In my application I have an off the shelf 600W 72V AC-DC converter that generates the main DC bus that powers a custom board that houses a 72V input, 9A output buck converter switching at 330kHz. A picture of the AC-DC supply internals are shown below:

My application is a bit unique in that my buck converter doesn't operate at a steady state output voltage. Instead it turns on and outputs a 9A constant current into a capacitor bank charging the capacitor bank from 10V to 65V in about 160usecs after which it turns off, another circuit discharges the capacitor bank back down to 10V and the buck converter turns on again. This process occurs at around 5kHz.

In my current design I have an EMI filter on the board that provides about 30dB of attenuation at my switching frequency. This drops the predicted conducted emissions to about 70dBuV (still above the ~50dBuV limit.) One issue with this design is the with the large value of the inductor combined with the rapid on-off nature of the converter the ripple current in the damping capacitors on the board is very high requiring many large electrolytic capacitors on either side of the filter inductor.

I am looking to see if I can't reduce the side and attenuation of this input filter to spare space on the board to allow for some higher efficiency parts as well as just more space to dissipate heat. I was wondering if anyone had any estimates on how much HF attenuation I can expect from the AC-DC supply and whether I could downsize this input filter without risking failing EMC compliance.

Otherwise any way that I could potentially reduce ripple current in my damping caps would also be useful

Unfortunately getting a full EMC precompliance test setup is not really feasible for me at the moment. (Would run around 3-4k all told between an LISN, spectrum analyzer, and various accessories.)

Any advice would be appreciated.

Edit: below is a rough schematic of my system, with the emi filter detailed

Questions:

• How much attenuation can I expect to get in the AC-DC supply? Do i need to filter the conducted emissions on my board all the way down to 50dBuV or is leaving some overage okay?
• Is there anything I can do to optimize the design of my EMI filter for the rapidly changing load conditions to reduce the bulk capacitance ripple current w/o sacrificing too much attenuation?

In my application I have an off the shelf 600W 72V AC-DC converter that generates the main DC bus that powers a custom board that houses a 72V input, 9A output buck converter switching at 330kHz. A picture of the AC-DC supply internals are shown below:

My application is a bit unique in that my buck converter doesn't operate at a steady state output voltage. Instead it turns on and outputs a 9A constant current into a capacitor bank charging the capacitor bank from 10V to 65V in about 160usecs after which it turns off, another circuit discharges the capacitor bank back down to 10V and the buck converter turns on again. This process occurs at around 5kHz.

In my current design I have an EMI filter on the board that provides about 30dB of attenuation at my switching frequency. This drops the predicted conducted emissions to about 70dBuV (still above the ~50dBuV limit.) One issue with this design is the with the large value of the inductor combined with the rapid on-off nature of the converter the ripple current in the damping capacitors on the board is very high requiring many large electrolytic capacitors on either side of the filter inductor.

I am looking to see if I can't reduce the side and attenuation of this input filter to spare space on the board to allow for some higher efficiency parts as well as just more space to dissipate heat. I was wondering if anyone had any estimates on how much HF attenuation I can expect from the AC-DC supply and whether I could downsize this input filter without risking failing EMC compliance.

Otherwise any way that I could potentially reduce ripple current in my damping caps would also be useful

Unfortunately getting a full EMC precompliance test setup is not really feasible for me at the moment. (Would run around 3-4k all told between an LISN, spectrum analyzer, and various accessories.)

EMI Filter Schematic (VDC is fed by the power supply imaged above): Buck Converter Schematic: Output Waveforms:

Questions:

• How much attenuation can I expect to get in the AC-DC supply? Do i need to filter the conducted emissions on my board all the way down to 50dBuV or is leaving some overage okay?
• Is there anything I can do to optimize the design of my EMI filter for the rapidly changing load conditions to reduce the bulk capacitance ripple current w/o sacrificing too much attenuation?

In my application I have an off the shelf 600W 72V AC-DC converter that generates the main DC bus that powers a custom board that houses a 72V input, 9A output buck converter switching at 330kHz. A picture of the AC-DC supply internals are shown below:

My application is a bit unique in that my buck converter doesn't operate at a steady state output voltage. Instead it turns on and outputs a 9A constant current into a capacitor bank charging the capacitor bank from 10V to 65V in about 160usecs after which it turns off, another circuit discharges the capacitor bank back down to 10V and the buck converter turns on again. This process occurs at around 5kHz.

In my current design I have an EMI filter on the board that provides about 30dB of attenuation at my switching frequency. This drops the predicted conducted emissions to about 70dBuV (still above the ~50dBuV limit.) One issue with this design is the with the large value of the inductor combined with the rapid on-off nature of the converter the ripple current in the damping capacitors on the board is very high requiring many large electrolytic capacitors on either side of the filter inductor.

I am looking to see if I can't reduce the side and attenuation of this input filter to spare space on the board to allow for some higher efficiency parts as well as just more space to dissipate heat. I was wondering if anyone had any estimates on how much HF attenuation I can expect from the AC-DC supply and whether I could downsize this input filter without risking failing EMC compliance.

Otherwise any way that I could potentially reduce ripple current in my damping caps would also be useful

Unfortunately getting a full EMC precompliance test setup is not really feasible for me at the moment. (Would run around 3-4k all told between an LISN, spectrum analyzer, and various accessories.)

Any advice would be appreciated.

In my application I have an off the shelf 600W 72V AC-DC converter that generates the main DC bus that powers a custom board that houses a 72V input, 9A output buck converter switching at 330kHz. A picture of the AC-DC supply internals are shown below:

My application is a bit unique in that my buck converter doesn't operate at a steady state output voltage. Instead it turns on and outputs a 9A constant current into a capacitor bank charging the capacitor bank from 10V to 65V in about 160usecs after which it turns off, another circuit discharges the capacitor bank back down to 10V and the buck converter turns on again. This process occurs at around 5kHz.

In my current design I have an EMI filter on the board that provides about 30dB of attenuation at my switching frequency. This drops the predicted conducted emissions to about 70dBuV (still above the ~50dBuV limit.) One issue with this design is the with the large value of the inductor combined with the rapid on-off nature of the converter the ripple current in the damping capacitors on the board is very high requiring many large electrolytic capacitors on either side of the filter inductor.

I am looking to see if I can't reduce the side and attenuation of this input filter to spare space on the board to allow for some higher efficiency parts as well as just more space to dissipate heat. I was wondering if anyone had any estimates on how much HF attenuation I can expect from the AC-DC supply and whether I could downsize this input filter without risking failing EMC compliance.

Otherwise any way that I could potentially reduce ripple current in my damping caps would also be useful

Unfortunately getting a full EMC precompliance test setup is not really feasible for me at the moment. (Would run around 3-4k all told between an LISN, spectrum analyzer, and various accessories.)

Any advice would be appreciated.

Edit: below is a rough schematic of my system, with the emi filter detailed

Questions:

• How much attenuation can I expect to get in the AC-DC supply? Do i need to filter the conducted emissions on my board all the way down to 50dBuV or is leaving some overage okay?
• Is there anything I can do to optimize the design of my EMI filter for the rapidly changing load conditions to reduce the bulk capacitance ripple current w/o sacrificing too much attenuation?

In my application I have a off the shelf 600W 72V AC-DC converter that generates the main DC bus that power a custom board that houses a 72V input, 9A output buck converter switching at 330kHz. A picture of the AC-DC supply internals are shown below:

My application is a bit unique in that my buck converter doesn't operate at a steady state output voltage. Instead it turns on and outputs a 9A constant current into a cap bank charging the cap bank from 10V to 65V in about 160usecs after which it turns off, another circuit discharges the cap bank back down to 10V and the buck converter turns on again. This process occurs at around 5kHz

In my current design I have designed an EMI filter on the board that provides about 30dB of attenuation at my switching frequency. This drops the predicted conducted emissions to about 70dBuV (still above the ~50dBuV limit) however one issue with this design is the with the large value of the inductor combined with the rapid on-off nature of the converter the ripple current in the damping caps on the board is very high requiring many large electrolytic caps on either side of the filter inductor.

I am looking to see if I can't reduce the side and attenuation of this input filter to spare up space on the board to allow for some higher efficiency parts as well as just more space to dissipate heat. I was wondering if anyone had any estimates on how much HF attenuation I can expect from the AC-DC supply and whether I could downsize this input filter without risking failing EMC compliance.

Otherwise any way that I could potentially reduce ripple current in my damping caps would also be useful

Unfortunately getting a full EMC precompliance test setup is not really feasible for me at the moment. (Would run around 3-4k all told between an LISN, Spectrum Analyzer, and various accessories)

Any advice would be appreciated.

In my application I have an off the shelf 600W 72V AC-DC converter that generates the main DC bus that powers a custom board that houses a 72V input, 9A output buck converter switching at 330kHz. A picture of the AC-DC supply internals are shown below:

My application is a bit unique in that my buck converter doesn't operate at a steady state output voltage. Instead it turns on and outputs a 9A constant current into a capacitor bank charging the capacitor bank from 10V to 65V in about 160usecs after which it turns off, another circuit discharges the capacitor bank back down to 10V and the buck converter turns on again. This process occurs at around 5kHz.

In my current design I have an EMI filter on the board that provides about 30dB of attenuation at my switching frequency. This drops the predicted conducted emissions to about 70dBuV (still above the ~50dBuV limit.) One issue with this design is the with the large value of the inductor combined with the rapid on-off nature of the converter the ripple current in the damping capacitors on the board is very high requiring many large electrolytic capacitors on either side of the filter inductor.

I am looking to see if I can't reduce the side and attenuation of this input filter to spare space on the board to allow for some higher efficiency parts as well as just more space to dissipate heat. I was wondering if anyone had any estimates on how much HF attenuation I can expect from the AC-DC supply and whether I could downsize this input filter without risking failing EMC compliance.

Otherwise any way that I could potentially reduce ripple current in my damping caps would also be useful

Unfortunately getting a full EMC precompliance test setup is not really feasible for me at the moment. (Would run around 3-4k all told between an LISN, spectrum analyzer, and various accessories.)

Any advice would be appreciated.