Devo me preocupar com descargas eletrostáticas?


33

Fazendo principalmente um trabalho de microcontrolador, nunca me preocupei em me aterrar com uma pulseira ou qualquer outra implementação.
Eu deveria me preocupar com isso? Qual é a gravidade da ESD ao trabalhar na minha bancada de trabalho em projetos?


3
Uma vez vi alguém tocar no radiador (doméstico) para descarregar antes de trabalhar dentro de um PC. Isso realmente ajuda?
Lars

Ajuda se o radiador estiver adequadamente aterrado. Caso contrário, pode funcionar como algo para ajudá-lo a cobrar.
Kortuk

Amarre-o ao terra com um fio, através de um resistor de 100k ou mais e você ficará bem.
Kortuk

Se você atravessar o tapete, do radiador ao PC, isso não ajuda em nada. Se você andar com roupas que geram estática, isso não ajuda.
Kortuk

Respostas:


24

Eu sei que a maioria das pessoas vai gritar comigo por isso. E honestamente tomo algumas precauções contra ESD dia a dia, mas a ESD é um modo de falha rara variado na eletrônica moderna.

Pela minha formação, participo da aula de design da capstone na minha universidade e eles precisam criar muitas coisas em quatro fases. Muitos estudantes (quase todos) não nos ouvem quando alertamos sobre as precauções contra ESD.

Eu já vi pelo menos 10 projetos em que eles culparam a ESD, apenas um que provavelmente era ESD. Nesse projeto, um quadro estava funcionando e um aluno se aproximou e tocou nele, e você podia ouvir o choque audivelmente enquanto ele descarregava a eletricidade estática. Isso falhou.

Para os outros 9, eles culparam a ESD apenas por descobrirem mais tarde que tinham uma falha de design que não era fácil de rastrear e apenas criavam sua cabeça feia em condições muito específicas.

Ouvi muitos mais de 10 projetos culparem o esd e encontrarem um problema em poucas horas, mas apenas 10 juraram que deveriam poder continuar para a próxima fase porque funcionou, mas a ESD o fritou.

Nessa nota, quando você estiver trabalhando, verifique as folhas de dados. A maioria dos dispositivos modernos possui proteção ESD, classificada com alguma voltagem usando HBM (modelo do corpo humano). Se estiver abaixo de 1kV, é bastante suscetível. se for avaliado em 2k, é semi-resiliente; na maioria das vezes você teria que ser burro para danificá-lo. se for 4k, é bastante resistente. se for 8k, você ouvirá uma faísca antes de machucá-la.

Desconfie daqueles que culpam a ESD, mas use uma pulseira para o caso. você pode obter um tapete e uma cinta por quase nada.


Nota lateral: se você tem piso de ladrilhos e seu prédio é muito úmido, você corre menos riscos ao descarregar o ar a uma taxa maior.
Kortuk

Boa dica sobre as classificações de ESD. Eles devem orientar a decisão sobre a quantidade de proteção necessária caso a caso.
JPC

Sim, temos um dispositivo com classificação <500V, o que significa que quase nada pode destruí-lo; felizmente, ele é soldado em instalações profissionais e outros dispositivos oferecem proteção.
precisa saber é o seguinte

1
+1 for modern ICs, but you're leaving out things like bare FETs and old MOS/CMOS IC designs, which are still quite susceptible to ESD damage.
Warren Young

11

Yes you should/must be worried about this. The simple case is where you blow the device completely. This is easy, it just does not work. The problem is where you damage the IO driver circuit - quite common as this is the bit that is exposed by the pin. In this case the driver may appear to work for some of the time. If it has to operate at high speed for part of the time then it may well fail giving intermittent and hard to diagnose faults. (Been there and got several t-shirts!!)

This is becoming more and more necessary as device geometries are shrinking and core voltages are dropping. Microcontroller manufacturing processes have often been slightly behind the bleeding edge but as oxide layer thicknesses are reducing with the device geometry size the required ESD voltage required to damage the device is reduced.

The bottom line is that you can get away with ignoring ESD precautions for a while but for how long? As Clint Eastwood said “You've got to ask yourself one question: 'Do I feel lucky?' Well, do ya punk?”


