Help understanding and using TVS diodes.

[sebmadgwick]

I will eventually need to add input protection to a design.  Space in limited and there a quite a few channels, I am hoping that a multi-channel TVS package will offer a convenient solution.  I have not used TVSs before and so have been reading around the subject.  I'm unclear on few details which I am hoping other forum users can assist with.

I recommend these two documents:
- On Semiconductor AND8230/D - Application Hints for Transient Voltage Suppression Diode Circuits
- On Semiconductor AND8232/D - PCB Design Guidelines that Maximize the Performance of TVS Diodes

My questions are:

1) Given that most device requiring protection specify absolute max/min voltages within a fraction of a volt of either rail, why do TVSs targeted towards 3.3 V or 5 V applications typically specify forward/breakdown voltages of the order of 10 V?  Surely this would mean the voltage is clamped 'too late' and the device risks damage.

2) Can TVS diodes be combined with a series resistor to provide protection against DC voltages (e.g. 5 V signal on a 3.3 V input)?  If so, then how would values be chosen given that TVS datasheets only specify transient characteristics?

Related existing posts in this forum:
- Topic: Rule of thumb for using TVS to continuously sink current
- Topic: Transient-voltage-suppression (TVS) diodes to protect IO Pins
- Topic: Bidirectional TVS a better choice?

[T3sl4co1l]

1.  Two things.  You can use clamp diodes, e.g. BAT54S is very common (or something smaller if capacitance is critical), so the input doesn't exceed the power rails by much.

In either case (clamp diodes or TVS), the voltage will still spike beyond whatever the nominal 'catch' voltage is (due to ESR and ESL), so you add some series resistance between TVS/clamp and the IC.

That way, instead of a 2kV ESD spike going into the chip (probably all it's rated for, if that), you can get an 8kV spike going into the circuit, which is attenuated by the outer TVS/clamp to maybe a 50-100V spike.  The trick is, if you connected *that* directly to the pin, it would be almost as bad as no TVS at all, because that 50-100V spike now has a very low impedance.  So you add a series resistor to let the IC's clamp diodes fill in the rest.

Any resistance is better than none, but 10-100 ohms for digital or non-precision analog outputs, or 100-10k for inputs, is usually good.

Oh, and don't forget ESD applies to inputs and outputs.  Just because it's driving a signal doesn't mean it's any more robust than the rest.

As a bonus, you can add a cap to ground after the resistor (for inputs), which accomplishes some filtering.  Or for resistors under 100 ohms, a ferrite bead may be helpful too.  Good for EMC.

2. Avalanche TVS are just big fat zener diodes rated for peak currents, so you could use them for clamping inputs (casually) or regulating supplies.  But I wouldn't recommend it.

For inputs, you can make excess-voltage-tolerant inputs by adding series resistance in front of the TVS.  The resistor will probably arc over under ESD conditions*, so you can't count on it to bear anything under that condition, but under DC conditions, a hundred volts on a high enough resistor value, sure, do that all day no problem!  (You also get a little implicit EMC filtering, because the TVS/clamp has some capacitance, which the resistor works against as a filter.)

Note that the input equivalent circuit changes depending on what voltage you apply, so it's no longer just a resistor and a pin capacitance kind of equivalent, but there's diodes and clamps and stuff in there.  If you need a very consistent input (e.g., oscilloscope 50 ohm / 1M + 20p matched), you'll want to consider the input circuit in more depth.

For protecting supplies, yes you could use a TVS as a regulator diode, but I would much rather suggest a proper method, like an LDO to ride out the swell plus a low voltage MOV to absorb the energy (e.g., automotive load dump), or a thyristor type TVS (or one of your own fashion) to crowbar the input.

*One example offhand:
http://seventransistorlabs.com/Monitor/Images/NeckBoard2.jpg
No idea if the 22.0 ohm 2W? resistors are, like, HV or arcover-prevention rated or something, but they certainly had some idea what they were doing here.  The problem with a CRT is, sometimes the high voltage arcs over internally (which would be equivalent to something like 30kV machine model -- NOT a kind discharge!), and that needs to stay away from the >100MHz bandwidth video amplifier: low capacitance, high precision, very sensitive!

What they do is, neon tubes (they're probably not actually plain old NE-2, but GDTs that happen to be made in that style) clamp the brunt of the spike, probably still passing a few nanoseconds over a thousand volts -- GDTs don't turn on instantly.  The resistors absorb this remainder, and some low capacitance diodes near, or inside, the amplifier handle the rest.  The amplifier internally is probably bipolar, so it won't be extremely ESD sensitive, but it will be made from small enough bits that it still needs protection against that remainder (I didn't get a picture of what parts are under the heatsink, and there are generic SMTs on the bottom side, but I think there are additional clamp diodes in there).

