Ability of TVS to suppress ringing?

[treez]

Please advise if the SMBJ48A TVS (datasheet below) will be able to quench the attached ringing waveform, which we see on the input voltage rail of our LED driver board, whenever it is “hot-plugged” into the 48V PSU.

So, the waveform you see (attached) occurs when the 48V external power supply is “hot-plugged” to the LED driver PCB…..as you can see, the 48V rail rings up to 71.6V, and this ringing is due to thte 1uH of inductance of the supply cable which connects the 48V PSU to the LED driver PCB.

We must be sure that the input rail does not ring up above 60V. Do you think the SMBJ48A can do this? If it can do this, how many times would you say it can successfully “quench” this transient before it blows up?



SMBJ48A TVS datasheet:
https://www.fairchildsemi.com/ds/SM/SMBJ48A.pdf

 

[betwixt]

It should be suitable.

Don't worry about longevity in a hot-plug situation. Another manufacturer of a similar product pulse them 2000 times to full rating and then re-check their performance hasn't changed as part of their production testing process. You should be good to several thousand hot-plug occurrences without deterioration.

Brian.

 

[treez]

Thanks, the breakover tolerance is from 53.3V to 58.9V,and I am wondering of the 58.9v parts will be quick enough to stop the rail going up to 60V?

We cannot add capacitance to make the voltage rise slower as we are at the output of an off-the-shelf PSU, and as you know, adding capacitance to the output of an off-the-shelf SMPS can make it go unstable.

 

[D.A.(Tony)Stewart]

TVS will not fix your problem. Preloading your SMPS for 5% may reduce overshoot. Improving design Phase margin may improve it more.

I think you have a measurement error as the LED current would be nonlinear and ringing would NOT be sinusoidal at this voltage >60V. ....Reason is the LED string will probably have a lower ESR than the 500W rated TVS @ 10us.

Also there is no long term degradation on TVS like MOV's, as it is a passivated glass semiconductor junction and not an epoxy coated metal oxide (MOV) part, prone to oxidation.

THe 48V TVS has an ESR of 2.7 Ohms which isn't much different than a string of 3W LED's

The knee rating is based on 1 mA is 56V +/-2.9 approx.

The clamp rating for 48V part is 77.4V @7.8A(=600Wp)
with a knee voltage of 56 gives ESR of (77-56)/7.8A = 21/7.8=2.7 Ohm

Send layout , photo of probe and signal.

Why 60V limit? where?

 

[treez]

sorry , the led driver has not actually got up and running at the moment of hot plugging, the transient you see is the effect of inrush into the caps on the led driver board....this inrush current traversed the cable inductance, and thus the ringing was produced between the cable inductance and the input caps on the led driver board.

As you describe though, the clamp rating is 77v.....and so there's no way the part is going to clamp it conveniently at 58.9v, its upper breakover voltage?

 

[betwixt]

As it's DC, perhaps a big 51V Zener might be a solution.

Brian.

 

[treez]

we'd like the tvs to limit overshoot to 60v, as it gives us some margin for using 63v caps, and not going above their voltage.
So you are saying that zeners are quicker acting than TVS's it seems?
Also, do you think the only way we will really limit ringing is by having an active inrush limiter.

Also can high faradic value ceramic caps not tolerate any voltages say 10% above there rating, ever, without being damaged?