Current mode Half Bridge converter is fine

[cupoftea]

Hi,
There is a rule that Current Mode Half Bridge (HB) converters dont work due to rail splitting capacitor imbalance.
This is true where "pure" current mode is concerned (ie with no added slope on the current sense input), and even the LTspice simulator shows this.
However, as long as a Current Mode Half Bridge converter has some added slope then its fine.
AYK, obviously sub-harmonic oscillation would cause rail capacitor unbalance.

The attached is a current mode Half Bridge (LTspice and jpeg) with a 0.7 Duty cycle, and added slope compensation to keep it from sub-harmonic oscillation. It is even hacked so that its peak current is significantly different for every other switching cycle, and yet the rail capacitor unbalance produced is insignificant. (!!!)

Most Half Bridge converters are going to have added slope anyway due to the magnetising current.

As such, it is believed that Current Mode Half Bridges (with added slope) are actually fine after all.

The things that make sub-harmonic oscillation worse also seem to increase rail capacitor unbalance....eg if the inductor di is high, then it tends toward greater capacitor unbalance.

Is it the vendors of LLC chips that are cranking up the slating of the Half Bridge?.....because there seems to be no "official" text claiming that Half Bridge in Slope-added-Current-Mode is bad.

Neither Basso, Morey , Billings, Cuk, Musk, Mammano or Ridley say anything of it.

I believe that the problem of capacitor unbalancing with Half Bridge is no worse a threat than that of Severe reverse recovery in the LLC....in fact, HB is safer because you can overcome cap unbalance (mitigate damage) by just rating the caps for the full rail voltage.
(by LLC i mean LLC without series diode and without added software to prevent rev rec ever happening)

I believe LLC (with proviso) is worse than HB in respect of possible danger....what we are seeing is a "hit" on the plain HB by LLC chip vendors.

Pull out your current mode HB design spreadsheet and enjoy that half voltage rail input, and the lesser dv/dt that results....and the two FETs that are easy to bootstrap drive.

 

[Easy peasy]

hahaahhaaaa, try building that ....

 

[cupoftea]

Thanks, we know current mode doesnt work with half bridge, though the lack of symettry in the bridge currents can happen in voltage mode too. For example, the fets etc may have different tolerances etc. Admittedly its not as bad as in current mode, but rail cap imbalance is possible.

Quite a lot of the isolated offline half bridges you see have no output inductor, have no feedback either....just a pri CST to do overcurrent....and a FET switching network involving so many components that you doubt that the on times are "exactly" the same in either direction...(theyre just stuck at @45% duty cycle permanently)..yet they have tiny failure rates, and no rail cap imbalance seen.

 

[cupoftea]

Maybe they (open loop half bridges) are made in places where they have tons of staff to do production testing, and those units which fail soak test due to rail cap imbalance, get scrapped/recycled, the rest get sold?

 

[cupoftea]

In this thread, it was discussed that current mode Half Bridge is possible without cap imbalance, as long as enough slope compensation is used.....

https://www.edaboard.com/threads/half-bridge-smps-in-current-mode-but-with-slope-compensation-is-ok.388021/

..Half Bridge is precisely in that power level where the West would not even research into it...so i guess we'll never find out.....since this sector of SMPS has been almost entirely outsourced to the Far East.

--- Updated Sep 7, 2022 ---

This article describes how it is possible to get f(xover) above the LC resonance frequency in a Voltage Mode Buck

https://www.renesas.com/eu/en/document/oth/tb417-designing-stable-compensation-networks-single-phase-voltage-mode-buck-regulators?language=en

...As such, the Half Bridge is a transformer derived Buck so there will be similarities.