Tim

[sebmadgwick]

Thanks for the explanation, Tim.  I am happy with the ESD side of protection but I am still not clear if TVSs can provide protection against other unwanted voltages.  For example, 5 V signal on a 3.3 V input, or a +/- 15 V RS-232 input on a 3.3 V UART input.

I am already familiar with how diodes clamping to the rails provide protection.  Such a design should ensure that the maximum forward voltage summed with, or subtracted from, the rails is within the absolute voltage limits specified by the device to be protected, and that a series resistor limits the current through the diodes.  Such a design would be straightforward to verify given a design specification and datasheet characteristics.

What I do not understand is how TVSs with forward voltages of ~10 V will protect a pin.  Either rail plus 10 V is well outside the expected limits of an ICs.  You suggest that the TVS will not protect alone and that a design should rely on the "IC clamp diodes" (via series resistor).  Most ICs do not characterise their clamping diodes in much detail, if at all.  How can such a design ever be verified?

[T3sl4co1l]

Well, if you need absolute verification, you can ball up in the fetal position and cry...   For the rest of us engineers, there are some things we can take for granted:

- If the chip is rated for ESD at all, it must have some sort of diode or controlled breakdown mechanism.  It needn't be very much, but there's something.
- The internal protection has a lot of ESR and voltage drop, so it accumulates damage quickly (dissipation, small area, hot spots..), but that also means it's not completely useless even just to put schottky diodes to the supplies, at the device, without any series resistors.
- If the chip is good for e.g. 2kV ESD, they mean a direct hit to the device.  Even if you bump that down to a paltry 500V spike (say with a GDT followed by enough inductance to allow it to turn on without the IC clamping all the ESD voltage), the pulse is still pretty short, and if you get the impedance comparable, it'll survive, albeit probably with some incremental damage.

- ESD itself isn't some nuclear blast; it's well defined (of course).  IEC 61000-4-2 specifies a test setup -- ways the pulse is applied, how it's created, and what the pulse should look like.  Namely, an e.g. 8kV test has a 16A peak (or something like that), and uses 150pF + 330 ohm, plus whatever architecture of the ESD gun (it has a long ground cable, so there's some inductance and transmission line effects, resulting in the double humped waveform).  The rise time is very sharp (a few ns).

- When you get down to using discrete TVS or clamps, with ESL, you'll get residual spikes probably under 100V, and not much longer pulse length, but much lower impedance.  So with 100-1000 ohms, it goes back down to regular ESD levels.

In specific cases, like 5V tolerant inputs, or RS-232: if you don't need much speed, you can drop the extra 1V on a resistor and call it good; or, RS232 is current limited, so it's technically perfectly acceptable to short it out with clamp diodes or whatever -- not that you necessarily want to do that, but you could.  (RS232 receivers generally have internal resistors dividing it down, so they don't have to worry about input range within a given supply range.  And beefy ESD stuff anyway.)

Tim

[sebmadgwick]

In specific cases, like 5V tolerant inputs, or RS-232: if you don't need much speed, you can drop the extra 1V on a resistor and call it good; or, RS232 is current limited, so it's technically perfectly acceptable to short it out with clamp diodes or whatever

As I said, I am happy with ESD.  I am specifically asking "if TVSs can provide protection against other unwanted voltages", and gave the examples of protecting a 3.3 V pin from a 5V DC signal or from a +/- 15 V square wave.

I'm fairly sure that TVSs don't have any extra magic in them that I have not yet understood and that they are not appropriate for the purpose I have suggested.  If the pin absolute limits are -0.3 V to 3.6 V, then I will need to find a 300 mV forward voltage diode pair.

[T3sl4co1l]

Oh, the qualifying that part, right...

The most strict interpretation of pin ratings would be only within the axes of current and voltage limits, i.e., -0.3 to +3.6V and -10 to +10mA inclusive.  If that's part of your qualification... TS, you'll have to do it the hard way.

A more lenient interpretation would read, well any voltage is okay as long as the current doesn't exceed +/-10mA: and you're only guaranteed -0.3V to 3.6V where that will be true.

Which has physical relevance, because those diodes are going to be pretty low Vf at 150C, so you can't guarantee that more voltage can be applied absolute.  The actual limiting voltage might be -0.7V to 4.0V at room temperature, and more might be fine for short duration (Altera FPGA datasheets specify a little bit of overshoot is tolerable for a few nanoseconds).

