1kW Flyback converter for battery charging

Status
Not open for further replies.
The currents and voltages are complex numbers (called phasors) and they are analyzed by, you guessed it, by complex analysis. The Laplace transformation is similar to Fourier transformation and is used to convert the time domain into the frequency domain (i(t)---> I(f)) and poles (and zeros) are frequencies where the transformation blows up. Just consider f/(z-a) where the function is having a pole at z=a. Most commonly you not be able to get rid of the poles but you can shift them to a point which is outside your design considerations.
 

You don’t necessarily have to derive the power stage small signal transfer function for a CCM, current mode flyback converter, because it has already been done.
See page 9, fig 8 of the attached.

This shows it for a buckboost converter, so all you have to do to apply it to the flyback is refer the output capacitor and load to the primary side..then you can use the buckboost transfer function.
 

Attachments

  • buk,boost,bukboo, cm, DCM _Pg 8,9.pdf
    595.2 KB · Views: 142

Thanx treez , c_mitra and Warpspeed.

1) Can the 2 switch flyback be used upto 1kW ?

2) for operation of 2 switch flyback in discontinuous the inductance equation:

Lp= ((Np/NS)^2*(1-D)^2*R) / (2*f)

and

Vo = Vi*D*squareroot((efficiency*R)/(2*Lp*f))

with

D_max= (Vo/Vi) * squareroot((2*Lp*f)/(efficiency*R))


are these equations enough ?
 
Last edited:

For dcm flyback , just work it as power out = power in.(OK theres efficiency to think about too)
Then you know that power in = 0.5*L.I^2.freq

Where:-
L = L(pri)
I = Ipeak pri
Freq = switching frequency

The rest of it is mainly use of the trapezoid equations and lenz’s law
Trapezoid equation also ok for triangle of flyback pri as triangle = trapezoid with one side zero
For 1kw you better using full bridge or LLC
 
Reactions: Ravi_H

    Ravi_H

    Points: 2
    Helpful Answer Positive Rating
Two switch flyback will work fine at 1Kw if the dc input voltage is high enough.

The peak current at full 50% "on" duty cycle in the primary will be 4x the dc input current.
From memory I think you said your dc input voltage was around 400 volts ?

Dc input current 400v x 2.5 A (assuming an unrealistic 100% efficiency)
Peak current in the ramp will only be 10A which should present no difficulties.

In practice it will be higher, but still not excessively high, and conduction loss should be absolutely minimal with only two fairly ordinary mosfets.

On the other hand, if you were attempting to do this with only 40 volts dc input, the current peaks would be 100+ amps which would make a DCM flyback impractical at 1 Kw.
 
Reactions: Ravi_H

    Ravi_H

    Points: 2
    Helpful Answer Positive Rating
Thanks Warpspeed and treez

If i increase my switching frequency to 500 kHz to reduce the primary inductance what problems I may face. Is it advisable to increase the PWM frequency above 100 kHz?

Are there problems related to availability of materials for 500 kHz or just an L-C filter at the input can take care of the operation at 500 kHz ?

Please do give ur opinion.
 

increase my switching frequency to 500 kHz to reduce the primary inductance what problems I may face.

Please note that inductive reactance is directly proportional to the frequency; the current in any given cycle will be correspondingly reduced.

To keep power output same, you need to increase the current in the primary and reduce the number of turns that will reduce the primary inductance.

Core losses are measured per cycle; at higher frequency core losses per unit time becomes higher. Winding losses are rather fixed.

First you need to select a proper core. Both core size and material will dictate the final parameters.
 

I would be heading in the other direction down towards 20Khz to 30 Khz.

Its going to need sufficient copper to carry the current, and skin effect will be problematic at 500 Khz. Your switching losses are going to be a lot higher, its just making everything a whole lot more difficult.
 
Reactions: Ravi_H

    Ravi_H

    Points: 2
    Helpful Answer Positive Rating
Yes 500khz gives you a switching period of 2us, so theres not much room in there for you to slow up those switching transitions if you need to, to diminish noise.
 
Reactions: Ravi_H

    Ravi_H

    Points: 2
    Helpful Answer Positive Rating
If you want to see how bad a single switch flyback can be at a power of 100W, then just see
the attached LTspice simulation. Its only a simulation, but there is much truth on display. The RCD clamp resistor dissipates 15W despite the extremely tight transformer coupling of k = 0.999 !

