cupoftea
Advanced Member level 6
Hi,
The attached LLC LED driver was once due to be the future of offline LED Drivers.
The old Fluorescent ballasts used PFC stages followed by Soft Switched Half bridge Fluorescent
ballast Drivers using simple bootstrap high side FET drive.
The "equivalent" to that is the attached Half Bridge LLC LED driver.
(LTspice and PNG attached)
But no-one has ever done the attached for offline LED driving, in spite of it being
the obvious "equivalent" to the old fluorescent ballast driver, which incidentally, was
also soft switching.
The reason the LLC LED driver never took off was the danger of the insidious LLC topology.
The LLC converter simply suffers too many failures due to startup, re-start,
return from brownout, etc etc.
Yet never do we see these ills being mentioned in the LLC controller datasheets.
The LLC is tolerated in server farm PSU's where there are zillions of PSU's
so efficiency trumps the high failure rate of the LLC....they just make them hot-swappable so the
many failures can be swapped out quickly and harmlessly. The LLC is also tolerated in HV
PSU's where the leakage of the Txformer can be utilised, and its by far the cheapest way
to get a HV PSU, so here again LLC is tolerated there.
Most other places and the LLC is a nightmare to be avoided. Even the GaN fraternity have stayed well
shy of the LLC....which you would have thought would have been there "prodigal son", since the GaN's
lack of internal diode (well, lack of 'low voltage' internal diode) means that a parallel SiC diode can be put with it and so avoiding the
reverse recovery nightmare of the LLC......the thing is, that the LLC has more bad points than just reverse recovery...it also suffers
terribly from spurious on-coming of the opposite FET unless an expensive gate drive set-up is chosen.
And have you ever used an LLC with a large output capacitor bank?
The soft start runs out of time, and then the LLC goes into
severe overcurrent pulsation, (due to cap bank still too low voltage st startup) and the overcurrent comparator
has to cut in, causing mayhem amongst the resonant L and C.
Several reverse recovery events are often seen during this malarkey...
reducing the lifetime of the LLC FETs.
Very often there is not a tight overcurrent comparator on the LLC bridge current...
(unlike in a current mode or even voltage mode controlled hard switched topology)
this is because they are scared of a low current turn off as it may upset this
wild horse resonant stage. Not to mention...just imagine if there was a sudden
overcurrent turn off in the LLC, and then the 48%/48% gate drive would be suddenly changed
and the ringing in the gate drive transformers primary cap and magnetising current would likely see a spurious turn ON of the
top and bottom fets together...shoot-through and Ka-Boom!
So overcurrents in the LLC are unfortunately allowed to run free for a lot longer than your typical hard switched alternative....and the components
of the LLC are thus stressed and blow up in earlier life.
So we can see exactly why the old fluorescent ballast soft switched drivers never got replaced by their "equivalent" Half Bridge LLC stage when LEDs took over the lighting world!!
Page 56, 57 of the below shows some of the perils of the LLC:
...This shows that a specific processor programmed for LLC is needed, and pages 56 and 57 show the features implemented using the processor, and that these protections are not available with a plain LLC controller such as the ICE2HS01G......i would hazard a guess that these software features are not available in any of the offTheShelf Analog LLC controllers.
People follow the App notes and think they have designed an LLC with op point away from the peak that gives the onset of capacitive region operation...however, when the output is hard shorted, you end up in this capacitive region even if you limited your lower Fsw to be above the peak of the gain curve during max normal loading.....this is why all LLC controllers implement short circuit protection by latching off...there is no re-try or hiccuping......this is not made clear to the readers of those datasheets. I was in a small, relatively non-wealthy company once and a certain pair of Apps Guys came in and tried hard to flog us their LLC silicon...they said nothing of the failure modes of the LLC.....it was only 300W and customer had access to load terminals so may short them....thank goodness that company did not pick LLC...and instead picked a plain old Full Bridge...
See page 8 of ICE2HS01G datasheet to see that short cct protection is by way of "immediate latch off"....no hiccuping into an output short with an LLC.....it'll likely blow up!...
Plain old Full Bridge has far better short circuit withstand than LLC AND can hiccup ON/OFF into a hard short no problem!...The Apps guys didnt tell us this when they came to flog their LLC silicon!
I once worked for a radar company who did LLC converters simply because they had to...due to the high voltage output
that they had to provide.....and I remember the top designer telling me about a particular consignment that
was sent to a military customer...….his words were.....they will fail, usually around a year (on average) into their
service life...….he said regarding their failure, "it's not if, but when".
I also visited a HV PSU company in UK....the MD was telling me that he had come away from the LLC and
was instead preferring the "LC Series Resonant converter".....he said that this was "more bullet-proof" than the
LLC. Obviously the "LC series resonant converter" is a slightly "Tamed" version of the full blooded LLC, which
has a low Lmag value in comparison, and a lot more circulating current.
As we also know, special FETs have to be used with the LLC, as eg the coolMOS CFD series...but even then, their parameters may improve reliability, but their is no talk of them removing the failure mode completely....
So when will we start to see the true nature of the LLC being admitted in the LLC controller datasheets?
