Input/Output filter of DCDC module is sub-optimal from EMC point of view?

[zenerbjt]

Dear Engineers,

I have been shown a schematic and PCB for a DCDC module power supply which has a vin of a 48V battery (14S LiPO) . Vout is 5V and 5.5W.

I believe the input and output filter is very sub-optimal, and will cause problems with EMC and connector sparking and subsequent degradation.

I attach schematics of the current setup, and the proposed setup. Would you agree that the proposed setup is better from an EMC point of view? (and connector lifetime point of view?)
--------------------------------------------------------------------------------------------------------------------------------------------
The nearest output cap to the DCDC module is some 3.5cm from the module and is a 47uF, 16v, X6S, 1210. You can see switching spikes on the Vout at the output terminals of the DCDC module which comprise 50MHz ringing, up to 5.5V. This appears to be due to the ESL of the 47u and the PCB trace inductance, ringing with the 47uF cap. I believe it needs a 100n,0603 right at the DCDC module output terminals to mitigate this.

Also, there is an 82uF input capacitor, which will mean high inrush from the battery when connected, and high sparking at the connector (it will get connected/disconnected once per day). Thus I proposed an inrush limiter and smaller input capacitor value.

Also, the capacitor at the input to the DCDC module is a alum poly A759KK225M2EAAE685 (2u2, 250V) with 0.685R of ESR. This to me , is more like an RC damper at the DCDC module’s input. From Middleman’s theorem, we know that at the input to a DCDC converter we need a very low impedance, (looking back from the DCDC’s input, to the input filter), so we need some ceramic capacitance there. As such, I am proposing a 100n ceramic and a 1u ceramic in parallel at the DCDC module’s input, as well as a 10uF film capacitor.

I appreciate that the DCDC module works OK as it is now…and successfully supplies the load which is some 5.5W (might go up to 10W in future). However, I believe for EMC reasons, when this product finally goes for approvals testing, I believe the proposed input/output filter is preferable…would you agree?

The DCDC module used is Traco “THN 20-4811WI”.

THN 20-4811WI-A1 | Traco Power

www.tracopower.com

(the DCDC module footprint is “doubled up” as sometimes the Battery will be 24S, and a higher voltage DCDC module will then be placed)

 

[danadakk]

There is an app note here on EMI considerations -

THN 20-4811WI | Traco Power

www.tracopower.com

Regards, Dana.

 

[zenerbjt]

Thanks, yes, and i attach it here......the app note has the nearest input capacitor to the module as an 1812, 2u2, 100V ceramic MLCC cap.....that has far less ESR than the nearest cap to the module input than the PCB i have here. I am certain that the 685 milliohms of ESR in the cap nearest the module is a significant error, would you agree?...after all, SMPS's need to see a low impedance when looking back into their input filter.
Attached is the PCB i have with me now.

 

[danadakk]

I scanned this out of a National Semiconductor 2007 Seminar handbook. It might be useful as
well for layout and other recommendations.

Note if your DC to DC uses any internal LDOs then ESR affects Cap selection, as many require a min
ESR in that cap to create a stable control loop. If you need it I will scan that selection as well, it explains
why this is true, insightful.


Regards, Dana.

 

[danadakk]

Capacitor technology relative ESR performance -

Regards, Dana.

 

[zenerbjt]

Thanks, we currently have the THN 20-4811WI DCDC module set up with a 47u,1210,X6S,16V capacitor 3.5cm away from its output terminals. Then there is a ferrite bead, then another 47u,1210,X6S,16V capacitor.

When we scope the output of the module on AC coupled , there is an 50MHz, 1vpkpk ring on the output. This is obviously due to the combination of the 47uF and its ESL and the PCB trace inductance ringing. We could solve this by adding a 100n right at the module output terminals…but surely there must be such a capacitor inside the module anyway?….i mean, the attached App Note says it passes EN55032 with no external components…so howcome there’s no small value output capacitors inside the module to stop the ringing that we are seeing?

 

[danadakk]

Curious, what is your load exactly ? Types of parts being powered ?

Using good probing practices ? See attached.



Regards, Dana.