1
most things like microcontrollers sport diodes in the pins that allow ESD to discharge harmlessly. They are called ESD protection diodes normally.
Kortuk

2
They are a palliative that provide some protection but are often themselves damaged as they protect the main device. On the second ESD event there is no protection left on the pin.
uɐɪ

I had never been told that, nor have i seen common failures were people do not use protection. I would like to know more about it. On that note Ian, do not take me wrong, I am a strong supporter of ESD protection, I just feel being honest with people about what protections already exist is worthwhile. You have said the same thing in your post, so no worries.
Kortuk

1
@Kortuk: As I understand it, the energy of an ESD event doesn't necessarily blow through the oxide layer, or not, full stop. It may push that layer around inside the device, like a bullet hitting an armor plate. Hit it enough times, hard enough, and you will break through.
Warren Young

10

The tricky thing with ESD is that even when it doesn't seem to affect anything, every small discharge does cause some damage, whether it's pitting of the silicon, vaporizing a little bit of the bonding wire, removing some of the metalization on one of the semiconductor layers, or some other effect, it will physically alter the product.

In most cases, most uses of a given part won't be measurably altered by the change. Most people overengineer their designs so they aren't using them anywhere near their limit, so there's a huge margin where a pitted part will work just fine for more uses.

Over time repeated discharges will change the part enough that it will behave slightly out of spec - but perhaps not enough to notice without thorough testing. This is really the worst case scenario - since it works most of the time, then it's really difficult to find out why it's not working when it fails. It's far better for a part to fail completely so it's easy to track down.

Alternately, the one small discharge that happened once damages it enough that when you do use it to full capacity, it then fails. The happens, of course, once you've deployed it in the field and it's a pain to get it back and troubleshoot it.

This kind of damage cannot be detected without removing the silicon from the package and inspecting it, or, in cases where the damage is hidden inbetween layers, by slicing the silicon into very thin sheets where the damage is likely to occur.

So yes, everyone here is right when they say, "I rarely follow ESD precautions and I have never seen an issue." They will never see it, and when it does fail there will be no reason to believe it was ESD related because there was no recent ESD event.

But that doesn't mean I'm going to trust the products that come from their bench, either.


1
If you go for the intractable then beware that silicon vendor ratings are most commonly statistical so you still get 1 part out of 1000 that is more than 3 sigma from the mean. It may ruin your design or it may not. :)
jpc

7

You should only skip ESD measures if you can bear the pain ($$) of replacing whatever it is. If you do go un-grounded, though, be aware that device failure is not all-or-nothing. Complete catastrophic failure is probably the best failure mode, because you won't waste as much time trying to figure out what went wrong. Personally, I won't handle anything bigger than 7400 series stuff without a grounding wrist strap.


7

If you don't take ESD precautions, you risk damaging things. The probability of damaging something varies all over the place (temperature, humidity and floor surface being important factors), and it's entirely possible that you'll never damage anything (or at least not such that you'd notice).

I HAVE damaged devices before with ESD. Heck, I once killed a serial port (something that's PROTECTED against ESD because the pins are hanging out there) by shocking to the pins.

Now, that's not to say that you have to have the be-all and end-all of ESD protection. Even if you don't go for the full array with ionizers and wrist straps and conductive tile flooring, try to get an grounded mat for your bench, and try to touch them mat when you sit down. You'll tend to brush the mat with your hands as you are working, and this will keep you mostly at the same potential as the parts on the mat and keep the possibility of problems to a minimum.


1
Seconded. The mat on the bench will do wonders.
Kortuk

1
Sorry, i meant to ask earlier, how the hell did you total an RS232 port?
Kortuk

1
Grabbed a cable connected to the port and somehow managed to shock to the rcv line instead of the shell. Shouldn't have killed it even so, but apparently I can build up one heck of a charge some days.
Michael Kohne

6

My experience seems to be that you'd probably be ok for months or years (depending on the exact components and how much static charge you build up in your environment) but it's not just a theoretical problem and one day it will bite you. Probably at the most inconvenient time and the most expensive way to fix.

And it's not very difficult to protect against it so just get into the habit of doing it. I've not taken my own advice there, but I really should and do intend to!


3
this is the case for most people. I still have never had a problem at work, and we are on carpet.
Kortuk

6

Yes, you definitely should be worried about it, especially if you're working with CMOS (which is about every digital IC nowadays).

Static electricity can easily build up to several kV on a dry winter day, walking with your leather sole shoes over a synthetic fiber carpet. Just reaching for the door handle may draw a spark from your fingertips and give you an unpleasant shock.

What's the risk with CMOS? CMOS inputs in digital ICs like microcontrollers have a very high impedance; if you apply a voltage it has nowhere to go. And a MOSFET's gate is isolated by only an extremely thin insulation layer from the FET's channel. So if you're charged to say 4000V and you touch the microcontrollers input pin the charge will find the easiest way out, and that may be the gate's insulation layer. If you're lucky the charge doesn't damage the insulation layer passing through it. But it's also possible that it gets damaged, destroying the MOSFET, possibly rendering your device useless.