The attached LLC LED driver was once due to be the future of offline LED Drivers.
The old Fluorescent ballasts used PFC stages followed by Soft Switched Half bridge Fluorescent
ballast Drivers using simple bootstrap high side FET drive.
The "equivalent" to that is the attached Half Bridge LLC LED driver.
(LTspice and PNG attached)
But no-one has ever done the attached for offline LED driving, in spite of it being
the obvious "equivalent" to the old fluorescent ballast driver, which incidentally, was
also soft switching.
The reason the LLC LED driver never took off was the danger of the insidious LLC topology.
The LLC converter simply suffers too many failures due to startup, re-start,
return from brownout, etc etc.
Yet never do we see these ills being mentioned in the LLC controller datasheets.
The LLC is tolerated in server farm PSU's where there are zillions of PSU's
so efficiency trumps the high failure rate of the LLC....they just make them hot-swappable so the
many failures can be swapped out quickly and harmlessly. The LLC is also tolerated in HV
PSU's where the leakage of the Txformer can be utilised, and its by far the cheapest way
to get a HV PSU, so here again LLC is tolerated there.
Most other places and the LLC is a nightmare to be avoided. Even the GaN fraternity have stayed well
shy of the LLC....which you would have thought would have been there "prodigal son", since the GaN's
lack of internal diode (well, lack of 'low voltage' internal diode) means that a parallel SiC diode can be put with it and so avoiding the
reverse recovery nightmare of the LLC......the thing is, that the LLC has more bad points than just reverse recovery...it also suffers
terribly from spurious on-coming of the opposite FET unless an expensive gate drive set-up is chosen.
And have you ever used an LLC with a large output capacitor bank?
The soft start runs out of time, and then the LLC goes into
severe overcurrent pulsation, (due to cap bank still too low voltage st startup) and the overcurrent comparator
has to cut in, causing mayhem amongst the resonant L and C.
Several reverse recovery events are often seen during this malarkey...
reducing the lifetime of the LLC FETs.
Very often there is not a tight overcurrent comparator on the LLC bridge current...
(unlike in a current mode or even voltage mode controlled hard switched topology)
this is because they are scared of a low current turn off as it may upset this
wild horse resonant stage. Not to mention...just imagine if there was a sudden
overcurrent turn off in the LLC, and then the 48%/48% gate drive would be suddenly changed
and the ringing in the gate drive transformers primary cap and magnetising current would likely see a spurious turn ON of the
top and bottom fets together...shoot-through and Ka-Boom!
So overcurrents in the LLC are unfortunately allowed to run free for a lot longer than your typical hard switched alternative....and the components
of the LLC are thus stressed and blow up in earlier life.
So we can see exactly why the old fluorescent ballast soft switched drivers never got replaced by their "equivalent" Half Bridge LLC stage when LEDs took over the lighting world!!
Page 56, 57 of the below shows some of the perils of the LLC:
...This shows that a specific processor programmed for LLC is needed, and pages 56 and 57 show the features implemented using the processor, and that these protections are not available with a plain LLC controller such as the ICE2HS01G......i would hazard a guess that these software features are not available in any of the offTheShelf Analog LLC controllers.
People follow the App notes and think they have designed an LLC with op point away from the peak that gives the onset of capacitive region operation...however, when the output is hard shorted, you end up in this capacitive region even if you limited your lower Fsw to be above the peak of the gain curve during max normal loading.....this is why all LLC controllers implement short circuit protection by latching off...there is no re-try or hiccuping......this is not made clear to the readers of those datasheets. I was in a small, relatively non-wealthy company once and a certain pair of Apps Guys came in and tried hard to flog us their LLC silicon...they said nothing of the failure modes of the LLC.....it was only 300W and customer had access to load terminals so may short them....thank goodness that company did not pick LLC...and instead picked a plain old Full Bridge...
See page 8 of ICE2HS01G datasheet to see that short cct protection is by way of "immediate latch off"....no hiccuping into an output short with an LLC.....it'll likely blow up!...
Plain old Full Bridge has far better short circuit withstand than LLC AND can hiccup ON/OFF into a hard short no problem!...The Apps guys didnt tell us this when they came to flog their LLC silicon!
I once worked for a radar company who did LLC converters simply because they had to...due to the high voltage output
that they had to provide.....and I remember the top designer telling me about a particular consignment that
was sent to a military customer...….his words were.....they will fail, usually around a year (on average) into their
service life...….he said regarding their failure, "it's not if, but when".
I also visited a HV PSU company in UK....the MD was telling me that he had come away from the LLC and
was instead preferring the "LC Series Resonant converter".....he said that this was "more bullet-proof" than the
LLC. Obviously the "LC series resonant converter" is a slightly "Tamed" version of the full blooded LLC, which
has a low Lmag value in comparison, and a lot more circulating current.
As we also know, special FETs have to be used with the LLC, as eg the coolMOS CFD series...but even then, their parameters may improve reliability, but their is no talk of them removing the failure mode completely....
So when will we start to see the true nature of the LLC being admitted in the LLC controller datasheets?
Attachments
Last edited:
