What is a spark gap, and why we need it on a PCB?

[Azhar]

What is  the benefit of having a spark gap on PCBs?
See Dave's video here  http://youtu.be/_EMAggN5H-w?t=15m43s
or check the photo below

[T3sl4co1l]

Think there was a Y-type cap missing nearby on that particular device, or something like that, so I think it did nothing...

But, on the version where it is used (or if I'm just misreading it, and it's always in use), the reason would be to bypass very fast surges or spikes (EFT and ESD) so they don't potentially damage the insulation or isolation of components (usually common mode chokes, transformers and Y caps).

I don't know what standards, if any, suggest this; I've seen it occasionally, and it never really made much sense.

Tim

[suicidaleggroll]

I think it's used as sort of a ghetto ESD or over voltage protection.  The idea is that the high voltage will ionize the air between the two terminals and spark across them before it damages the rest of the circuitry.  As far as I know, you can think of it like a significantly less sophisticated MOV.

[jahonen]

Putting a spark gap on the PCB is much cheaper alternative to having a gas discharge tube component, but not as robust.

Regards,
Janne

[Wytnucls]

A good Microchip article on the subject of ESD protection:
http://ww1.microchip.com/downloads/en/DeviceDoc/EMC%20Newsletter%20Issue%205.pdf

[Azhar]

Thanks all, that was helpful

[Neilm]

Not just for ESD - I have seen some multimeters with spark gaps to protect against transients on the lines. A fuse upstream was used to extinguish the arc

[DeweyOxberger]

A few lessons learned about spark gaps on pcbs:

1)  You probably don't want a sharp point.  Sure, physics says you do but physics is whack.  Reality says the point carbonizes after just a few arcs.  The longest lasting spark gap footprint looks just like an 0402 component that is missing.  One spot across the gap will naturally arc first (think "micro sharp spot arcs first"), burn out after 2-3 arcs, and some other spot will become the new sharp spot.  You will get a few thousand hits before the entire thing wimps out.  Gold flash holds up better than solder finish.

2)  Any line that can see ESD could use one.

3)  It takes a variable amount of time for the gap to kick in.  Once it kicks in it shunts some of the transient.  That can really help a circuit survive repeated ESD hits.

4)  Spark gaps alone never work.  You need still need series elements (resistors) and clamps to help deal with the initial fast edge and the (lower than without an spark gap) sustained voltage.

5)  In dry conditions the gap's arc voltage is really inconsistent.  I'm fairly sure charge can splatter across the surface of the PCB and get stuck there.  That charge changes the ambient e-field at the gap and changes the arc voltage.  Wiping down the board after an arc can remove the charge and improve the performance of the gap.

Here is an arc gap calculator I worked on years ago:

http://www.cirris.com/learning-center/calculators/50-high-voltage-arc-gap-calculator

[SeanB]

A lot of older alarm panels included them, as they are free with the PCB design, and cost nothing aside from a small amount of board space, which might be unusable in any case as it would be there for the terminals. As the breakdown voltage is not controlled well, they are only used where you need some protection and the input will survive this poorly defined transient, and the wiring resistance acts as the current limiting. Ground side was always a thicker trace going to a substantial earth wire ( at least compared to the 20SWG alarm wiring) that went to mains earth. For those inputs and outputs that were more sensitive they included 'expensive' zener diode protection.

[DeweyOxberger]

A few more items I remember.

6)  They can cause problems in manufacturing.  They are likely spots to short if your solder process gets a bit sloppy.

7)  To use them properly they need to be right on the edge of the board where the signal comes in.  They should shunt to ground and that ground should connect back to the case or Faraday shield right at that point.  That will keep as much energy off the board as possible.

[WarSim]

Why does every high tech industry always mix up terms over the years.  What it meant then is not what they mean now.  Now if you read a 50year old technical document with today's terms the details are lost. 
For example originally a spark gap was a device for the intentional exchange of energy through a medium.  An arc gap was a device to accept energy anomalies.  Ergo spark gaps where operational ccts arc gaps where safety ccts. 
Yes spark gaps have been and are used as functional devices.  The most common being a spark plugs in a car. Less commonly used in very high power transmission
The term differences are important to differentiate between the different types.  Spark gaps had more stable mediums resulting on lower ratings but longer life.  Arc gaps had much higher ratings but only survived about 100-5000 events.
Now the only term I hear now is gas discharge tube with no defined function.  Non of the other types mentioned for years even though they are still manufactured. 
The newest term usage of Spark gap for the PCB sacrificial break over gaps is amusing.  It uses the arc gap concept with no stabilizing medium nor physical contractors.  It is extremely low duty, contaminate medium (free air makes it not calibrated), very low repeatability (well into the sacrificial rating range).  These board elements are not ESD protection features they are barely better than a used fuse wrapped in tinfoil.  They do have their uses to protect against one catastrophic event, but other methods need be also implemented for actual reliable protection. 


Sent from my iPad using Tapatalk

[orion242]

I still see them on almost all our commercial controllers today.  Just as described, right on the edge of the board, tied to ground and as close to the inputs as they can get.

[EEVblog]

Just shooting a video on this now!

[Azhar]

That's great, thanks a lot Dave! what about these neon lamps I've seen them in TVs, Old phones and they are not used as lamps, they are used as something else I believe, as a spark gap maybe???!

[mmirabent]

Those neon lamps are probably part of a relaxation oscillator. Commonly used for sweeping the beam on CRTs or other places where a sawtooth signal is needed.

[T3sl4co1l]

Those neon lamps are probably part of a relaxation oscillator. Commonly used for sweeping the beam on CRTs or other places where a sawtooth signal is needed.

Oh my, nowhere near stable enough! 

