What are diodes for across LLC split resonant caps?

[cupoftea]

Hi,
On pages 23 and 44 of the below App note on 3.3kw LLC....

https://www.infineon.com/dgdl/Infineon-Evaluationboard_EVAL_3K3W_LLC_HB_CFD7-ApplicationNotes-v01_00-EN.pdf?fileId=5546d4626cb27db2016d3a60583725dc

...There are diodes across the split resoant capacitors. Do you know what these are for?
LTspice sim attached shows them to be not of any use, in fact, deleterious?

 

[FvM]

Purpose of clamping diodes is described in the application note.

 

[cupoftea]

Thanks, the reasons they give for the "clamping diodes" on page 9, i am sure all would agree, is bogus.
I am not knocking Infineon here......just whichever disty put this App Note together was in error.
It says the diodes are there to limit overcurrent......but a current sense transformer needs to be used for that reason.

If you run the attached LTspice simulation, you see why the clamping diodes are very bad.....you loose gain...your vout drops severely....ie the power output goes down...this will tend to push your f(sw) down, getting you closer to the dreaded capacitive region.

Page 9 also says that the clamping diodes prevent operation in the capacitive region.....this is not so....The entire reason d'etre of the App Note is a digital control of an LLC converter, the big advantage of which is that their software algorithm helps you avoid the capacitive region...then they are saying the clamping diodes are needed for the same reason...doesn't make sense.

The best way to avoid hard reverse recovery from capacitive region of operation in an LLC is to use SiC FETs. However, this App Note is advertising the IPW series FETs , which are not SiC, but do have very very low rds(on)...

 

[FvM]

Page 9 also says that the clamping diodes prevent operation in the capacitive region

No, they don't state this. I think you are still on the way to fully understand the application note.

The clamping diode are preventing rising of resonant capacitor voltage beyond the supply rails.

If you run the attached LTspice simulation, you see why the clamping diodes are very bad.....you loose gain...your vout drops severely

In my understanding, they'll never conduct in nomal operation. There's something wong in your design if you reach this operation point.

 

[cupoftea]

No, they don't state this. I think you are still on the way to fully understand the application note.

Thanks, though here is the quote from page 9.....

QUOTE...
***Under resonance and operating with heavy loads the converter risks entering the so-called capacitive mode
where the ZVS of the primary-side MOSFETs is lost and they become hard commutated. This undesired
condition can be prevented in several ways: limiting the minimum switching frequency (45 kHz in this design);
limiting the maximum output current at the higher output voltages; using clamping diodes in the primary side,
which avoids that the primary-side current increases out of control [5].***
UNQUOTE...

In my understanding, they'll never conduct in nomal operation. There's something wong in your design if you reach this operation point.

Thanks, but these clamping diodes are a couple of very cheap 3W, SMC , ES3J's.......if they were of any use....everybody would use them in LLC converters, and they do not.....there are countless articles over the internet of split cap LLC converters, and they dont use clamping diodes with them.

Hard reverse recovery in LLC's is a "point of terror" in the LLC converter...if it could be eradicated by a couple of cheap diodes that dont even need heatsinking, then i believe that everyone would be doing it.

As you know, clamping diodes across C(res) cannot be used in a full bridge LLC.....and nobody has ever said this is a disadvantage of full bridge LLC's

The reasoning behind split cap LLC's is found in many internet articles...none of these docs states that an advantage is that you can use clamping diodes across the C(res).

The clamping diode are preventing rising of resonant capacitor voltage beyond the supply rails.

Thanks, this is prevented by a suitable current sense transformer and comparator shutoff. Also, these caps are usually well over-rated in voltage, as is common for film caps.

The following App note points out the reasons for split cap HB LLC's

https://www.st.com/resource/en/application_note/an2644-an-introduction-to-llc-resonant-halfbridge-converter-stmicroelectronics.pdf

...but gives no mention of the clamping diodes.
...............__________________________________________-
Also, the LLC of the Top post App Note does have a relatively large resonant capacitor....its 528nF.
The L(res) is 10.5uH.
That gives SQRT(L/C) = 4.5.
However, the Load impedance referred to the primary is 9.2 Ohms (8.n^2.Ro/pi^2).
This is slightly unusual, as SQRT(L/C) is usually wanted to be some 1.5x Load impedance for an LLC.
............................___________________________________

 

[Easy peasy]

you can easily put "clamping" diodes (4) each side of the res cap in a full bridge,

the effect is the same as for the half bridge cousin, i.e. limiting volt build up & hence limiting max power transfer - one could put the caps across Lres too .... or both ....

 

[cupoftea]

. limiting volt build up & hence limiting max power transfer

...thanks, though would those things better be served by a current sense transformer and an output overvoltage monitor?
Is it worth designing an LLC with the extra constraint, that at your required max power and voltage, you should not invoke the clamping diodes.
And would you agree that the clamping diodes do nothing to stop hard reverse recovery happening if ever the converter goes into the capacitive region?

--- Updated Apr 25, 2022 ---

It will be interesting if these cheap diodes are of significance in LLC converter though. Must admit ive trawled through a lot of LLC literature on the internet, but never seen anything on diodes across the res caps. With the PSFB you see diodes to the rail in the primary.....but they have the good "raison d'etre" of being there to vastly reduce ringing and overvoltage on the output diodes.

--- Updated Apr 25, 2022 ---

Just imagine...you design an LLC converter...you then find that when you put diodes across the res caps, that the diodes short the caps out near your max load....so you go then have to go back and start all over again?....doesnt sound right?

 

[cupoftea]

I actually suspect that the diodes across the res caps only come into usefulness if you've made the res caps too small Faradic value......ie, the Sqrt(Lres/Cres) ratio is too big.