A new resonant DC-DC converter: Pe17 circuit

A

Atal

Junior Member level 3
Joined
Mar 25, 2018
Messages
28
Helped
0
Reputation
0
Reaction score
3
Trophy points
3
Activity points
161
I hereby present my newly invented resonant DC-DC converter circuit, named 'Pe17 circuit'.





Features:
* Resonant half/full bridge inverter, utilizing ZVS operation.
* Constant frequency switching.
* Wide input/output voltage range operation capability, suitable also for EV battery chargers.
* Higher efficiency and power density.

Brief intro:

For more details, see US patent - US11139734:

Attached:
Presentation providing comparison between the Pe17 circuit and the LLC resonant circuit.
LTSpice simulation files for evaluation.
 

Attachments

  • Comparison of the Pe17 circuit and the LLC resonant circuit for power supply DC-DC conversion.pdf
    3.7 MB · Views: 336
  • simulation.zip
    70.2 KB · Views: 226

Sort by date Sort by votes
D.A.(Tony)Stewart said:
Still room for the improvement with 6% ripple Vpp
Click to expand...
This ripple can be halved by replacing the 2x 1500 µF output capacitors with 4x 2200 µF ones (cost more money and space).
In detailed simulation, set C6=150n, C11=8.2n, R40=6k, R44=11k, R55=1k, Tdelay of V1 to 2.6 ms, and remove D15 and R42.



It goes without saying that 3000 µF is an absurdly low amount of reservoir for a 12 V output at this power level, and 6% load transition ripple still complies with ATX PSU specifications.
 
Last edited:
Upvote 0 Downvote
Regarding the Onsemi appnote for Gan-based 300W LLC DC-DC converter, 1650µF were used for 19 V output while the full-load to no-load transition ripple was 1.2 V (~6%) even with the additional current feedback. Scaling it for a 500 W 12 V output, ~6800µF are required. Of course that a typical 500 W PSU employs at least 10 elcaps at the output, amounting to much more than 6800µF.

 
Last edited:

Upvote 0 Downvote
Take for example the HDPLEX 250W GaN and the PlayStation 5 ADP-400DR, which are not standard ATX power supplies, yet these amass respective 9000µF and 13000µF.



 

Upvote 0 Downvote
How do you convince OEM buyers which supply is better on metrics?

Cost vs Performance vs size vs reliability for high volume benign consumer environments.

cents per Watt? vs % overshoot vs % ripple vs tolerances vs??

From my past experience in the 80's & 90's, Engineering would try to qualify 2 of the suppliers for performance and test all of the 30 pages of specs to and beyond limits then allow changes to comply with all environmental limits and perform 20 unit elevated temp. power cycling MTBF tests. Then if they passed DVT tests to get qualified, purchasing would negotiate for best price and delivery. These were the likes of Hammond, Brown, Lambda, Shindengen, PowerOne and Fujitsu.
 
Last edited:

Upvote 0 Downvote
It's difficult for me to answer, since I've no marketing skills or affinity.
I believe your engineering experience back then was different and more qualitative than what is nowadays.
 

Upvote 0 Downvote
But you must choose a design based on cost vs performance, right?
So how do you compare?

e.g. Cost of passives vs active parts relative to efficiency and %error+noise over range of load?
or cost of Rdson vs performance for SiC FET , IGBT vs Si and
ESR of C's vs ESR*C for gang effects vs Tr or f
 
Last edited:

Upvote 0 Downvote
See the Pareto surface in the graph below - there's an optimal spot to which the design is supposed to converge.



It was taken from a presentation of ETH Zurich university:
 
Upvote 0 Downvote
You must log in or register to reply here.

Similar threads

Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…