incidentially, the external PSU will only supply
power in the first place, if it has higher volatge.

Thats wrong too if the USB powered device is designed properly.

The multiple USB power sources arent just electrically in parallel.

This means that for fused USB ports it is entriely useless.
Nope.

These would also be able to supply something like 1.5A per port-pair anyways.

Plenty of laptops cant do that.

For current controlled ports, the situation is a bit better,

Its completely safe in fact.

but still problematic. You may well damage hardware there as well.

Nope, thats the whole point of the current control.

Bottom line: These Y-Cables are only ever safe to use if the
used USB ports are on the same 5V supply. I would strongly
advise you toi stop using the external PSU in addition.

More fool you.

 

[Mike Tomlinson]

For that reason I've taken to connecting a wall-wart that provides 5V USB
power to the dummy plug on the other end of that "Y" connector.

It's not something I would advise doing.

How about buying a powered USB hub, plugging both connectors of your Y
cable into that, then plugging the hub into the PC? That way, the wall
wart for the hub is powering the external drive, not the PC's PSU.

 

[Arno]

There normally is.

I am talking about one for external overvoltage. That is advanced
functionality and not necessarily there.

It isnt that many now that have nothing but that.

So? I have serveral ASUS boards and one Gigabyte board with that
and nothing else.

Nope, there should be a diode at least in the USB powered device that stops that from happening.

A diode in the USB powered device does not help, it would need to be
in the Y-cable. And it would need to be a high-current low loss
Shottky type to limit voltage loss to a level that the attached
HDD still spins up.

The USB ports arent electrically paralleled.

Are you sure? This whole construction is unspecified (and disallowed)
by the USB standard. Hence every vendor can do what they see fit.
And if both ports are driven from the same 5V line, there actually
is no need for any circuitry, you can just put the two power lines
together.

Thats wrong too if the USB powered device is designed properly.

Oh? There is just ONE power input in an USB powered device, not two.
How would that work?

The multiple USB power sources arent just electrically in parallel.

Again, how do you know that? I think you are deep in the real
of wishful thinking here...

Yes.

Plenty of laptops cant do that.

Laptops typically do not have fused USB ports. They have proper
current control circuitry.

Its completely safe in fact.

No. External overvoltage circuit need to be present for it to
be save.

Nope, thats the whole point of the current control.

More fool you.

I think you just failed EE 101. Your undertsnading on how
USB power circuitry works is clouded by limited knowledge,
wishful thinking and plain misunderstanding basic DC
circuitry design.

Arno

 

[Arno]

It's not something I would advise doing.

How about buying a powered USB hub, plugging both connectors of your Y
cable into that, then plugging the hub into the PC? That way, the wall
wart for the hub is powering the external drive, not the PC's PSU.

That would be pretty safe and in the worst case destroy only
the hub, not the USB port on the PC.

Arno

 

[Mike S.]

It's not something I would advise doing.

How about buying a powered USB hub, plugging both connectors of your Y
cable into that, then plugging the hub into the PC? That way, the wall
wart for the hub is powering the external drive, not the PC's PSU.

Yep. Sounds like a logical compromise.

 

[Mike Tomlinson]

Again, how do you know that? I think you are deep in the real
of wishful thinking here...

You're talking to another Woddles morph. Never engage in a battle of
wits with an unarmed man.

 

[John Weks]

I am talking about one for external overvoltage.

You dont need that with basic diode protection in the USB powered device.

That is advanced functionality and not necessarily there.

It is not the only way to protect against that particular situation.

So?

So you are talking about something that is not seen much at all.

I have serveral ASUS boards and one Gigabyte board with that and nothing else.

Thats fine for the situation you are talking about. You wont get anything
like that sort of current even with a wall wart USB power device and
nothing but diode protection in the USB powered device.

A diode in the USB powered device does not help,

Corse it does.

it would need to be in the Y-cable.

Wrong. It can be electically just where the Y cable connects to the USB powered device.

And it would need to be a high-current low loss Shottky type to
limit voltage loss to a level that the attached HDD still spins up.

And its completely trivial to use one there.

Are you sure?
Yep.

This whole construction is unspecified (and disallowed) by the USB standard.

Sure, but it wont work with just electrically paralleled.

Hence every vendor can do what they see fit.
And if both ports are driven from the same 5V
line, there actually is no need for any circuitry,
you can just put the two power lines together.

