High frequency rectifier diode characteristics, forward secondary rectifiers.

[David_]

Hello.

I'm in the process of designing my first offline forward converter and I am looking at secondary rectifier diodes and have some questions related to the characteristics needed.

Like the current rating, I am going under the assumption that the DC output will be maximum 10A but the current going from the secondary winding to the storage capacitor can be larger if I am not mistaken.
In any case diodes has two or three current ratings, maximum average current, maximum non-repetitive peak surge current and but I would think there are also a maximum repetitive peak surge current. Now the names seems to be different from different manufacturers but they should be adequate.

Lets take MBR1080G as a example:
https://www.onsemi.com/pub_link/Collateral/MBR1080-D.PDF
It has:
Average Rectified Forward Current = 10A
Peak Repetitive Forward Current(square wave 20kHz) = 120A

My converter will run at 25kHz, is 10A the value I should consider to be the limit or can this diode manage >10A at 25kHz?
if so what about 50kHz?

This is related to the power dissipation, how do I approximate the power dissipation when using <50% duty cycle?
I know how to calculate heat sink and junction temperature but it feels as the situation would be quite different when PWM signals are involved, or is it?

And finally, the reverse leakage current. What does that actually influence?
Losses seems a obvious influence but by how much and are there any other or is the losses high enough to be of concern?

When looking at diodes I find some that have 4,5µA reverse leakage and some that have 100µA or 250µA all with pretty much the same forward voltage drop, then there are those with >1mA of reverse leakage current.
Do I care if the one I choose has 4,5µA or 100µA reverse leakage current?

By the way, is it normal to only specify recovery speed as <500ns?

Regards

 

[Warpspeed]

As this is a forward converter, there will be an adequate filter choke in the circuit, and the ripple current through that choke effectively controls the rectifier forward current waveform.

The result will be some average dc value of rectifier current which should not contain any huge repetitive current spikes that some other topologies sometimes produce.

It would be safe to assume that if the device is rated to carry ten amps average forward current, it should be well capable of doing that.
This assumes some kind of current limit that will not allow significantly higher average current to flow under abnormal conditions for any length of time.

Reverse leakage current is so small it can readily be ignored in normal power rectifier service.

Some designer may decide to use that diode for an unusual purpose where leakage becomes a very critical parameter.
And so it is always included in the published data, but for most general purpose power rectifier applications is of little direct interest.

 

[FlapJack]

Is this a one switch or a two switch forward converter. How about some specs. input voltage, output voltage and current.

 

[Easy peasy]

A lot of switch-modes have near 50% current pulses in the diode, hence 20A pk = 10A ave

as the ON drop doesn't go up greatly with I forward, it is often done to calc the power dissipated as Vf x Iave.

Almost all switchmode diodes are designed to handle 2 x the Iave (or more), the reverse leakage current is not an issue unless you want to run at T > 120 C,

You will need a diode with Trr < 50nS (ideally < 35nS) unless you want it to overheat with reverse recovery (ans assoc high PIV), watch out for volt spikes on the diode at turn off - these will easily kill a schottky diode, so overrate your diode PIV and use an RC snubber to limit spiking...

 

[David_]

Okey, I see.

As to what package I shall use is not clear jet, I would like to use a SMD typ, TO-277 or D2PACK but in both instances I worry about heat sink capability of the PCB and if I would go for a TO220 then adequate heat sink would be a simple matter.

This is for a offline 2-switch forward converter running from the mains through a PFC-stage that will output 380-400V(not jet done), the secondary will be(hopefully) regulated with the help of magnetic amplifiers to give me multiple fully regulated rails but if that fails then I will ha one regulated... You know this is not clear to me jet since I have been thinking about making the hole thing into a low power winding and a high power winding but in ether case the end result would be ≈10V-55V which then would be feeding a post-regulator composed by a P-Channel MOSFET enhanced Linear Regulator controlled through digital means, and that should deliver 0-50V and 5A at 50V.

I have not jet decided on a solution for the following but I would like to be able to draw 10A at <25V.

Regards

- - - Updated - - -

About that <50ns(or <35ns), I don't think they have any schottky diodes for my purpose, 55V 10A. I'm sure that there are such schottky diodes but not at digikey which is where I will place a order.

 

[treez]

do you need a post regulator David_?