You'll find them under GDTs (gas discharge tubes).  Both uses are for transient protection: the advantage is high peak current capacity, low voltage drop (conducting state), and reasonably fast response time, while looking like essentially an open circuit (hardly any leakage, a few pF capacitance).  The downsides: they aren't too stable (you need to choose a device with a typical breakover voltage significantly above your nominal operating voltage), the voltage overshoots significantly before turning on (the residual looks something like an ESD zap), and they stay on as long as current is maintained (a latching element, like a thyristor TVS, or related devices -- SIDACs, SCRs, TRIACs).

In phones, they work for lightning protection (along with the isolation transformer).  In TVs, they're most likely to be found near the CRT (from http://seventransistorlabs.com/Monitor/index.html ).

Tim

[orion242]

Just shooting a video on this now!

Let me know if you would like some commercial HVAC/Lighting controllers for a tear-down.  I have plenty of material I would be willing to send over dating from the late 80s to today.  Dissection of the input / output protection and power supply might be interesting.

Come to think of it I have a wad of 4-20ma sensors to go with it.

[a210210200]

Spark gaps are not meant to protect against sustained over-voltages and since most will latch on activation they are meant to blow a fuse, circuit breaker or something to protect the downstream devices. I'm not sure if they are relevant to something like ESD protection for human static discharge cases as their properties are pretty hard to control especially PCB made ones.

I find it very interesting in protecting against locally direct lightning strikes or being near a substation or HV fault is actually very hard to protect because it can cause even the ground potential to go crazy which can cause a lot of damage. (Which is aptly called ground potential rise)

[kolbep]

When I worked at a local ISP,
there was one customer that had a lightning strike somewhere along the overhead telephone line leading to his house.
It melted the cable all the way down the wall inside his house, and literally split his 56k modem in half.
I don't think the built-in gas arrestors or spark gaps helped too much.

[kolbep]

Dave already did that on one video,
But the sparkgap he was testing was the range switch on a cheap POS multimeter.
It did spark across the gap quite nicely.

[coppice]

When I worked at a local ISP,
there was one customer that had a lightning strike somewhere along the overhead telephone line leading to his house.
It melted the cable all the way down the wall inside his house, and literally split his 56k modem in half.
I don't think the built-in gas arrestors or spark gaps helped too much.

Just because you try to arrest a spark, that doesn't mean it will go to the station quietly with you.

One of the annoying failures in 1980s modems was a modest discharge passing through the line relay contacts causing a pulse in the relay coil that would blow out the driver transistor. When modems changed from relays to opto couplers for the off-hook switch a lot of people expected the modems to be a lot more fragile. The opposite was actually the case.

[EEVblog]

but what kind of voltage do they arrest? 1kv ?

In theory the lowest you can get is about 1500V
Wait for my video... Patreon & Forum supporters have already seen it 

[DeweyOxberger]

Dave,  The video is spot on.  A few things I'd add to the discussion:

1)  It's always better to keep the spark gap feature away from points being soldered.  Those darn spark gaps tend to create solder bridges.  So your second pcb layout is better than your first by far.

2)  If you really want to become a guru on this stuff get a cheap set of amplified, powered speakers.  Use them to make a "current to audio converter".  (series resistance for the current sense, then ac couple to the powered speakers, put anti-parallel diodes to clamp the signal heading to the speakers).

You will find as you increase the voltage you will hear an increase in the hissing noise.  Well before you see an arc you will hear it shunting energy.  It will pop fairly loudly even with no visible arc.

3) I consider these "lawn mower" devices.  They cut off the top of the transients and shunt the top energy to ground. Even with an active arc you are getting major energy heading toward the pin.  You will still need a PI or T clamp structure on the pin to shunt the remaining energy.

4) Finally, it isn't really an air gap.  The dielectric of the FR4 warps the field in a big way and some truly amazing stuff can happen with charge splattering across the surface of the pcb (think partial discharge "corona", relaxation oscillators).  If you get the magnitude of the electric field at the exposed surface of a metal to 16500 V/inch you'll start sucking electrons off the metal but they won't go very far.  They will land on the FR4 and create gates that block future electron flow.  Given time those electrons will dissipate and more electrons will barf out of the gap. 

Fun stuff.

[Azhar]

Thanks for the video Dave! awesome!

[SeanB]

They are used as spark gaps, as often the drive signal is around 60-80V, and thus the breakover voltage of 100-150v is perfect as protection. This is combined with the built in spark gaps in the CRT base, which is hidden inside the base ring, and which has small gaps and a grounded ring that handles the majority of any flashover, diverting it to a thick ground wire going via a short lead direct to the CRT grounding straps that are held against the outer conductive surface of the CRT glass

[bruce0126]

I know this is an old topic, but I'm just wondering why the solder mask should be removed on the spark gap?  The relative permittivity of solder mask is higher than air, so doesn't the solder mask actually makes it easier to arc over?

[T3sl4co1l]

It could exhibit partial discharge, but then that discharge moves the voltage across the dielectric, and no further action results.  In other words, the tiny capacitance of the air gap (~0.05pF?) discharges, rather than the full load current.  It's like the static crackle when rubbing a balloon on hair.

If this happens enough times to cause both solder mask layers to break down, then a proper arc can start.  Since the solder mask thickness is highly variable, this could be many kV more than intended!

Not that spark gaps are at all well controlled to begin with, but that's just making bad to worse.

It's worse still when you consider the second strike: the solder mask has been pierced, so now it's only an air gap.  Except a small amount of solder mask has been carbonized, and more on each strike after that.  Rapid tracking ensues, and you get a glowing burning ~short circuit.

Tim

[Azhar]

Dave made a video about spark gaps and here is a link in case someone didn't watch it