# Adding extenal input and output capacitors to a DCDC SMPS module?

#### zenerbjt

##### Full Member level 2
Dear Engineers,
We are using the UWE-12/10-Q48NB-C DCDC module for our application.

UWE-12/10-Q48NB-C Datasheet:
https://www.murata.com/products/productdata/8807040483358/uwe-100-120.pdf?1583754815000

Spec:
Vin=42-59VDC
Vout=12vdc
Iout = 10A
Isolated (500V)
No fans
Enclosure 50x30x10cm (totally enclosed carbon fibre 1mm thick)

This module has an internal PI input filter and internal output capacitors. However, it says nothing of the values. Our contractor reasons that we can use it with no external input or output capacitors.

However, I believe this is unwise. Specially since we don’t know our exact wiring input impedance in all cases of use. All the datasheet specs (transient response, output ripple, etc etc) were recorded whilst using an external Cin of 33uF and external Cout of 11uF. Therefore, I suggest that we use those values in our application. Would you agree?

As you know, the input filter should have an output impedance (as “seen” from the SMPS ) of value ten times less than the SMPS input impedance at the crossover frequency. The SMPS input impedance at the crossover frequency is approx equal to [Vin]^2/Pin. In out worst case that’s 42^2/134 = 13.2 Ohms.

Its anyone’s guess what the internal PI filter is. So I suggest that we should use an external input filter with impedance of <1.32 Ohms at the crossover frequency. However, the crossover frequency is not stated in the datasheet. The transient response graph suggests that when Cout = 11uF , the crossover frequency is approx. 86kHz. This value was arrived at by A(step) / {2 x pi x [delta V] x Cout}. Where “delta V” is the undershoot after the load transient step of 5A.

However, since the F(sw) is only 200kHz, an 86kHz crossover sounds unrealistic. As such, I believe we should assume crossover of f(sw)/10.

Would you agree with this?

Another quicker way would be to just use the datasheet suggested values for cin and cout of 33uF and 11uF respectively…though the datasheet says that these were simply what they used for the datasheet spec’ing, and that it was not necessary to use these.

#### zenerbjt

##### Full Member level 2
Hi, sometime between feb2020 and june2020 there was a great post on this started by the person "treez". It concerned the DCM design guide by vicor, and "asdf" gave excellent comments on the ZOUT vs F graphs, but i cant find this post now, anybody know where it is?