Input LC filter inductor is being shorted out

[treez]

Hi,
In the Vicor DCM3623 DCDC module datasheet (below) , page 21, there is, at Figure 23, an LC input filter to some DCM3623 modules in parallel. Each Inductor of this LC filter has a 0.3 Ohm resistor across it.
The Capacitors (“C1_x”) are all ceramic capacitors.

Do you agree that the 0.3 ohm resistor across the filter inductor is basically there to stop LC ringing at the sudden application of Vin?…ie, this 0.3R resistor is there to prevent overvoltage damage to the DCM3623 DCDC module?

I say this because the 0.3R resistor is almost shorting out the inductor, ruining the effectiveness of the input LC filter and potentially jeopardising EMC.

Would you agree that a better way (from an EMC viewpoint), would be to have a much higher value resistor across the inductor, and then to damp out the startup LC ringing by adding an RC damping circuit across the input, just downstream of the LC input filter?

….Or…Alternatively, a comparator “looking” at a divider across the input, which then turns on a FET which switches a low-ish value resistor across the input would be another way to dampen the LC ringing without ruining the EMC performance.? (the FET would only be switched on for the very brief overvoltage ring)


DCM3623 DCDC module datasheet
https://www.vicorpower.com/documents/datasheets/DCM3623x50M53C2yzz_ds.pdf

 

[asdf44]

I suspect the damping is desired for more than start up transients. Simulate their network and see how the bode plot looks.

To damp transient and fault scenarios I’ve clamped inductors with parallel diodes/zeners - a simple form of switched damping.

 

[Easy peasy]

Yup - mainly to stop ringing at 1uH, 20uF 35.5kHz ...

 

[treez]

Thanks, from the Bode plot etc, the inductor is being pretty well shorted out, so i believe we will need to get rid of the 0.3R parallel resistor and implement damping in another way (eg the diode/zeners) which you kindly deccribe.....the Vin is 50v and overvoltage damage occurs at 65v.

The 0.3R damper surpises me, ive damped inductors before with a parallel resistor, but never with so extremely low as 0.3R...the EMC effectiveness is just getting trashed.

 

[Easy peasy]

Actuarrrly, the Zo of that filter is 0.22 ohms - so 0.3 ohms is just a bit higher than critically damped ( i.e. less damped )

 

[FvM]

The filter has still a fair attenuation at the 700 kHz fundamental. If you want stronger high frequency filter effect, you can go for other topologies like L-C||RC, or 2-stage. I guess, you didn't yet calculate, just stumbled upon the 0.3R?

 

[Easy peasy]

To attenuate the switching noise of the converter the C would have to be on the left of the L, as it is the filter is filtering what is coming thru from the left - also isolating the converter from the supplying bus, the 0.3 R is recommended in the data sheet ...

 

[mtwieg]

Overvoltage during transients is likely not the main point of the resistor. Rather its to prevent the filter output impedance from peaking strongly, thus helping loop stability. An undamped LC filter on the input of a SMPS is a recipe for trouble.

 

[treez]

yes i believe i see your point, and in addition, it looks to me that upstream of this heavily damped inductor, there will be a need for a CLC filter, (not shown in the datasheet) the "L" of which will be far less heavily damped, otherwise i believe passing EMC could be tight.

 

[Easy peasy]

The main emc requirement will be lots of damped Y caps to earth ...

if there is no earth then you are reliant on the modules being quiet - as CM chokes would add too much weight if they were to be effective ...

 

[treez]

The main emc requirement will be lots of damped Y caps to earth ...

The main emc requirement will be lots of damped Y caps to earth ...

Thanks, in our case, it’s a tethered drone, so we don’t have an earth connection to it…just “power” and “return”. As such, we have nothing to connect Y caps to.
Having said that, we have an isolated 700VDC power supply on the ground, and are thinking of earthing its output negative….having said that, its not the usual setup for a common mode filter connection, which always relies on an earth connection, and an earthed chassis.

 

[treez]

To damp transient and fault scenarios I’ve clamped inductors with parallel diodes/zeners - a simple form of switched damping.

Thanks, I like this idea, eg e diode, cathode to source, across the input filter inductor….there is some concern over the diode constantly conducting on the switching frequency off-stoke, hence we would like to use a zener of say 10V or so, as you also kindly discuss.

 

[asdf44]

0.7V of steady state ripple would surprise me. Run a simulation to see if it’s possible or not. Could get a dual package and put 2 in series if you’re worried. And/or put an R in series too.

Zener is fine just might not be necessary (and makes more watts when it’s in action).