TPS92314: FET selection

Hello,

Are you considering an isolated flyback topology for this? At your maximum Vf the output power will be 85V * 0.35A = 29.75W. It will be difficult to design a transformer that fits in the T8 tube.

Also are you trying to replace a T8 florescent bulb and rewire the lamp to remove the ballast?

Picking the MOSFET VDS rating is part of the design process. If your input is 240 VAC that would equate to 373V peak for a 10% high tolerance on the line. A flyback reflects the secondary voltage back to the input on top of the rectified AC, for example if the transformer was 2:1 ratio, at 85V the primary would see an additional 85V * 2, or 170V. Add this to the 373V and you are at 543V. This does not include overshoot due to leakage inductance or margin for AC line surge. The calculations need to be done to find the worst case drain voltage for the MOSFET. This will include the leakage damping or clamping circuit which needs to be above 170V in this example and any surge requirement which would be clamped with some limiting device such as an MOV, TVS or gas tube if your requirement needs this. For this example the MOSFET will need to be at least rated for 800V VDS or possibly higher.

Figuring out what the rating needs to be is all part of the design process, if the transformer ratio was 1:1 the reflected voltage would be lower but the input current peak the MOSFET sees will be higher and at lower duty cycle and probably a less efficient design.

MOSFET VDS depends on Vin maximum peak voltage plus reflected LED voltage through the turn ratio of the transformer plus worst case leakage spike on the drain waveform plus any surge requirement (and how well it is clamped on the input) plus margin.

Regards,

Dear Mr Nederbragt

 

Thank you for your adivce precisely and quickly!!

Sorry to bother you  again, I had missed the info. about topology, but my customer  will use non-isolated. ( NOT isolated).

So, I think following  PMP4362 (slura48.pdf) ; 600V/4A  FET is okay for customer's case.

 

http://www.tij.co.jp/jp/lit/df/slura48/slura48.pdf

 

Would you please give me your comment again?

Best Regards

Hello,

In this topology the maximum voltage at the peak of the rectified AC and 264 VAC (240 VAC + 10%) would be 373V plus the LED stack voltage of 85V = 458V. This is if there is no line disturbance so it should work fine in the lab and during testing but can fail in use due to surge. The MOV used in PMP4362 is ERZV05D471, it's maximum voltage is 810V during at surge at its rating. In this case even a 800V MOSFET isn't high enough. To pick the MOSFET voltage rating needs to take into account what the input surge requirement is.

Regards,

Dear Mr Nederbragt,

Thank you for your answer quickly/ precisely again !!!!

I am talking with customer and I will ask customer tol use 800V or more  .

Please allow me to have customer's qustion about selection of Idrain again .

Q1 Id(MAX)

My LED current is 350mA, but how do we think about Id(MAX) for FET selection?

For example, If Id >> 4~8times ? of LED If, it should be okay?

If so, for axample, we will select 2A or 4A/800V.  

: FCD1300N80Z 800V/4A Coss(eff.):effective output cap is 48.7pF

: FCD3400N80Z 800V/2A Coss(eff.):effective output cap is 41pF

To minimize switching loss, 2A/800V is better?

Q2 Cdrain in the calculation tool sluc446

TPS92314_TPS92314A_Calculation_tools__REV1.1_version: sluc446 shows

following column.**

But is  "Cdrain"  means  Coss in the FET data sheet?( effective output cap?)

***
Cdrain (typical = 22-150p) from MOSFET datasheet

>>> Coss example

Best Regards

Hello,

The MOSFET optimization would come from the design calculations.  The current in the MOSFET is not the current in the LEDs.  In the case of a buck-boost it is easier to understand since the peak current in the MOSFET is also the peak current going to the LEDs and output capacitor.  If it is power factor corrected the output power is about double at the peak of the rectified AC.  If your output is 85V and the output current is 350 mA you are at about 30W output power, at the peak of the AC this is about 60W output power.

This means the average current at the output at the peak is about 700 mA.  The peak of the rectified AC is about 340V, the output 85V.  The duty cycle for the MOSFET at the peak is about 20% so the output is seeing about 80% duty cycle.  Since this is transition mode the peak current has to be twice the average current and factoring in the duty cycle the peak current will be 700 mA * 2 / 80% or 1.75A.  To find the dissipation of the MOSFET conduction loss it would be necessary to figure out the RMS current over 1/2 of a line cycle.  Without doing all of the math I would start with something around one ohm RDSon, maybe more or less but that requires doing all the math, knowing the switching frequency, etc.  Qgate shouldn't be high and I saw parts wtih less the 20 nC that were around one ohm.  The STD4NK60 is a two ohm part so during optimization (the math calculations and testing) that would be discovered.

Coss is non-linear and calculating switching loss is a bit difficult due to the changing frequency and current.  The operating frequency tends to be fairly low on these designs to help reduce EMI and losses.  Picking the output rectifier is a more critical item in the design since the reverse recovery affects switching loss, especially at high temperature (unless using silicon carbide).  It will have to be rated at a high voltage as well.

50 pF at 100 KHz and 400V would cause about 0.8W of loss, it will be less since the input is not always at the peak of the rectified AC.  Having Coss also helps with EMI since it slows the rise and fall time of the MOSFET drain, this is also why a gate resistor is sometimes added though it can add to efficiency loss.

The MOSFET attached looks like it would be a good starting point for a design like this (FCD1300), the other may work as well.  Note that RDS on increases with operating temperature and on high voltage MOSFETs it can double or more (see curve in datasheet).

Regards,