Why use super low trr diodes in an LLC converter secondary?

[treez]

Hello,

We have had some of our 3Kw offline battery chargers failing so we opened them up and looked inside. Battery voltage is 250-400VDC

We notice a half bridge LLC converter running after an interleaved PFC stage.

The LLC secondary diodes are ISL9R1560P2. these have 29ns reverse recovery time but a higher forward voltage than many diodes.

ISL9R1560P2 UF diode datasheet:
https://www.fairchildsemi.com/datasheets/IS/ISL9R1560P2.pdf

Why have they not used output diodes with greater reverse recovery time, that have significantly lower forward voltage?
..eg they could have used VS-15ETL06

VS-15ETL06 UF diode datasheet:
**broken link removed**

So, in an LLC converter, which has very minimal reverse recovery of the output diodes, why have they opted for a super fast diode with higher forward voltage, rather than a slightly slower diode, with a much lower forward voltage?

(there are no synchronous FETs used)

 

[D.A.(Tony)Stewart]

It depends on your power budget for conduction loss on diode vs dynamic loss during recovery. Often dynamic losses exceed conduction losses forcing designers with 100kW SMPS to use very expensive HV low Trr parts to get 98% efficiency.

 

[treez]

thanks, though the LLC converter is such that there is very minimal reverse recovery losses on the output diodes. The diodes have virtually no current flowing in them at the point when there off state voltage starts to build up.
An LTspice simulation here of an LLC shows what I mean by this..

 

[treez]

In other words, why have they not used some of the multitude of low Vf fast diodes on the market? It would have meant less FET losses, and would have been cheaper too.

 

[Easy peasy]

Perhaps they didn't have time to experiment before the product had to go on the market, so they chose the faster diodes to be on the safe side, perhaps they did in fact find a real problem with slower diodes.. you have the units, put some slower diodes in and see what happens, you may find it instructive...!

- - - Updated - - -

p.s. you didn't say why the units were failing - I for one would be interested to hear...

 

[treez]

These are reports that have come back from trials, all I have heard is that they stop working, then we get them back here, and they are working. Something is tripping out, temperature, not sure.

 

[Easy peasy]

These are reports that have come back from trials, all I have heard is that they stop working, then we get them back here, and they are working. Something is tripping out, temperature, not sure.

Not really a quality item designed to run at full power then? I guess if it was that easy to design, then every body would be doing it....

 

[treez]

There was something unusual about the gate drive circuit for the half bridge LLC FETs of this battery charger..
The FETs were both driven by their own toroidal gate drive transformer, the gate drive controller being on the secondary side of the power supply.
The gate drive transformers’ primary and secondary were simply coils of triple insulated wire. What was remarkable was that they were not bifilar wound, but instead, the primary and secondary coils were messily and sloppily just wound round the torroid “any old how”. This is in adverse to the suggestion by Laszlo Balogh on page 33, in his paper (below), “Design and application guide for high speed mosfet gate drive circuits”. Here Balogh recommends bifilar winding of the primary and secondary to reduce leakage inductance. This is certainly not done in our battery charger’s FET drives.

Laszlo Balogh gate drive article:
http://www.radio-sensors.se/download/gate-driver2.pdf

I scoped the gate drive waveforms with a diff probe (Pico TA057) into a Tek DPO2024 oscilloscope and they were surprisingly poor. The LLC was on max power and switching at its LrCr resonant frequency. I could see a distinct fairly longish miller plateau on the FET turn off transient. Also the gate drive voltage was bipolar, and was only getting up to around 5 – 6V peak maximum. Pretty poor when you think that the Vgs(th) of the FETs was 3-5V.

The LLC FETs were IXFB100N50…
http://ixapps.ixys.com/DataSheet/DS99496F(IXFB100N50P).pdf


(LTspice sim of the LLC is attached)
Do you know why they didn’t bother bifilar winding the gate drive transformer?