Is this 4-layer PCB good for mains?

Does it say somewhere in the specification that it can handle 2500 V or just 1500 V between layers ?

Those are the peak voltages that should be expected in mains electricity.

Some requirements might require such test voltages for a minute.

More like 4kV for industrial mains. Unless you've got a surge protector upstream of this.

Typical breakdown (or maybe minimum?) 20kV/mm, so I wouldn't recommend it.

Personally I've seen 650V across a 7 mil internal gap, which survived most of the time (testing of several protos), but one article did fail shorted.

Could also order a 6-layer proto and ignore the mid top/bottom. Unless the middle layers are too close together. Those might not be proto though. Custom boards aren't expensive in any case.

Tim

Just goes to show you need to check each fab's stackup - mine for a 4 layer has 1-2 and 3-4 with fr4 cores and prepreg between layers 2-3.

Personally I wouldn't trust prepreg above 60V for sustained periods.

piglet

er:

I'm unclear how they can have copper on top and bottom over prepreg. Core comes with copper already on both sides, that's why it would normally be co re-prepreg-core rather than prepreg-core-prepreg.

I'm wondering if they actually are showing a correct stackup.

I suppose they could plate the prepreg the same way they plate vias, but wh y bother with that? It has to be easier to just use two layers of core and one prepreg.

urer:

e comes with copper already on both sides, that's why it would normally be core-prepreg-core rather than prepreg-core-prepreg.

why bother with that? It has to be easier to just use two layers of core a nd one prepreg.

1. Copper foils The copper for the outer layers of multilayer boards is applied in the form of copper foil and pressed together with the prepregs / cores. PCBWay have copper foil of 35?m and 70?m in stock. "

cturer:

ore comes with copper already on both sides, that's why it would normally b e core-prepreg-core rather than prepreg-core-prepreg.

t why bother with that? It has to be easier to just use two layers of core and one prepreg.

tml

rm of copper foil and pressed together with the prepregs / cores. PCBWay ha ve copper foil of 35?m and 70?m in stock.

I guess it seems odd to me they would want the extra effort of doing that r ather than just using two cores. I suppose they etch two boards either way . A core and the final stackup or two cores. This particular stack up giv es pairs of layers that are very thinly spaced which is good for power plan e coupling, but not so good for impedance control unless you work with rath er fine trace width as the impedance is controlled by the ratio of trace wi dth to dielectric height. I believe most useful impedances will require ra ther fine traces if you aren't already using them.

My present board is six layer and I had a bit of back and forth with the PC B house to get the inner layer spacing as thin as I wanted as it is not so easy to make the core layers thin. The board is not very large and I wante d it to have as much inter-plane capacitance as possible. To do that with this stackup would require power/gnd on layers 1 and 2 or 3 and 4 since 2/3 layers are 45 mil spaced.

facturer:

se

he

Core comes with copper already on both sides, that's why it would normally be core-prepreg-core rather than prepreg-core-prepreg.

but why bother with that? It has to be easier to just use two layers of co re and one prepreg.

.html

form of copper foil and pressed together with the prepregs / cores. PCBWay have copper foil of 35?m and 70?m in stock.

rather than just using two cores. I suppose they etch two boards either w ay. A core and the final stackup or two cores.

it also mean they can make boards with thicker copper on the outside layer s without extra plating

We have never seen problems in 3 phase designs

IPC-4101 defines tests for PCBs which manufactors needs to comply with.

For FR-4, it is listed as 40kV/0.78mm, which is close to 50kV/mm

OP has 1kV at 0.18mm, which is about 6kV/mm, so many factors below the test limits

Cheers

Klaus

No one needs to comply with IPC. They are a voluntary, private (industry organized) group.

If you need it UL or CE certified, order boards fabricated and labeled accordingly, and have the designs reviewed by whatever certifying body is responsible (UL, ETL, etc. in the US). Pricey (but so are lawsuits).

Tim

This is an interesting one: does any certification body take any responsibility ? This is a question I started thinking about while gathering some signatures for an IECEx project.

All I've seen they state that your item was tested against some standardsor state that your stuff has been found to comply with the standards.

If the shit hits the fan, do they offer any support at that stage ? (Not that I'm expecting).

