400V TVS should be fine for UK/European/Australian mains product?

[ocset]

Hello,
Our contractor tells us that we cannot use a 400V TVS in  our offline (216-265VAC) Power factor corrected LED driver.
He says that we will need to use the 440V TVS of the same family, since he says that the 400V TVS would be damaged by the leakage current that it would experience.
Here is the SMCJ family TVS datasheet:
http://m.littelfuse.com/~/media/electronics/datasheets/tvs_diodes/littelfuse_tvs_diode_smcj_datasheet.pdf.pdf
Our contractor insists that we must use the SMCJ440A instead of the SMCJ400A. Surely the SMCJ400A conducts less than 1mA  of leakage current all the way up to 447V, and so should be fine?
The schematic of  where we use the TVS is as attached.

[T3sl4co1l]

I wouldn't use TVS anywhere near a mains circuit, period...

Tim

[dmills]

Indeed, and especially when there is not exactly much impedance upstream.

Also, 240V (UK Reality) * sqrt (2) * 110% is 373V which I would hold to be more then somewhat tight for a 400V part (At 1mA, that is a third of a watt of dissipation, so the thing would run hot).

regards, Dan.

[Benta]

I wouldn't use TVS anywhere near a mains circuit, period...

Tim

Me neither.

[ciccio]

I think it is not a good idea: you need some impedance upstream for the TVS to work as expected: adsorbing peaks and surges on mains.
There must be another component that can dissipate the power when the TVS is conducting, even for short times, when there are spikes on mains voltage.
If there is no resistive components, there is nothing to dissipate the energy of the spikes, so the TVS will overheat, and fail, even if rated at an higher voltage.
When the TVS will fail, and my experience says it will, the inductor will act as a fuse, I hope....
An option, if you really need a TVS,  could be to replace the fuse with resistor, as in many CF or LED lamps  I've dismantled.
The resistor will dissipate excess energy and will act as a fuse (hope it will).

[Marco]

I've seen some suggestion that the main thing holding back good surge protection is the fact that the existing standardized waveforms for testing them are just complete and utter bullshit. Which seems reasonable, since they pass the tests but as exemplified in this thread fail in practice.

The main problem seems to be that the continuous (pulsed) power is much higher than the standards assume, with most TVS designed for just accumulating the heat from the single test pulse internally ... when it should be effectively heatsinked

[T3sl4co1l]

I've seen some suggestion that the main thing holding back good surge protection is the fact that the existing standardized waveforms for testing them are just complete and utter bullshit. Which seems reasonable, since they pass the tests but as exemplified in this thread fail in practice.

What tests, what practice?

I can't think of any TVS, SMC package or smaller, that wouldn't explode into a million bits when hit by a 2 ohm, 1.5kV, 8/20us surge.  (Probably not literally a million pieces, but it's going to be a short circuit afterwards.)

They meet the specs they meet.  They're rated by peak current, which is a 2 or 20 ohm surge at whatever voltage delivers that current into the part.  Much less voltage in the 2 ohm case than the 20 ohm case!

Tim

[Marco]

Hmm, yes I see what you mean. I guess I was severely overestimating the current carrying capacity of these devices at higher voltages.

[ocset]

I wouldn't use TVS anywhere near a mains circuit, period...

Thanks, i get your actual point, though we do the standard thing of using the TVS downstream of an overcurrent clamp, as in the schematic in the top post, so surely we  should be OK?....the FET of the overcurrent clamp takes most of the associated overvoltage.
We have a littelfuse TMOV module upstream of this all (not shown)