Ancient CMOS was prone to latchup and ESD, but anything modern claims "freedom from latchup", which usually means something like 100mA into the inputs [probably] won't cause problems, and they're rated for 10mA DC (or whatever) forever.

TVS diodes aren't specified for DC clamping so no, you wouldn't be able to guarantee such things with them.  You'd use regular zeners, which are fine for low currents and have stable knee voltages.  Not that 3.0V zeners are very sharp; if you need 0.3V strictness, you'd be better off with a diode to a rail slightly below VCC, or a BJT clamp (similar idea but base voltage can be set with a resistor divider pretty easily).

And to combine both (ESD protection on a clamped input):

Zeners, in turn, aren't rated for TVS duty, though they aren't fundamentally different, and you could probably use a 1N52xx in place of a small TVS just fine (a 1N47xx or slightly larger is probably on par with an SMAJxxA part).  Which was done, back in the days before such protection became necessary.  Again, for quality tracking purposes, you'll probably need to use both; on the plus side, the TVS can be whatever voltage, since the inner zener, and resistor, and input diodes, and whatever else, will contribute even more resilience.

Tim

[Mad ID]

1.  Two things.  You can use clamp diodes, e.g. BAT54S is very common (or something smaller if capacitance is critical), so the input doesn't exceed the power rails by much.

In either case (clamp diodes or TVS), the voltage will still spike beyond whatever the nominal 'catch' voltage is (due to ESR and ESL), so you add some series resistance between TVS/clamp and the IC.

From your posts I can see that you are an expert and would like to discuss this topic a bit more.
In my current design I have a 32 analog inputs (0-10V) going to CD4051 muxes (VCC is 13V) which I need to protect somehow. My first idea was to put an ESD diodes like PESD12.. on each channel (as you say, to absorb the energy).

Question 1) Is the series resistor after the TVS diodes really needed? The internal IC protection should not have a problem with 10's of volts leftover from a huge ESD spike.. Right? Can you provide some details or prof that this resistor is needed and in what case? I know that a resistor between the connector and the TVS plays no role in ESD spikes due to arching (it can probably be done with small SMD resistor if some kind of spark gap is formed at the connector pin).

Question 2) Do you think that a schottky clamp like the BAT54S you mention for each pin is a better protection method in this situation and why? I presume that a TVS diode is also needed to prevent the VCC bouncing?

In general, when would you opt to absorb the energy via TVS diode versus diverting the energy via schottky diodes? Here I'm referring only to the ESD spikes, I know that surges cannot be diverted and must be absorbed (like IEC 61000-4-6).

Thanks.

[SeanB]

Put the ESD diodes on the inputs, with series current limiting resistors before and after them, then put in overvoltage protection after that. Use a 10V 5W zener diode, and place a 1N4001 in series with it to increase the forward voltage by 0.6V, and power it from the 13V rail so that the zener is dropping 10.6V, but not dissipating 5W. Then use a diode (1N4001 again or a low leakage diode if the leakage current will cause errors) from each input to the zener diode, so you clamp any input that goes above 11.2V with the zener diode. That way you have protection against minor overvoltage, and protection against very high events, while still not doing much to the input signals.

The reason for biasing the zener on is because it normally is slow to turn on, and biasing it on means leakage will not cause errors, and it will clamp what the transient suppressors let through via the second resistor.

Using SMD parts for all aside from the 5W zener ( no way you will get a 5W device in SMD easily unless it is in a Dpak) it will not take up much extra board space.

[T3sl4co1l]

From your posts I can see that you are an expert and would like to discuss this topic a bit more.
In my current design I have a 32 analog inputs (0-10V) going to CD4051 muxes (VCC is 13V) which I need to protect somehow. My first idea was to put an ESD diodes like PESD12.. on each channel (as you say, to absorb the energy).

Question 1) Is the series resistor after the TVS diodes really needed? The internal IC protection should not have a problem with 10's of volts leftover from a huge ESD spike.. Right? Can you provide some details or prof that this resistor is needed and in what case? I know that a resistor between the connector and the TVS plays no role in ESD spikes due to arching (it can probably be done with small SMD resistor if some kind of spark gap is formed at the connector pin).

I'm expert enough to be dangerous. I don't have data on that (maybe some manufacturers or real-falootin' ESD/EMC experts have data down to the waveform and microscope slides kind of level?), it's more of an informed hand-wave.

