cupoftea
Advanced Member level 5
We all know of the Phase shift full bridge (PSFB).
We know that its got better EMC performance than eg the Full Bridge etc etc.
The thing is, no-one cares about EMC as long as its cheap and passes the regs.
Many also dont care that much about efficiency.
(BTW, this thread has already been touched on before....)
www.edaboard.com
So Full Bridge can replace PSFB in very many applications, without being that much bigger.
The leakage inductance of a Full Bridge means that you switch on at almost zero current anyway..
..even if in CCM.
OK, PSFB means that you dont discharge the Cds through the FET at turn on....but Cds is not that big anyway.
And with a Full Bridge, a proper turn-off snubber means you dont have that much current going through the FET at turn on anyway,
since there is a relatively high value resistor across the diode of the turn off snubber.
And we all know, the TURN-OFF switching losses in a Full Bridge deffo wont be any bigger than
the TURN-OFF switching losses in a PSFB.
The "actual PSFB" does nothing to reduce turn-off switching losses.
(OK...many are going to call me out on that one and say you can put a big cap across the CDS with a PSFB, (and then turn the fets off super quick) and garantee it
getting naturally discharged just before turn-on....but as discussed, a Full Bridge with a proper turn-off
snubber can compete well with that)
OK, many are going to say that if you are looking for the lowest volume space possible in an SMPS then the PSFB will help
you shave off the cm^3's, more so than with a Full Bridge.
...OK Yes, but most industrial power supplies of a few kW do not have to have "mobile phone smallness"......the factory floor
is likely to be pretty big, and if you have a big old unit standing on the floor somewhere....who cares....as long as its cheap.
And speaking of secondary side diode snubber losses. Again, the PSFB is not better at making this loss smaller.
Now lets talk OffTheShelf controllers.....far far more and cheaper of these available for Full Bridge than PSFB.
Now lets talk potential failure mechanisms. The PSFB, when driven with cheap Si FETs, is well known ( i can post the docs if you want) for
a reverse recovery related failure. This seems more prevalent at start-up, transient load, etc etc.
You can get round this with SiC FETs, or GAN, but they are more pricey and have slightly more complex drive requirements.
But lets talk about the unmentioned advantage of the PSFB....yes....even the marketeers dont bluster about it....the PSFB, even in no load
, switches the FETs with 50% duty cycle....the duty cycle does not change...this is great , because it means you dont get the "sudden duty
cycle change" problem that you can get with the Full Bridge which is driven by a gate drive transformer.
..But even then, you dont have to drive the full bridge FETs with a gate drive transformer....you can give the hi fets a high side
supply, and drive them through a digital isolator. (or a damped gate drive transformer just to send the drive signal) Lets leave bootstrap drivers out of it for now, because they have potential problems
when doing multi-kW converters with 600W input voltages (output of a 3 phase rect).
But nobody does 3kW SMPS from a 600V input voltage (3 phase mains rect input) with a Full Bridge do they?......doesnt everyone use a PSFB or LLC for this?
(LLC may suffer if the Vin may be wider ranging).
...No they dont, having taken apart 3 phase input electro-plating SMPS's with 600Vin and 3 kW power level (vout 24v)....i noticed they were using
a Half Bridge converter. Yes , a very plain Half Bridge with NO OUTPUT INDUCTOR...(just using the leakage inductance in the transformer).
And the UC384x controller. Parallel FETs high and low. The cheap wires going to the transformer from the drive board hadnt even been twisted...and they
were forming a wide-ish loop. All power semis screwed to a centre metal tube, through which a fan blew air. The diodes were those Rectangular things, look
size-ish of a matchbox.
We know that its got better EMC performance than eg the Full Bridge etc etc.
The thing is, no-one cares about EMC as long as its cheap and passes the regs.
Many also dont care that much about efficiency.
(BTW, this thread has already been touched on before....)
Phase Shift Full Bridge SMPS is massively over-hyped?
Hello, The Phase Shift Full Bridge (PSFB) converter is a hoax. –The LTspice simulation attached of a “plain” Full Bridge SMPS bears this out. PSFB claims to be able to reduce switching losses compared to the “plain” Full Bridge SMPS. However the PSFB does have Turn-OFF switching losses, just...
So Full Bridge can replace PSFB in very many applications, without being that much bigger.
The leakage inductance of a Full Bridge means that you switch on at almost zero current anyway..
..even if in CCM.
OK, PSFB means that you dont discharge the Cds through the FET at turn on....but Cds is not that big anyway.
And with a Full Bridge, a proper turn-off snubber means you dont have that much current going through the FET at turn on anyway,
since there is a relatively high value resistor across the diode of the turn off snubber.
And we all know, the TURN-OFF switching losses in a Full Bridge deffo wont be any bigger than
the TURN-OFF switching losses in a PSFB.
The "actual PSFB" does nothing to reduce turn-off switching losses.
(OK...many are going to call me out on that one and say you can put a big cap across the CDS with a PSFB, (and then turn the fets off super quick) and garantee it
getting naturally discharged just before turn-on....but as discussed, a Full Bridge with a proper turn-off
snubber can compete well with that)
OK, many are going to say that if you are looking for the lowest volume space possible in an SMPS then the PSFB will help
you shave off the cm^3's, more so than with a Full Bridge.
...OK Yes, but most industrial power supplies of a few kW do not have to have "mobile phone smallness"......the factory floor
is likely to be pretty big, and if you have a big old unit standing on the floor somewhere....who cares....as long as its cheap.
And speaking of secondary side diode snubber losses. Again, the PSFB is not better at making this loss smaller.
Now lets talk OffTheShelf controllers.....far far more and cheaper of these available for Full Bridge than PSFB.
Now lets talk potential failure mechanisms. The PSFB, when driven with cheap Si FETs, is well known ( i can post the docs if you want) for
a reverse recovery related failure. This seems more prevalent at start-up, transient load, etc etc.
You can get round this with SiC FETs, or GAN, but they are more pricey and have slightly more complex drive requirements.
But lets talk about the unmentioned advantage of the PSFB....yes....even the marketeers dont bluster about it....the PSFB, even in no load
, switches the FETs with 50% duty cycle....the duty cycle does not change...this is great , because it means you dont get the "sudden duty
cycle change" problem that you can get with the Full Bridge which is driven by a gate drive transformer.
..But even then, you dont have to drive the full bridge FETs with a gate drive transformer....you can give the hi fets a high side
supply, and drive them through a digital isolator. (or a damped gate drive transformer just to send the drive signal) Lets leave bootstrap drivers out of it for now, because they have potential problems
when doing multi-kW converters with 600W input voltages (output of a 3 phase rect).
But nobody does 3kW SMPS from a 600V input voltage (3 phase mains rect input) with a Full Bridge do they?......doesnt everyone use a PSFB or LLC for this?
(LLC may suffer if the Vin may be wider ranging).
...No they dont, having taken apart 3 phase input electro-plating SMPS's with 600Vin and 3 kW power level (vout 24v)....i noticed they were using
a Half Bridge converter. Yes , a very plain Half Bridge with NO OUTPUT INDUCTOR...(just using the leakage inductance in the transformer).
And the UC384x controller. Parallel FETs high and low. The cheap wires going to the transformer from the drive board hadnt even been twisted...and they
were forming a wide-ish loop. All power semis screwed to a centre metal tube, through which a fan blew air. The diodes were those Rectangular things, look
size-ish of a matchbox.
Last edited:
