PCB Creepage and Clearance requirements from around the world....

[Smokey]

Since there are engineers from all over the world and from all different industries here, I thought it would be interesting to hear what you guys use as far as creepage and clearance rules in your PCB designs and why.  What sort of voltages do you work with?  Are the rules in your country/industry not as restrictive as others?   What do you usually design to: UL, IPC-2221, something completely different?  Do you use different rules for different markets?  Horror stories about getting products certified.  That kind of stuff.  Could be interesting.

[Smokey]

Hmm.  Seeing how much talk there is about meter voltage clearance and safety issues I thought I would get at least a couple hits.  Go figure. 

[PChi]

Just to get the ball rolling.
It's many years since I was involved so I can't remember the document numbers.
For general purpose home and office I think that European creepage distance is 5 mm. It used to be 8 mm in the UK.
Insulation thickness is 1 mm or any two layers surviving some voltage which is how the triple insulated transformer wire passes.

[Neilm]

I work designing equipment to IEC61010. The clearances are not only dependent on voltage, but also on altitude, operating conditions,  pollution degree, material types and so on.

The highest voltage I worked with was 100kV and the clearance for that was massive.

Neil

[Mechatrommer]

some of the threads i brought this thing up
https://www.eevblog.com/forum/beginners/let's-review-mains-voltage-breakdown-and-min-copper-clearance

hvalac provided the http://www.smps.us/pcbtracespacing.html
in https://www.eevblog.com/forum/projects/oshw-diy-1kv-100mhz-differential-probe-(dilemma-vs-hope)/msg117052/#msg117052

its just me.

[Niklas]

I have made some circuit boards for household appliances that have been third party tested for CE compliance. The EN60335 standard for household appliances points towards some other, more specific standards about isolation, EMC etc. Then you might have product or country specific addendums on top of that.

2 layer PCB
230/400Vac, 3 phase and PE
PCB inside a IPX5 rated box, pollution degree II
Functional insulation, 2.5 mm, between L1, L2, L3 and N
Basic insulation, 2.5 mm, between PE and all of L1, L2, L3 and N

Traditional 50 Hz transformer and 12Vdc SELV on the secondary side, reinforced insulation. Transformer specified and tested to 4 kV between the primary and secondary windings.
Creepage distance between primary and secondary on FR4 laminate was set to just over 5 mm. I can not remember if we specified the CTI value of the FR4 laminate.

4 layer PCB
Frequency inverter
230 Vac, PE input
PCB inside a IPX5 rated box, pollution degree II
PFC with stepup to a bit over 400 Volts DC, the PCB was designed to handle 500 Volts.

Problem areas:
Heatsink at PE potential fixed with metal screws through the PCB
MOSFET drivers with bootstrap capacitors and a diode. SO16W packages and only 2 NC pins bridging 400V+
SMD ceramic snubber capacitors on the MOSFETs. SMD because of the layer stackup with GND+GND+GND+400Vdc
Three output phases with 400V AC
Phase voltage feedback to a CPU PCB at GND and +5V. Several 1206 resistors in series to handle voltage rating and creepage
Current feedback with Allegro ACS712 in SO8 packages (2.5kV insulation)
The grounding solution made us insulate the external control signals with optocouplers

L-N, functional insulation, 2.5 mm
L and N to PE, basic insulation, 2.5 mm
Output phase to output phase, approx 4 mm
Generally I used just over 4 mm and some problem areas were treated with conformal coating after soldering. The board itself used 70 um base copper without extra plating and our board supplier did not recommend us to combine that together with tougher CTI specifications.
To handle the event of PCB delamination, the layer to layer creepage was measured as: Copper->hole/edge, thickness = 0, hole/edge->Copper

Don't just enter these numbers into the DRC and hope for the best. Do check which standards and addendums that apply for your product! In one of the addendums we had, there was a paragraph that read something like this: "Take the creepage from the next higher voltage class in table YY and then add X mm to adjust for mass volume production".

Regarding certification body testing, make sure that you are prepared and have done your home work. By selecting components, like plastic film capacitors, connectors etc, that already have been tested in some way helps a lot. Glow wire ignition tests takes time and had to be done on components without a trusted test certificate. UL and VDE have online databases listing components and manufacturers with valid test reports. Whatever you do, do not trust any chinese supplier as they tend to "borrow" test reports from each other. Always double check test reports that you get from a supplier.
Panasonic did not want to release anything more than the datasheet for their ECQUL X2 capacitors. The datasheet claimed that they will not burn, but without any reference to UL 94 V0 or something similar.

[AlfBaz]

Here's an uncited document from 10 to 15 years ago that was used in a workshop that manufactured industrial control panels. I beleive it was for reference only
Bear in mind that this is/was an uncontrolled document and as I don't have access to AS/NZS 3439 (the Australian/New Zealand standard that I beleive holds this information) I can not verify this imformations accuracy, or lack there of. As such I am providing this soley as a loose reference point.
   