The series resistor may not be strictly necessary, but you might as well since it improves things that much more, and is usually practically a freebie.  The source impedance of the TVS/clamp is probably in the 1-10 ohm range, so adding teeny bits <10 ohms isn't going to help much but over 10 ohms it gets useful.

Cases where you can't tolerate a resistor include most outputs*, and precision analog and high speed inputs.

*Although if you need a series terminating resistor, you can use that to your advantage.  You may need a lower drive impedance to compensate (e.g., using a 74AC or 74ABT bus driver over a 74HC standard gate output), to maintain the characteristic source impedance.

USB drivers are usually a good example of this: the internal drivers are plain ordinary CMOS (for low / full speed), with about 30-50 ohms channel resistance.  It's not well controlled, because it's monolithic.  You add the remaining 22 ohms externally (or some have them internally now, possibly achieved with laser trimming or fuses) to get the 60 ohms per side that matches to the 120 ohm twisted pair.  Being source terminated and point-to-point only, no end terminator is required.

Anyway, since you're talking PESD thingies, the residual isn't real great (~30V?), so you'll want series resistors there.  Even if your circuit is precision analog, I'm guessing its load resistance isn't all that low (CD4000 analog switches, ~100 ohms), so an extra 100 ohm there wouldn't hurt too much.  And if you really want to gild the lily, maybe a BAT54S or BAV99 or something like that, besides (at the chip input).

As for the choice between something like BAT54S and BAV99, the former has a lot of leakage so you might not be able to use that.  Then you have to accept the higher Vf of a silicon diode, and deal with it elseways (like having more stages, or knocking it down beforehand, etc.).

Question 2) Do you think that a schottky clamp like the BAT54S you mention for each pin is a better protection method in this situation and why? I presume that a TVS diode is also needed to prevent the VCC bouncing?

BAT54S is handy when the input range equals the supply range, which is a lot of cases, and basically everything digital.  An example where you might use a zener / TVS could be an analog 0-5V control input, in a 0/12V supply domain -- no use clamping it to the supply, so you use a zener (and input series resistor so it's overvoltage tolerant) to keep it limited to the desired 0-5V range.  The zener does double duty as ESD protection and constraining the operating range.

VCC bounce, you at least need a 0.1uF bypass as near to the ESD clamp(s) as possible.  That keeps the peak down, and turns it into a small surge instead.  For ESD, a TVS in the supply probably isn't important (it's not much charge compared to the low impedance supply network), but yes, I would recommend it anyway, and besides, you're protected in case of supply surges or accidental reversal.

Note that, again, low voltage avalanche sucks, so although there's such thing as a 3.3V TVS, they have so much internal resistance that they're basically 5 or 7V devices under surge.  If you have sensitive stuff on low voltage rails, you might be better off building a crowbar.

Tim

[T3sl4co1l]

Put the ESD diodes on the inputs, with series current limiting resistors before and after them, then put in overvoltage protection after that. Use a 10V 5W zener diode, and place a 1N4001 in series with it to increase the forward voltage by 0.6V, and power it from the 13V rail so that the zener is dropping 10.6V, but not dissipating 5W. Then use a diode (1N4001 again or a low leakage diode if the leakage current will cause errors) from each input to the zener diode, so you clamp any input that goes above 11.2V with the zener diode. That way you have protection against minor overvoltage, and protection against very high events, while still not doing much to the input signals.

The reason for biasing the zener on is because it normally is slow to turn on, and biasing it on means leakage will not cause errors, and it will clamp what the transient suppressors let through via the second resistor.

Using SMD parts for all aside from the 5W zener ( no way you will get a 5W device in SMD easily unless it is in a Dpak) it will not take up much extra board space.

Prebiased zeners / TVS are a good approach.  The reason: zeners big enough to handle spikes/surges have a lot of capacitance, which kills high frequency signals.  So, you dump into it with low capacitance junction diodes.  Downside: junction diodes have forward recovery, meaning you still get maybe 30V for a few ns.

Warning: DON'T USE 1N4001 or anything like that.  Forward recovery will eat your pulse!  Use something fast (and small, besides) like 1N4148, BAV99, etc.

You can buy monolithic packs e.g. for USB, HDMI, etc. with the clamp diodes and TVS included.  The TVS usually goes across the supply, doubling as local surge protection.

Tim

[SeanB]

True, but 0-10V says industrial, and 16 or so multiplexed inputs means they change slowly, so big junction capacitance will not be much of an issue, and the beefy current capacity will count more.