Hi guys.

In some datasheets i have seen this specification: Non-Repetitive Peak Forward Current. They say the time that this peak can last. What i want to know is how much the device must wait to be able tu bear this peak again, 'cause i don't think the "Non-Repetitive" means for all its life.

So, how can i know the time the device require?

 

I worked in a semiconductor company in my previous life. Back then there wasn't a standard on what the maximum duty cycle on that. However we calculate that NRPF current from the thermal characteristics, the bonding wires and the hotspots of the device amongst other things. The way we derive the Pdmax and the Imax mean that anything lower than 5% duty cycle is considered safe. However, I would suggest for safety reason to go below 1% duty cycle. Bear in mind that if you subject the device to the NRPF current periodically then you have selected an unsuitable device and you should have gone with one that has higher ratings.

 

Thanks, but you are telling me about some 5% duty cycle and 1% duty cycle, however a can't understand. The duty cycle of WHAT. What time?

 

Originally posted by Iodem_Asakura@Dec 29 2005, 01:55 AM
Thanks, but you are telling me about some 5% duty cycle and 1% duty cycle, however a can't understand. The duty cycle of WHAT. What time?

[post=12744]Quoted post[/post]​

Duty cycle is the percentage of the high pulse to the total period (high + low). If you have 1us pulse, then 1% duty cycle needs to wait 99us before the next pulse.

 

Ok, now i understand what are you telling me.

So if the datasheet says the component can bear a NRPFC of 4 Amps for 1uS, you suggest me to wait 99uS for the component to be able to support this Peak again?

Thanks, that was helpful.

 

It's safest to consider the 'none repetitive peak' as something that should only generally occur under rare or fault conditions.

Remeber that power increases with the square of the current.
If your peaks are just ten times the device's continuous maximum rating, then at a 1% duty cycle it will be having to dissipate around the same power as it would at maximum continuous load!

As a rough example:
A hypothetical device has a continuous rating of 10A and a forward voltage drop of 1V, its effective resistance is 0.1 Ohms & it's maximum continuous dissipation 10W, or 10 Joules per second.
Under normal conditions it's only dissipating 1W at 10A.

If you hit it with a surge of 1000A for 1uS, it's dissipation while the current flows is 100,000W (I2R) which will cause massive local heating, even though the overall surge energy is only a tenth of a joule.
It could possibly stand this several times per second.

If you repeated that same spike on a 1% duty cycle (10,000 times per second), which initially may look reasonable (100X current, 100th the time) you would be trying to dissipate 1000 Joules per second or 1KW...

Even if spikes or surges are very rare, they can cause long term damage due to thermal shock, a bit like pouring boiling water on part of a cold glass; one part rapidly expands more than another & things start cracking up.