Electronics > Power & Renewable Energy

Common Power Supply (DC/DC) Topologies

(1/3) > >>


I am interested in your opinion and/or experience regarding the "state-of-the-art" power supply topologies used these days in the power range of about 1W to 20 kW. Although I work now for >10 years on power supplies, there are some topologies, which I think, nowadays are just found in textbooks and not in the industry. E.g., I have never seen a hard-switching half-bridge dc/dc converter (just ZVS LLC) in a project, a hard switching full-bridge (just as ZVS phase-shift or as H-Bridge for DC motor drivers but not in an actual SMPS), and also the single-ended forward converter seems to vanish (although I have seen them in older days 10 - 15 years ago in some designs).

Therefore I am curious about what you think, which topologies are up-to-date. My perception is that the following topologies are most frequently used:

- buck and synchronous buck
- flyback (mostly DCM, quasi-resonant, and active clamp)
- boost converter (mostly for PFC and sometimes in low voltage applications)
- LLC resonant converter (half-bridge and full-bridge)
- ZVS phase-shift full-bridge
- two-transistor flyback and two-transistor forward
- active clamp forward
- SEPIC (low power only <50W mostly battery-operated equipment)

What is your opinion is my perception wrong or am I biased? Which topologies do you think a designer should know these days?

You missed dual active bridges for Isolated bidirectional applications but other than that I agree. I have also heard from more senior folks that for a given "type" of application and power level there's not much choice of topology.

I was surprised to find a single-ended forward converter in a Mean Well SP-500-24, a 500W 24V power supply.


Switching transistors are two 2SK3878 in parallel. If I remember correctly it connects the clamp winding and diode through a capacitor to the power winding/mosfet drain to effectively clamp turn-off voltage spikes.

I measured switching frequency and it was 62 kHz.
PWM controller TL3845P.
PFC controller LT1249.

SP-500-24 is out of production since a few years back, but I wonder if the PSP-600-24 in current production with the same form factor is similar (or if they went to a two-transistor forward maybe?).

Hard-switched half bridges are definitely not as popular as they used to be with that self-starting BJT circuit found in almost all AT and old ATX power supplies. The topology doesn't lend itself easily to current mode control without special arrangements and more or less requires a current transformer to monitor switch currents and not blow up during overload.

And as switching frequencies rise and transformer flux-swing becomes limited by core losses rather than saturation, the bipolar excitation isn't much of an advantage any more either.

While the winding utilization is less in a forward converter because of the non-power-transferring DC current component in the windings, at the same time the DC current isn't subject to skin/proximity effect losses. As DC resistance can be quite a bit lower than AC resistance, the DC current in the winding doesn't cost that much either.

I think these reasons - easier current mode control and current sensing in addition to that unipolar excitation and DC current in the windings don't give much disadvantages at higher switching frequency - are what has led to the widespread adoption of different types of unipolar-drive forward converters (dual switch, active clamp, single switch) and the decline in use of hard-switched bridge converters.

The choice of PSU topology depends on many factors, efficiency, cost, EMI, required input and output voltages, etc.

The trend in recent years is towards higher F and soft switching, resonant and Class E.

The micgation of frequencis seems to be ~ 10X per 1 - 2 decade, eg 1960s 20 kHz, 1970s 100K 1990s 500K 2020 1 M-50M

The recent GaN and SiC devices have also made the higher frequencies proactical.

We have used foward and double foward at ~ 500W..3 kW in 1990s.

Of course the EMI treatment, component choice and ease of semiconductors also affect the design.

Bon Journee


I actually sent a big reply to this post but its disappeared.
Anyway...please find my SMPS course, wherein there is a folder on half bridge.

Half bridge in current mode actually works ok as long as theres enough slope compensation.

...there are controllers like LM5039 which has average current mode in overload, (for half bridge) but it never really caught on, as standard current mode with overload protection and slope comp can get through overload.

(BTW i am wondering how my approx 30 line reply went missing from this thread...is there any way you can tell when you logged in to this site in the past?...as someone may have hacked my password and deleted my post)


[0] Message Index

[#] Next page

There was an error while thanking
Go to full version