In PCB if the voltage is pulse like 800 V 50% on , when i calculate the clearance in PCB can i consider this voltage is 400v not 800V and also what is the clearance between primary side and secondary side for SMPS and it 480VAC is 6mm is enough and what is the clearance between the input and chassis is 6mm is enough and for 480VAC we do HIPOT at 2100VDC so 6mm will be ok between primary and secondary ?
http://www.smpspowersupply.com/ipc2221pcbclearance.html

 

In PCB if the voltage is pulse like 800 V 50% on , when i calculate the clearance in PCB can i consider this voltage is 400v not 800V

The reference you gave stated that the peak voltage was what mattered. So, what makes you think using average voltage would be okay?

 

The reference you gave stated that the peak voltage was what mattered. So, what makes you think using average voltage would be okay?

if this peak is only 50% on so can we take average, many engineer do this and i need to make sure if this acceptable or not?

 

Based on the requirements given in the IT product safety standard (IEC/EN 60950-1), at the voltages you cite, you require the following minimum creepage/clearance distances.

Basic clearance = 2.6mm
Reinforced clearance = 5.1mm

Basic creepage = 4.8mm
Reinforced creepage = 9.6mm

The above satisfy the requirements for a mains input voltage of less than 300Vac, with a peak circuit voltage of 800V and an rms of 480Vac.

Your electric strength test primary to secondary should be conducted at 3000Vac or 4243Vdc.

A switch mode power supply generating peak voltages of 800V and rms voltages close to 500Vac is likely to be generating significant rf emissions – a well designed SMPS should not be generating peak voltages exceeding 420V.

 

if this peak is only 50% on so can we take average, many engineer do this and i need to make sure if this acceptable or not?

When you're told to use peak voltage, it is not acceptable to use average.

 

... – a well designed SMPS should not be generating peak voltages exceeding 420V.

It depends entirely on the topology. A flyback converter can very easily produce a leakage inductance spike of 800 V (which is why there are FETs rated at a kilovolt). With active power factor correction, a DC input voltage of around 400 is common and a flyback could be producing 600 V or more in the interval between the leakage spike and start of ring-down (which occurs only in discontinuous mode). It would be impossible to make any practical flyback convert where the maximum voltage seen, even with the leakage spike wiped out entirely, would be anywhere near as low as 420 V. It is common to design a flyback inductor turns ratio on the basis of the allowable "reflected" secondary voltage (when the flyback inductor actually is behaving like a transformer, though transfer of energy between windings is absolutely not wanted).

Single-switch forward converters and push-pull (rarely used at high voltage) will produce voltages of double the supply rail.

===

Don't forget that the creepage and clearance requirements also apply to the transformer. This can be difficult to achieve and cost a lot of window area for winding. Triple insulated wire can be a boon, but it is expensive.

===
I certainly never used "average" voltage. The average of AC mains voltage, wherever you are and whatever its magnitude, is zero.

===

Almost all switchers generate high RF. even if the voltage swings are not huge. It is a matter of "containing" the energy.

 

f this peak is only 50% on so can we take average, many engineer do this

No.
No knowledgeable engineer would use average voltage value when determining required high voltage clearance distance or dielectric strength required.
Why would you say they do?
Do you really think if you have a 1000V peak pulse at a 1% duty-cycle, you only need to allow for a 10V maximum voltage?
Peak voltage is always used for those calculations.