This may happen when handling a part lying on your desk, but it's also possible in finished products. Buttons are meant to be touched and often leave an air gap through which a discharge can reach a PCB. You can apply mechanical measures like lengthening the path by making it a labyrinth. And you can protect susceptible traces with fast ESD protection devices like transient suppression diodes or varistors.

ESD threats you hear about are real. When working with ESD-sensitive devices at least use an antistatic mat and an antistatic wrist strap.


4

It interesting how people say something to the effect of "I don't follow ESD precautions, but you should" It's OK ... we never live up to our own ideals. I think your question is like "should I get seven to eight hours of sleep every night?" The answer is "yes, getting enough sleep is important ... but sometimes it doesn't happen"

ESD precautions are particularly important in manufacturing settings where you are making a product and you don't want to deal with field repairs. Typically damage from ESD will not be immediately apparent, but might cause a product to fail after months or years of use.

It's a good thing to be in the habit of using basic ESD precautions. That being said, I wouldn't be terribly stringent with yourself about it for most basic work. If you occasionally pick up a chip without a wrist strap on, don't feel like you've committed a sin.


I actually intentionally work on a bench that has conductive paint. It is tied to ground properly and I always keep my forearms or elbows touching it as I work.
Kortuk

1
It is not that I think I will fry anything. It is that bugs take weeks to find already, if I spend weeks and it ends up something gets ESD'ed I may think it was a bug and lose a real one. Or actually waste weeks on ESD.
Kortuk

3

If you're a hobbyist, then you don't really need to care about ESD protection.

If you're building prototypes for an actual project, then you should be concerned about ESD protection...but in reality you probably won't bother too much, aside from periodically touching the nearest convenient grounded piece of metal.

If you're building production equipment that will be sold to paying customers, then you must be concerned with ESD protection. That means a strap and a mat at all times in an area with no carpet.


2

My answer is pretty straight forward. I learnt my lesson the hard way.

Just use ESD protection.

Wrist straps are annoying, yes. At the very least, go and get a big and good quality ESD mat and put it on your desk. It makes it much more unlikely that you will touch anything before putting your hands on the table. I tend to rest my arms on the table when working anyway.

My story is that I thought everything was fine, until I got a particularly cold winter and wore the wrong sweater just one day. I messed up all my AVR programmers and dozens of my own boards so that they now work "sometimes". this was actually because my ESD mat was bad quality. This cost me weeks. Its just not worth it.


1

We never had any problems with ESD neither in prototyping nor in production and we don't use any wrist-wraps or other stuff. OTOH we don't have any carpets and do not wear wool or synthetic clothing. Because of this we never had a static discharge from you bodies at work.

I would suggest fighting with static build-up at the source if possible (check you clothes, furniture and humidity) and after you've done this, you may buy some human grounding equipment for double protection.


1
Most devices are damaged at a level of static discharge far below what you can feel. On that note, I turned on an LED with static once and it still works.
Kortuk

The board still works I should say, it was one part.
Kortuk

To damage anything "normal" you must cause quite a current. To do this to modern, protected chips through standard human body resistance you need quite a voltage so I still think that if there are no sparks then you should be safe. In production we sometimes get returns from our customers but never any of this could be tracked down to a possible ESD issue. For an example of an issue you should worry about (albeit this one did not pass our Q&A): we once had some eerie problems with one NAND gate which was left with floating inputs. It worked with a chip from one vendor and failed with another.
jpc

SUMMARY: In prototyping, unless you got full-custom ASICs or a chip worth more than $50 (translate to you own currency and financial threshold), you should not worry about ESD too much. We find normal precautions, without straps, sufficient. There are multitudes of other very cool failure options you should fear! (like mistaking Vcc & GND ;-)
jpc

0

old thread but for anyone reading now. someone mentioned touching the radiator to discharge, that's a very common misconception! please excuse my grammar and terminology, not a native speaker. but basically the damage occurs when differently charged objects come into contact, the sudden equalization of electrons causes the damage. if you earth yourself, you might be at ground charge, but the chip itself might have a charge build up, if you're earthed and you touch it there will be a surge of electrons. that is why ESD straps are not just a copper wire, there's a high value resistance!!! so bodies will equalize in charge gradually rather than instantly. so many people think they are ESD safe if they are holding on to an earthed body. WRONG!!

Ao utilizar nosso site, você reconhece que leu e compreendeu nossa Política de Cookies e nossa Política de Privacidade.
Licensed under cc by-sa 3.0 with attribution required.