best cost/design effective ESD protection method and advices

[mcinque]

I need to protect some TTL circuitry from human ESD that could be discharged thru some pushbuttons.

Wich of these three solution is the most cost/design effective in your opinion?

#1 TVS
#2 M.O. VARISTOR
#3 LOW PASS FILTER

Assuming that I must use only TH components (at the moment I'm not still in SMD design  ) and the small size is an important aspect of this design, these are my considerations:

TVS: fastest reaction time, handles easily high voltages and currents, cheap and small in size.
VARISTOR: cheapest solution, but I don't know how to choose his value to use it as ESD protection. Small in size and space-saving mounting.
LOW PASS FILTER: I don't know much about his reaction time. Since the component should be rated at least 1000V his size would be not small and since I must protect at least 5 inputs this could be a space waste.

at last, I've heard form some friends that the most effective solution is to use a 1M resistor in series with the inputs...

Can you give me some advices?

[rob77]

actually it's not so easy to damage TTL chips with ESD those can withstand a decent charge. (even the HC & HCT ones can withstand a lot).
and regrading the push button concern - usually you're pulling down an input to ground with the push button (an input which is pulled up through a resistor to the vcc rail) so the ESD discharge will find it's way to ground or vcc rail (it will not go straight into your input pin).
i've been doing a lot of stuff with TTLs back in time and i never run into a problem with ESD damage.

and quite honestly - i never seen a design where all inputs were protected with varistors - that makes sense only when talking about inputs directly coupled to external world - but who the hell would do that with TTL levels ?

[Niklas]

I would vote for the RC solution. The series resistance will limit the current, lower the risk of latchup and if the current is limited the internal diodes will do the rest. 100 nF seems a bit too large. We used 10k + 100 pF on a multiplexed keypad that was driven from a 3.3V PIC18.

[lewis]

For the ultimate low cost solution, and assuming your device has internal ESD substrate diodes, use a high value resistor in series with the input, in the range of 220K-470K. That's it. (Don't do this on the MCLR pin of a PIC, there is no ESD diode to the positive rail and a transient will put the MCU into programming mode.)

The better solution is to use the RC circuit you posted, but I'd change the 1K to 100K, the cap to 10nF and add a diode in parallel with the cap (anode to ground) and another diode from input to VCC (anode to input). 1N4148's will do. Any transient glitches, positive or negative, will be bypassed to the supply rails via the diodes. The 10nF will help a bit with switch debouncing too.

All this assumes you're using a CMOS MCU, but you mention TTL in the post. The previous paragraph still applies, but 1K is about right depending on the TTL family you're using.

You don't need to use any components that are rated to 1000V, normal components will be fine. A small radial ceramic capacitor will easily survive an ESD event if there is a series resistor in place. After all, the capacitor is there to limit the voltage at the input pin. The resistor bears the brunt of the transient voltage, and any bog standard 1/4W axial metal film resistor will be absolutely fine.

Varistors are great for high energy transients but are relatively slow, and TVS diodes are great for very high speed and medium energy applications but will almost always fail short given a large overload. But for simple switch interfacing, large series resistor and cap down to ground is very reliable and effective.

[rob77]

For the ultimate low cost solution, and assuming your device has internal ESD substrate diodes, use a high value resistor in series with the input, in the range of 220K-470K. That's it. (Don't do this on the MCLR pin of a PIC, there is no ESD diode to the positive rail and a transient will put the MCU into programming mode.)

The better solution is to use the RC circuit you posted, but I'd change the 1K to 100K, the cap to 10nF and add a diode in parallel with the cap (anode to ground) and another diode from input to VCC (anode to input). 1N4148's will do. Any transient glitches, positive or negative, will be bypassed to the supply rails via the diodes. The 10nF will help a bit with switch debouncing too.

All this assumes you're using a CMOS MCU, but you mention TTL in the post. The previous paragraph still applies, but 1K is about right depending on the TTL family you're using.

You don't need to use any components that are rated to 1000V, normal components will be fine. A small radial ceramic capacitor will easily survive an ESD event if there is a series resistor in place. After all, the capacitor is there to limit the voltage at the input pin. The resistor bears the brunt of the transient voltage, and any bog standard 1/4W axial metal film resistor will be absolutely fine.

Varistors are great for high energy transients but are relatively slow, and TVS diodes are great for very high speed and medium energy applications but will almost always fail short given a large overload. But for simple switch interfacing, large series resistor and cap down to ground is very reliable and effective.

for TTL the 1K in series with a input might result in a weak pull-down - to have a good noise immunity you should pull down the TTL inputs hard. and as i mentioned TTL is  quite immune to ESD - it's bipolar. if we're talking about 74HC/HCT devices - then those are not TTL - those are pin compatible CMOS equivalents of the 7400 series made all in cmos. the HCT series have logic levels compatible with TTL, but those are still not TTL.

overly protecting TTL circuits make no sense IMHO.

[SeanB]

220R in series with the input, and a 1k pull up, along with 100n to ground and clamp diodes (1N4001) to both Vcc and Gnd. Will work with all TTL, even plain old 74 series. Has quite good noise immunity, and does not rely on the internal diodes of the TTL. Adds a little debouncing and current is high enough to keep the switch contacts clean.

[mcinque]

thanks to all!  Your suggestions are greatly appreciated

 I will definitely try with the series resistor and cap to ground.