Why not just have an external error amplifier, and change the reference voltage into it?..that will give you varying output voltage.
Yes with a forward you are stuck with D<0.5, and so you will and up likely with too high reverse voltage for a schottky with that 55v output.....I think high current (non sic) schottkys only go up to about 100v, and you'll likely be over that..especially with spiking already spoken of above by Easy Peasy.
So youll probably need the 35-50ns ultra fast diode.

 

[Easy peasy]

STTH20200 diode form ST 200V 20A " schottky" a very good device with low reverse recovery

mag amp control requires a min load on the o/p

- - - Updated - - -

sorry MBR20200CT used to be an ST part....

- - - Updated - - -

SMT e.g. D2pak or D3pak only good for 1.5W and 2W respectively on a pcb with 1 sq inch Cu

 

[treez]

Anyway David_, how about getting into full bridge for your application with LTC3723-1 and LTC3901....full bridge with sync rects......here is a simulation of it, I am nearly there with getting the sync rects to work proper..
https://www.edaboard.com/threads/344951/

 

[David_]

Oh, my bad about not being able to find such a schottky.
There are plenty 100V 10A schottkys, MBRF10H100 for one.

But do I count with Forward voltage drop * average current = power dissipation?
Its here where I am hoping that <50% Duty cycle will ease up on the heating somewhat.
Although even if I could take half of that it will still be to much for a SMD to handle without giving away to much space for copper area for heat dissipation.
But I have seen heat sinks for D2pak if I don't remember wrong.

treez, by post-regulation do you mean the mag amp idea or the linear regulator stage or both?

I never know for sure why I want to do certain things, but although mag amps has the benefit of regulating multiple secondary windings it might mostly be motivated by me thinking its a cool idea that I want to implement, but I have started getting a feeling that mag amps will be to complex and it would certainly increase the circuit size so much that the PCB would become hard to manage. I am designing for a 4-layer PCB since the cost seems to rise quite a bit if I want 6-layers but maybe that is something that I should consider since the layout and noise management would be better and in any case I am more or less just flipping a coin and hope that I get it mostly right from careful study and careful layout.

Its funny that I started out with a LM317/LM337 circuit that then with time came to change into two positive linear regulators isolated into two floating channels with digital control and that evolved into using a buck-converter between the mains transformer and linear regulator, and that evolved into a offline flyback to get rid of both the main transformer and buck, then I changed into a forward and now I am considering changing topology again.

I start to feel that maybe I already have bitten off more than I can chew and that a full-bridge would perhaps simply be to much, also I have not been doing things as I should.
Before jumping to a forward I should have at least prototyped a flyback and at least proven that I could make that work and then the same with forward before going ahead to the next topology but alas its a question of finances and I can't afford all I would like, my finances are really bad right now and also I feel that I really need to do some hardware work soon, it itches in my fingers from all this theoretical work.

However I will take a look at that thread and simulation and think about it, also try and read about pros vs cons of these two topology's.

Regards

- - - Updated - - -

You mentioned over sizing the secondary rectifier voltage rating, by how much is necessary?

- - - Updated - - -

I ask since the voltage drop of the diode seems to rise quite heavily if I want a TO220 package. also price, a 100V 10A diode costs 1USD while a 300 10A cost 1,7 USD and has 1,9V drop instead of 700mV drop.

 

[treez]

by post regulation I was referring to your linear regulator. I think you are right to knock the mag amps idea on the head for now. It may be difficult to get the brick-wall style ferrites that you need for them.

 

[David_]

Yes I have had difficulties locating easily purchasable cores with very square B/H loop. It's a very neat concept though.
I've gotten somewhat lost in my thought for what I need since I started aiming for mag amps, but I need more than one rail(though only one fully regulated, others will be used to feed 5V, 3,3V rails). I have begun pondering the idea of using a converter to produce the high voltage rail and then use a separate much smaller lower current regulator to deliver the low voltage rails.

By the way, in a simulation I checked the forward design and with my values of output choke/cap it showed that 200V diodes will do good but I wonder if there anything I can do to tame those nasty rising edge spikes?
Or is that something that comes with the use of a inductor and can't be helped.

 

[treez]

best to put an RC snubber across the diodes....start with say 220p,100R, but it depends on the leakage inductance etc etc

 

[David_]

I shall explore that, thanks.