We are making a device with 18 inputs and 12 outputs, plus USB signals. We are concerned about EMI/ESD protections. There is ferric core (for EMI) and clamping diode (for ESD) on the USB line. Should we also protect the I/Os against ESD? Are inputs more critical than outputs? We can also change the buffer/decoder chip. Is older fab tech (0.5 vs. 0.35 micron) better for ESD, with bigger components? Thanks.
c
Absolutely. Spark gaps, the biggest series resistor that will still allow the circuit to function correctly, caps to ground. Anything to slow down that energy pulse and keep the peak low.
Optoisolation is better still if you can manage it.
Your hunch that older, bigger geometries survive ESD better is in general true but only 100% guaranteed if you're talking about "old version of chip A" vs "new die shrink of chip A". Different parts, even on the same geometry, can have wildly different ESD survivability.
ric
I am thinking about clamping diodes. Caps might slow it down too much and not fast enough for xxxxKV ESD.
Too expensive.
Might have to go with HC, HCT or LS buffers/decoders. The CMOS micro ports are definitely not good enough.
Cost/function trade off... For us the majority of inputs requiring protection are DIP switches and the like, so they are read very slowly. Our standard protection circuit is spark gap to ground (poor man's voltage clamp :), cap to ground and then a series R going into the IC.
Think about where your I/O's are going. E.g., chances are, the USB port is "exposed" to the user "casually". If, OTOH, the other I/O's are tied to the field in such a way that they aren't
*intended* to see "service", you can dramatically change the signal conditioning that you deploy to better suit the service they *will* see.
ric
The I/Os are routed to a connector, then a flexible cable of 30cm. But it could be subject to ESD while connect/disconnect the cable during installation. ESD protection will double the cost (mainly 4 layers vs. 2 layers PCB) for added components. Short term (5 to 10% returns due to damages) is manageable, but long term might not.
USB is fully protected, since there are only two wires.
Bullshit.
OK, instead of 30 BAV99 at 5 cents each. What would you suggest?
Maybe not. Look, there is a spectrum of protection vs cost, I can accept that. In fact in my day job we do not use optoisolation for this reason exactly. But when I've done a contract vending machine project, I used optos on every input because my experience with coin- op arcade games was that they frequently die when people use piezo zappers on them...
Unless you are making xM numbers of this device, you shouldn't be pissing and moaning over making it ultra reliable when the difference isn't that great. Especially if you indeed intend to incorporate reliability and integrity into the design.
Of course transzorbs will work, but if you DO want guaranteed isolation, you want opto-isolation.
You asked the question, and got the answers. It is up to you to decide what level of protection you actually want or need.
In your requirements analysis, you decide where the crossover between cost of manufacture and protection level compromise each other.
Coin mechs and the doorways have been isolated from game circuitry for years.
That zap crap only worked on older machines.
Reliable at reasonable cost only, if the device fails, the customer can just dump the $50 device or replace the $10 PCB ($12 in the future). $1.50 in diodes is already 15% of cost, not to mention another 50c for PCB. I am still open for suggestions up to $2. Of course, if it's all up to me, I would redesign the housing for better shielding and grounding. But I can't change the world imposed on me.
cide
... and the newest machine I ever owned was Legend of Hero TONMA (or maybe Thunder Fox - whichever came out later). Most my machines were System 16 or older (Ghosts'N'Goblins, etc)
I have been around since the first upright machines hit the market.
Just get Mame and the ROMs and your set.
Why? ESD will kill an LED or a phototransistor, so you've got to protect the inputs and outputs of the optos. So you may as well save money and just protect your ICs.
RCs are good. There are also some specific ESD protector chips, essentially a lot of zeners in one small package. It's generally good policy to never run an unprotected IC pin signal out of a box.
John
yes, unless you have optical protection on the inputs, it's very important to protect the inputs with TVS type diodes.. And you may want to use the unipolar type, if the input is sensitive to reverse polarity.
Hi linnix,
The principle I generally use is to limit the voltage peak with a parallel clamp, then limit the current pulse from the remaining over-voltage with a series element. The parallel clamp can be e.g. TVS array, SMS05 etc. Perhaps a capacitor if that is too expensive. The series element can be a resistor for hi-impedance signals (inputs) or a ferrite bead for low impedance (outputs, power).
You can get all of the above in array format which is generally cheaper and saves space.
(I have never needed 4 layer to achieve this but YMMV...)
-- John Devereux
ESD shouldn't reach the phototransistor, in theory; the whole purpose of an optoisolator is that there is only an optical path between the tx and rx side, and the breakdown voltage between them is a device parameter. Yes an LED will fry, as all things do, but it's much more robust than a CMOS input.
Also when properly employed, the LED will be completely isolated from the circuit (i.e. running a single wire out of the box, going to say the anode of the LED, with the cathode going to a resistor straight to the same ground as your IC is NOT the right way to do it). So the ESD pulse will reach both sides of the LED at the same time.
John was referring to outputs. The phototransistor is hanging out in space.
Nonsense. ESD protection has to be designed for a single-pin hit. You cannot rely on common mode.
You should really do some SPICE analysis on your circuitry to understand what is going on in your input and output circuits. Both are vulnerable (and this includes the Power Supply as an input). You will need some added capacitance to slow down the rise for the clamping diodes anyway (to allow them to operate properly).
In comparison to losing the whole unit?
Think of your precious circuitry as needing the protective barrier of isolation from the nasty real-world. Opto-isolators, isolation transformers, series resistance, capacitance filtering and clamping diodes all working in harmonious unison will keep your sensitive micro-controllers safe from harm. Changing a burnt out opto-isolator is cheaper than dumping the whole board- full of components.
I would normally point you at my articles "Reading the World" and "Writing the World" but I am between web-sites at present and I don't know what has happened to the UK Forth Interest Group web-site.
[%X]-- ******************************************************************** Paul E. Bennett............... Forth based HIDECS Consultancy Mob: +44 (0)7811-639972 Tel: +44 (0)1235-510979 Going Forth Safely ..... EBA. www.electric-boat-association.org.uk.. ********************************************************************
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.