Yes, I should have worded it differently. In our company we need to comply with IPC and I suspect any not garage firm would do the same ;-)

Cheers

Klaus

AFAIK you can't plate the prepreg, or there'll be nothing to glue the cores together during lamination. Plus the prepreg flows a bit, which would be tough on the plating.

Cheers

Phil Hobbs

IEC 62368-1:2014 has the following:

G.13.5 (page 240)

Insulation between conductors on different surfaces The requirements for insulation on the different layers of a multilayer board are specified below. For basic insulation there is no thickness requirement. Supplementary insulation or reinforced insulation between conductive parts on different surfaces in double-sided single layer printed circuit boards, multi-layer printed boards and metal core printed boards, shall either have a minimum thickness of 0.4mm provided by a single layer or conform with one of the specifications and pass the relevant tests in Table G.14.

Table G.14 Insulation in printed boards

Spec of insulation | Type tests (a) | Routine tests for electric strength (c)

Two layers of sheet insulating | No | Yes material including prepreg (b)

Three or more layers of sheet | No | No insulating material including prepreg (b)

,,,

a Thermal conditioning of G13.6.2 followed by the electrical strength test of 5.4.9.1. b Layers are counted before curing. c Electric strength testing is carried out on the finished printed board.

...

Unless otherwise specified elsewhere in this standard, the test voltage for the electric strength of basic insulation, supplementary insulation or reinforced insulation is the highest value of the following three methods:

...

Table 26 Test voltages for electric strength tests based on transient voltages

Required withstand voltage | Test voltage for basic | Test voltage for kV peak | or supplementary ins | reinforced ins | kV peak or d.c.

.33 | .33 | .5 .5 | .5 | .8 .8 | .8 | 1.5 1.5 | 1.5 | 2.5 2.5 | 2.5 | 4.0 4.0 | 4.0 | 6.0 ... etc ...

Table 27 Test voltages for electric strength tests based on peak working voltages

Peak working voltage up to | Test voltage for basic | Test voltage for kV peak | or supplementary ins | reinforced ins | kV peak or d.c. .33 | .43 | .53 .5 | .65 | .8 .8 | 1.04 | 1.28 1.5 | 1.95 | 2.4 2.5 | 3.25 | 4.0 4.0 | 5.2 | 6.4 ... etc ...

Table 28 Test voltages for electric strength tests based on temporary overvoltages

Nominal mains system voltage | Test voltage for basic | Test voltage for V rms | or supplementary ins | reinforced ins | kV peak or d.c. 250,

Great input, much appreciated!

In this case it is all about reliability. Safety will be implemented by proper creepage, clearance and slots. 9mm or more. I just don't want the board to explode and I am not able to route it on 2 layers.

Best regards, Piotr

There will be a 20kA MOV per phase at the very input, ETFV25K320E4. But I habitually don't trust MOVs.

This is my primary concern. And the manufacturer doesn't want to tell if they are willing to make a 2- or 3-layer thick prepreg stackup. After a lot of googling I am basically where I started -- custom stackup and proto quantities don't come tohether.

According to their stackup specification, the outermost layers in

Best regards, Piotr

Not following what you mean. A 6 layer stack up is two cores and three pre pregs. The 4 layer board is one core with the outer most copper layers sep arated by prepreg. What's wrong with that? If you can use a 4 layer board , why would you even be considering a 6 layer board? What am I missing?

The four layer boards we use are two cores and one prepreg.

Cheers

Phil Hobbs

You are 100% correct here, I didn't pay enough attention to their

The structure of a 4-layer one combined with the unlimited current capabilities of mains scares me off. This is not about routing complexity, but about getting sufficient dieelectric strength. The 6-layer "signal" one would make a decent 3-layer "power" board due to the presence of 2 cores (connect layer 1-2, 3-4 and 5-6). The 4-layer board would suffice, but should preferably have double prepreg layers in order to increase the breakdown voltage and handle prepreg voids.

Best regards, Piotr

prepregs. The 4 layer board is one core with the outer most copper layers separated by prepreg. What's wrong with that? If you can use a 4 layer b oard, why would you even be considering a 6 layer board? What am I missing ?

That's what I'm accustomed to as well, but I still don't understand your co mment "Just give me 2 prepregs and call it a 4-layer board."