Once again, Google proves it's bought out.

On Sun, 08 Oct 2017 13:26:19 -0500, Char Jackson
wrote:

On Sat, 07 Oct 2017 13:11:18 -0400, wrote:

Years ago, I installed a USB2 card in this computer. It lasted several
years, then it just quit working. Two years ago, I bought another one.
That one lasted about one year, then that one quit working. I have just
given up on them since. The USB2 is a lot faster than the 1.1. In fact I
dont see any reason to go to USB3. I think USB2 is fine.

I agree that USB2 is fine...for things like a keyboard or a mouse. I
certainly wouldn't want to use it to connect an external drive,
including a thumb drive.



I use USB2 to connect external drives all the time, since I don't have
any USB3 ports. But since I use external drives only for backup, and I
do other things (sometimes going to sleep) while the backup is
running, I really don't care how slow it is.

wrote:
On Sun, 08 Oct 2017 12:25:37 -0700, Ken Blake
wrote:

On Sun, 08 Oct 2017 13:26:19 -0500, Char Jackson
wrote:

On Sat, 07 Oct 2017 13:11:18 -0400, wrote:

Years ago, I installed a USB2 card in this computer. It lasted several
years, then it just quit working. Two years ago, I bought another one.
That one lasted about one year, then that one quit working. I have just
given up on them since. The USB2 is a lot faster than the 1.1. In fact I
dont see any reason to go to USB3. I think USB2 is fine.
I agree that USB2 is fine...for things like a keyboard or a mouse. I
certainly wouldn't want to use it to connect an external drive,
including a thumb drive.


I use USB2 to connect external drives all the time, since I don't have
any USB3 ports. But since I use external drives only for backup, and I
do other things (sometimes going to sleep) while the backup is
running, I really don't care how slow it is.

USB 2 can copy a rather large partition of mine in about a half hour.
Thats acceptable to me. USB 1.1 takes 16 or more hours for the same
drive. That's not acceptable, but when it's all I have, I use it to
backup and try to do it when I am sleeping and since that extends into
the next day, I try to do it when I know I wont need the computer that
next day.

I have never had any USB 3, so I dont know how long that would take, but
I do question if copying stuff as fast as they claim USB 3 is, I have to
question if the copy is safe and reliable.

It's just like driving. Driving 200 miles at 10mph is going to take very
many hours. It can be done, but it sure is slow. Driving at 60mph is
pretty safe and gets the trip done. Although driving at 140mph will get
you to your destination fast, it's dangerous and risky.

Given the choice, I'd choose the 60mph.


Believe it or not, people worry about this stuff.

When they design USB chips, they do this kind of test.

(See figure 9 here...)

https://www.mouser.com/api/CrossDoma...07.bncrdwbpXHs

http://www.mouser.com/images/microsi...use2-fig09.png

(In this example, the red areas are gray instead, and this one
has more room between the top of the waveform, and the gray
rectangles at top and bottom.)

https://blog.nxp.com/wp-content/uplo.../04/Fig-5a.jpg

The eye diagram and mask, the measured waveform isn't
allowed to touch the "thing" in the center, or the
fat red rectangles at the top and bottom of the diagram.
That's how you know the design has margin. The cables can still
ruin it of course, by being poorly made.

Some people who make USB3 goods, they'll send their product
to an independent testing lab, to have the verification done.
That's because the set of instruments used, can cost a fortune.

The software that makes that measurement, costs $1500 and comes
on a floppy :-) I got to make a measurement similar to that one
once, and that's what the eye diagram software costs for the
instrument. It's an optional extra on the storage scope. So I got
to verify my design was working too. Without sending it out
to a third party, as we had the gear for it in the lab. (The
fastest scope we had, was a 40GHz sampling scope with
rigid plumbing electrical connections. Our bench scopes
didn't go quite as fast, but the bench ones cost $35K
a piece.) But that's the kind of dosh you spend, when you
want to see whether electronics are working or not.

So while you may spend $10 for something, and wonder
"did the guy making this know what he was doing", well,
an independent lab may have verified that he got it right,
using a million dollars worth of test gear.

Many hardware interfaces now, use packets, and the packets
have checksums. And that's how the receiving end knows the
data is good. SATA uses packets. USB3 uses packets.
PCI Express uses packets. And so on. And the protocols
can request retransmission. That kind of treatment can
improve the undetected error rate by three orders of magnitude.

Paul