The leakage inductance also puts a very large overvoltage ring on the secondary diode. The off-state voltage of the secondary diode would only be 790Vdc, but the leakage ring gives it a horrendous peak of 1400Vdc. (up to 1500Vdc at the non-soft startup). You can try and snub this down more with the RC snubber, but then you start dissipating too much in the snubber resistor.

Also, notice that the off-state voltage spike of the FET goes up to 840Vdc. (up to 870Vdc at start up due to no soft-start)
 

Attachments

  • Single Switch Flyback 100W.txt
    7.9 KB · Views: 104
Reactions: Ravi_H

    Ravi_H

    Points: 2
    Helpful Answer Positive Rating
All this is exactly why the two switch circuit is far superior.
All that horrendous 15W (1.5%) power loss is returned to the dc input bus so there is no voltage spike on the drains.

And as the flyback transformer will have an appreciable step down ratio (for battery charging) although the secondary peak current will be high, its nothing some appropriate shottky diodes cannot handle.

I do agree though, that generally very high power DCM flybacks are not practical, but in this case a combination of factors make it quite workable.

I have been working on a fairly novel 1.5 Kw 20Khz buck boost circuit, still in prototype form,
180 volt input, +/- 220 volt output rails, only a single mosfet and over 99% efficiency.
The thing runs barely warm flat out.
It works so well because the voltages are high, the currents low, and the drain voltage of the one solitary mosfet is very solidly clamped.
There is no snubbing at all.

Input is straight from solar panels, it provides +/- 220v regulated dc output rails.
 
Last edited:
Reactions: Ravi_H and treez

    T

    Points: 2
    Helpful Answer Positive Rating

    Ravi_H

    Points: 2
    Helpful Answer Positive Rating
1) Please guide me on the flyback design calculation, especially for the transformer. The reference I'm using has been attached. In that from page number 16 the calculations are given.

For 500 kHz switching I'm planning to use RM 14 (N49 material- EPCOS).

Please share proper reference for flyback calculations.

2) Is the design calculations for a Dual switch flyback the same as a the single switch flyback? Cause I'm unable to find seperate calculations for a Dual switch flyback.
 

Attachments

  • Transformer-and-Inductor-Design-Handbook_Chapter_13_Trafo design (1).pdf
    725.1 KB · Views: 136

For dual switch flyback just put the total power divided by 2 through each one.
The equations u need are in the attached.
 

Attachments

  • Core Mathematics and Equations for SMPS design.doc
    194.5 KB · Views: 109

Do I understand correctly you're looking to build a 1 kW flyback, 400V in, 57V out, 500 kHz?
Yikes.

I wouldn't build an LLC at 500 kHz and 1 kW.

Sorry I'm late to the party, but why a flyback instead of an LLC or full bridge?
 

only do flyback if you use many flybacks in parallel, and not doing at 500khz, try lower fsw.
 
Reactions: Ravi_H

    Ravi_H

    Points: 2
    Helpful Answer Positive Rating
The increases losses, heatsink and magnetics size will not make a flyback the lowest cost solution. Especially not a discontinuous one.
Windings in big flyback transformers are very lossy due to the gap. You can't run them at high frequency. Which will make them even bigger.

Forward and bridge topologies store energy in an iron powder choke, which has a lower cost per joule of energy storage, than ferrite (flyback) transformers.

What is your input voltage range and do you need a PFC?
 
Reactions: Ravi_H

    Ravi_H

    Points: 2
    Helpful Answer Positive Rating
Thanx treez, Warpspeed Sir and jcuadra for your help.

Which mosfet driver is used for a two switch flyback?

I'm currently using a high side and low side mosfet drivers separately. The PWM is having noise as the drain voltage is increased, which makes the PWM to deteriorate.

I'm using an op-amp to convert the 3 volt PWM signal to 5 volt, which are input to the mosfet drivers.
 

you could use this for the high side driver (only)
https://www.onsemi.com/pub_link/Collateral/NCP5181-D.PDF
..or better still if you can find a bootstrap high side driver with an UVLO feature (under voltage lockout)

..and then just drive the bottom one through a similar driver with the same delay..any low side fet driver of your choice.
to make 3v into 5v, then maybe its better to use Schmitt trigger type thing.
 

Status
Not open for further replies.

Similar threads

Cookies are required to use this site. You must accept them to continue using the site. Learn more…