"Arrows" pads?

Neato..

I'll guess the PCB arc points is to protect the magnetic wire lamination from arc damage (punched holes). It's a tradeoff.. Arcs fk up the PCB but not the CM xformer.

That would help maintain the creepage rating of the common mode transformer. (Why that would be important, I dunno yet.) However, the creepage on the PCB is allowed to get worse. Once points get a nice carbon spot, arcing can be achieved at lower voltages. But that might be ok.. Who cares about a loss of CM noise filtering during a freakish line spike anyways..

D from BC myrealaddress(at)comic(dot)com British Columbia Canada

Looks like (only) one of the spark point structures may be clear of solder resist, for some reason?

From back in my TV repair days I remember a certain Philips chassis that had a single choke in the mains input circuitry, a common fault was random blowing of the chopper transistor at switch off - the fuse would blow next time it was switched on. The cause was an open circuit damping resistor in parallel with the choke.

Obviously its possible then for a back emf to be generated by a mains input choke, although on a common mode choke I'd expect the spike to also be common mode and probably not particularly dangerous to the PSU.

The deliberate equal spark gaps help ensure the common mode back emfs are clamped at similar peak voltage.

Huh.. I didn't notice that.. Yeah..there's mask on one but not the other. That means one will arc at a higher voltage than the other. This implies L and N are treated differently.

Perhaps spikes (ex: lightning) are generally greater on L (line) then on N (neutral)?? Dunno..

Note the cooling holes for the main xformer and the long creepage notch in the PCB, The PCB designer is doing some smart stuff.

D from BC myrealaddress(at)comic(dot)com British Columbia Canada

Higher energy CM and DM impulses can break a CM inductor in half, though it's more common with E-core or U-core structures. It's not necessarily a freakish level, either, merely one not often experienced in class 2 household wiring.

Transtectors and Sidacs typically employed to provide safe low-impedance paths for this energy are fairly expensive. Controlled natural (free) breakover methods are hard to configure reliably. While printed wiring is reproducible, it's performance here may not be sufficiently predictable. The assumption is that it did work in physical testing, at some point. I'd read the fine print in the spec before accepting that assuption.

RL

D from BC ha scritto:

It seems they are common practises in commerciale off line PSUs.

I looked inside a PSU of another manufacturer and I found the same "arrow" pads and cooling holes.

-- Per rispondermi rimuovi _removethis_ dall'indirizzo

John Devereux ha scritto:

No, both are clear of solder resist. In one case there is a copper trace in the top layer.

-- Per rispondermi rimuovi _removethis_ dall'indirizzo

legg ha scritto:

Ok, but why is it so important to assure such path? What could happen of so dangerous without that path?

-- Per rispondermi rimuovi _removethis_ dall'indirizzo

I think it is better to have a brief spark, than a permanently damaged component (as "D from BC" suggested).

Do you mean a spike could break both the coil and the core?

Transtector? On wikipedia that's a cartoon. Is a transtector like a TVS device?

I dunno how someone figured out a typical spike energy level so the PCB gap doesn't get fried to the point of uselessness.

D from BC myrealaddress(at)comic(dot)com British Columbia Canada

It will attempt to force the coil to 're-orient' itself so as to shorten the magnetic path length of the relevent field. For E-core CM chokes, the core tries to fold in half, joining the distant ends of the wound structure. For toroids, the windings want the core to 'shrink' ~ producing more distributed force.

So's a transformer.

It's typically a 'gas filled tube' or 'gas tube arrestor' with a controlled breakdown voltage that is dV/dT dependant.

If the arc occurs, the energy released locally is minimized. The test waveforms for these surges has a characteristic amplitude, waveshape, duration and characteristic source impedance. The arc occurs across terminals of the circuit that normally do not experience high stress, ie across the coil ( a DC ~short circuit for creepage and cleerance purposes ).

Only testing will demonstrate effectiveness and durability - hence reference to assembly spec sheets. It does no harm, so even failed attempts at design (if it was a serious design effort and not just a whim) may have been left in place, by default.

To be effective, the arc must bypass all components that do not exhibit controlled avalanche, low transient inpedance, or known energy absorption characteristics.

RL

Sounds like what some might call that a 'cheap shot' or 'a better than nothing' philosophy. But that's ok.

D from BC myrealaddress(at)comic(dot)com British Columbia Canada

Very often, opportunities for 'probing the limits' of a design iteration are only available to the tightly-scheduled new product developer in this way. If it works ('free' samples for testing!), it's a new value-added feature that can be advertized and re-used; if it doesn't, there's no skin lost and no budget blown - just a gradual accumulation of savvy for the next attempt (if the developer lasts that long).

RL

...

That may be good electrical isonation, but that "long creepage notch" really reduces the mechanical strength of the board at a bad place, right next to the transformer (probably the heaviest component) in the middle of the board. I think that notch makes the board much more suceptible to breaking if the equipment is dropped.

I think the xformer helps to strengthen the PCB about the notch.

There's 2 large pads near the notch. I think those also belong to the xformer.

D from BC myrealaddress(at)comic(dot)com British Columbia Canada