USB key wearing out?

[ahengst]

This is a General "electronics" question.
A vendor gave us a USB stick a few years ago and I've been using it daily.
Lately it is not connecting to my Mac, but it still connects to the PC at work. If I hold it in my hand for a minute to warm up it does work in the mac, and once it's on it stays on. This trick has worked for months but I probably better retire this one soon.

I'm not sure if this is some component just aging or if the circuits are deteriorating, and I don't quite understand how warmth would help (some resistance value might change or.... ?? )

Since this behaviour *could* appear in other circuits, or in colder applications, I thought it might be interesting to ask an Engineer !

Thanks
Andy (happy adafruit customer, so I meet the criteria!)

[adafruit_support_bill]

Contacts wear and springs deteriorate a bit with each plug-in. Tolerances on plugs & sockets vary a bit, so one combination may fail before another. Warming it up could change things enough to make a difference. But if the connection is getting flaky, it is probably time to look for a new one.

[cstratton]

Could be a cracked solder joint - especially between the USB connector and the PCB where there is a lot of stress.

Could also be an out-of-spec crystal. That could bring the temperature difference in, and hint at why it works on some machines but not others (depending on if the host's clock is on the same or opposite side of the allowed variation).

You could probably have an interesting time devising experiments to figure out exactly what is wrong, but unless your job is in the manufacturer's QA department or designing similar products, there may not be any point to this beyond curiosity.

While it doesn't seem an aspect of your problem, I'd expect many who saw the thread title, and read that it was a promotional giveaway probably initially suspected a bargain basement (or even gray market) flash IC with a lot of bad blocks from the factory accumulating even more in usage.

[ahengst]

Update: this USB key still works, but with the newly replaced PC at work I now need to warm the key up there as well before it will activate. I use a very short cable that I don't disconnect from the USB key, to extend its life with daily plugging. So the warming up only should affect the circuitry of the key (and since it was working on the previous PC neither of the connectors were out of tolerance). Thanks cstratton for the tip about the crystal.

This key has blinky lights (7-12 very tiny SMD LEDs inside translucent body) that clearly show whether the key is "on" and/or "active".
It's true I don't really need to know what causes this. But this might be of interest if people find their USB sticks suddenly stop responding.

p.s. critical data is safe!

[thefatmoop]

http://en.wikipedia.org/wiki/Flash_memory#Memory_wear

My brother's friend did a lot of flash work in the computer industry back when flash drives were just hitting the market. All flash drives are expected to have distributed wear, and all should also have bad block management (BBM) - although he called it something different.

you'll see BBM in action if the flash drive slowly decreases in size over time.

[adafruit_support_mike]

Flash memory does wear out eventually.

Flash memory is an adventure in packing as much functionality into as small a space as possible. One bit of storage in static RAM is made of six transistors. One bit of storage in dynamic RAM is made of a capacitor with a mosfet controlling whether charge enters or leaves. One bit of flash memory is made of a single mosfet with two gates, one sitting above the other:



The blue material is P-type silicon, the red material is N-type silicon, the grey is silicon dioxide (glass, an insulator), and the yellow parts are the metal contacts. The S contacts are the sources, the D contacts are the drains, the G contacts are the gates, and the W contact on the flash transistor is the read/write input. That blob of N-type material completely surrounded by glass is the flash mosfet's second gate.

In both cases, putting a positive charge in the gate pulls electrons toward the surface of the P-type material, producing a conductive channel between the N-type blobs at the source and drain.

If you apply a relatively high voltage to the read/write contact of a flash mosfet, electrons will enter or leave the encapsulated chunk of N-type material by means of quantum tunneling. Once you remove the high voltage, that charge remains trapped, but it continues to have an influence on the P-type material. If you trap a positive charge, it pulls electrons toward the surface of the silicon, making the transistor easier to turn on. If you trap a negative charge, it pushes electrons away from the surface of the silicon, making the transistor harder to turn on.

You read the stored value by applying a weak voltage to the gate and seeing whether it's enough to turn the transistor on or not.

Every time you change the trapped charge, you have to fight the trapped charge that's already there. Quantum tunneling isn't as reliable as regular through-a-conductor connections, so eventually the encapsulated material can reach a midpoint where it doesn't read or write very well. For today's flash memories, that usually happens after about a hundred thousand read/write cycles.

If you have an older device and have been using it frequently, it's possible that you're starting to rub up against the limits of what the transistors can do.