Nope, because the wire guage in the Y adapater is normally
different in each leg and at those currents, the voltage at the
USB powered device end will be significantly different.

Oh? There is just ONE power input in an USB powered device, not two.

Thats not correct. Most will work fine with a normal USB
cable when the single port can supply enough current.

How would that work?

Basically they arent just electrically paralleled.

Again, how do you know that?

It cant work that way, because the voltage at the USB
powered device end will be significantly different in each leg.

I think you are deep in the real of wishful thinking here...

Nope, you are.

Nope.

Laptops typically do not have fused USB ports.

Quite a few of the cheap and nasty ones do. Or nothing at all.

They have proper current control circuitry.

Thats over stating it with the crap.

Yes.

External overvoltage circuit need to be present for it to be save.

Nope. Basically because the USB wall wart voltage sags significantly as the current increases.

I think you just failed EE 101.

Nope, you did.

Your undertsnading on how USB power circuitry works is clouded by limited
knowledge, wishful thinking and plain misunderstanding basic DC circuitry design.

Thats actually your problem. The two legs arent just electrically parallelled at the USB powered device end.

 

[John Weks]

That would be pretty safe and in the worst case destroy only
the hub, not the USB port on the PC.

No reason why the hub would be destroyed.

 

[Arno]

You're talking to another Woddles morph. Never engage in a battle of
wits with an unarmed man.

Agreed. I had some suspicion, but only later and did not want to
delete the already written answers for possible value to others
here.

Arno

 

[Rod Speed]

William R. Walsh wrote

That is anyone's guess. Have you seen some of the USB cards that are
on the market today? Seriously, you can buy some of them for $1 *new*
and a little shipping from someone in Hong Kong, China or similar locations.

Yeah, tho you can often inspect those visually for that particular question.

Many have VIA chipsets, and the later ones (VT6212, 6214) seem to do
what they should when put on a reasonably well built card. Still, I doubt
there's much change from that $1 left to put much quality into the card...

They dont need to, they just steal the design.

probably just enough to throw some solder at the components
and hope they stick in approximately the right locations.

Some chipsets (in particular the ones from Acer Labs)
don't really seem to work in USB 2.0 mode, especially
on a non-Windows OS. (BTDTGTTS*)

Never seen that myself.

I've seen some of these cards that have excellent USB chipsets
(NEC) and yet the card maker still does things they should not--like
splitting off one USB port into two that may not be used at the same
time. (It works fine until you connect a USB 2.0 device, and then you
get random failures. The other, normal ports work fine.)

Yeah, the cheap chinese stuff can be a hell of a problem in that regard.

But its so cheap you can just chuck it in the bin when you find one of those.

I'd hope motherboards were better designed than these disasters.
I wouldn't bet on it, especially with cheaper or no-name boards.

I normally stick with Asus or Gigabyte etc.

The same can be said for a lot of storage enclosures as the USB cards above.
They're only just good enough a lot of the time.

I use HD docking stations myself. Mainly because my main need
is for overflow from the PVRs and they work very well for that.

 

[Rod Speed]

There's usually a reference design in the datasheets for the IC they use.

That isnt what the best of the designs use tho.

It's the other parts that I'm talking about--they probably leave off
everything that isn't strictly necessary, like 'suggested' capacitors

They're mostly designed not to need those now.

or circuit protection.

Thats mostly built in too now.

I can't say for sure about Linux, as I only have it on a few
machines. But FreeBSD and the MacOS really don't like
the Acer chipsets for USB or USB/Firewire. I've had a
few different boards that used some Acer chip and none
of them worked for USB 2.0 on those operating systems.

Presumably someone ****ed up with the driver given they work fine with other OSs.

Most frequently data transfers would just stall.
It wouldn't usually take the OS down, but the failures
would produce tons of complaining on the system console.

Like I said, presumably someone ****ed up with the driver.

Oddly enough, the Firewire part of those chipsets
that did both USB and Firewire worked perfectly.

More evidence that someone ****ed up with the driver.

And they worked on Windows (probably
the only OS tested by the chipset maker?).

Very likely.

Not everyone will. Some folks only care about the price, and therefore
get what they deserve. I hate it when machines with no-name parts of
dubious quality come in for repair, because it just can't always be done.

Sure, but the replacements are so cheap it doesnt really matter much.