Minimum Clearances & Creepages in air
	Up to and including 125V	Above 125V up to and including 250V	Above 250V up to and including 660V	Above 660V up to and including 1100V	Above 1100V up to and including 3300V	Above 3300V up to and including 6600V	Above 6600V up to and including 11000V
Clearance Phase to Phase direct	10	13	19	25	51	89	127
Clearance phase to earth direct	10	13	19	25	51	64	76
Creepage phase to phase over insulation	13	16	25	32	76	146	216
Creepage Phase to earth over insulation	13	16	25	32	51	89	127
Creepage phase to earth over cable surface	19	25	33	51	76	127	178

EDIT: all measurements in mm

[AlfBaz]

Yes they are mm. Edited the post to add that.

Even condsidering Niklas specs I don't understand how something like a BTA20 triac with 600/700V Vdrm in a TO220AB package can pass without additional protection between pins.

[nctnico]

The absolute minimum distance is determined by the dielectric strength of a material:
http://en.wikipedia.org/wiki/Dielectric_strength

For (dry) air that comes down to 3000V/mm. So its no problem for a TO220 package to handle 700V. The minimum creepage distances in safety regulations take polution and codensation into account. And for a good reason: when I got my TDS744A scope I didn't let it warm up long enough so the first time I turned it on it went out with a loud bang. It appeared the wires on the mains filter where too  close to the case causing a discharge to the casing. I cut the wires shorter and applied insulation on the casing.

[AlfBaz]

...So its no problem for a TO220 package to handle 700V. The minimum creepage distances...

Creepage is the part that concerns me. Where the pins enter the body and A1/A2 are side by side the creepage distance can be as low (taking into account tolerances) 0.7mm
It would be interesting to run a 500 or 1000V insulation tester between the two pins and compare the difference between a bare part and one coated with some conformal coating

[Smokey]

Awesome replies.  This is the good stuff.
Ya the specs usually don't have anything to do with breakdown through clean dry air, which needs a surprisingly small clearance to keep from arcing.  It's more about the environment and potential for other stuff which can make an arc like particulate or humidity.  That's the part that makes the specs hard to define and apply since you need to understand all the possible conditions you might encounter, and all the various spec standards that govern those.

I'm not sure it could even be done, but there should really be a flow chart that starts with your operating environment and voltage range, has all the various standards and in which countries they apply, and ends with the appropriate spec you need to follow.  I would pay for that.  Without that you have to know what all the spec standards are, have access to them(some you have to pay for), and go through them individually to figure out what value from what table is applicable.  That's a lot of boring work.

Then you run into the manager who say things like "Well look in the lab.  This is only X-mm spacing and it's not arcing right now is it??  It's FINE! Don't worry about it."   

This is almost one of those cases where you could make a ton of money by charging way too much being an expert consultant doing nothing but advising companies on what spec to follow.  It would probably be cheaper for the company than paying someone to figure it out from the beginning.

[Niklas]

Yes they are mm. Edited the post to add that.

Even condsidering Niklas specs I don't understand how something like a BTA20 triac with 600/700V Vdrm in a TO220AB package can pass without additional protection between pins.

Sometimes you can see TO220 packages with the middle pin bent out a bit to handle clearance and creepage, on the PCB. The distances are still the same just where the pins enter the package. A guess from my side is that a triac och a transistor is perhaps not considered as an insulating device as A1/A2 or Drain/Source can be connected together by the device. Compare to a relay's coil and switches.

Generally the minimum creepage distances are larger than the minimum clearances so a TO247 might pass on clearance but fail on creepage. We had a lot of 1 mm milling lines in a prototype to increase the creepage distances in the inverter I mentioned. It was a pain to handle all the layer to layer creepage distances and also the mechanical board strenght was affected. By using a board trading house instead of a local manufacturer also ment problems as the prototype batches differed in flexibility. Removing the milling lines and adding conformal coating was our solution on that PCB.

[Neilm]

I'm not sure it could even be done, but there should really be a flow chart that starts with your operating environment and voltage range, has all the various standards and in which countries they apply, and ends with the appropriate spec you need to follow.  I would pay for that.  Without that you have to know what all the spec standards are, have access to them(some you have to pay for), and go through them individually to figure out what value from what table is applicable.  That's a lot of boring work.

I tried to make a spreadsheet that did exactly that for IEC61010. It did work reasonably well, but then they changed the standard to edition 3 and I haven't figured it out. Bits of what I did are OK (with tweaks) but that are other bits that are totally not understandable - I found I had to work backward through the standard (cross referenced to other parts) to get figures I thought were correct.

Then you run into the manager who say things like "Well look in the lab.  This is only X-mm spacing and it's not arcing right now is it??  It's FINE! Don't worry about it."   

Fortunately, you can counter that if it is sold in the EU there is almost certainly some standard it will have to meet. The company has to produce a "Certificate of Conformity" to say that the product is safe. It has to be signed, usually by the MD. If someone is killed due to it not conforming, then (theoretically) he could go to prison.

Neil