48V DC Lithium-ion Battery Flyback SMPS to 12V-1A, 5V-0.5A Suggestions

Hello All ;

I have 2 design which have not been applied in real life. Actually, i don't completely know the circuits but for experts in SMPS design please critisize these two design.

Here are in pdf format(2 of them in one pdf document) :

This is progress now that you have something definite to show.

1.

I see that U1 is powered from an auxiliary winding. In order for the transformer to convey any power, M1 needs to be turned on by U1. Can U1 do this before it is powered up?

Instead it may work better if you re-connect R10 so it takes power directly from 48V DC. Adjust component values so nothing overheats.

2.

Sudden load changes at Vo1 may affect output at Vo2 (and vice-versa.) It is because they draw from the same transformer. I suppose the solution is to raise both their output voltage and current.

3.

The TL431 is frequently used as an adjustable zener. Therefore you may be able to omit Z2.

BradtheRad said:

1.

I see that U1 is powered from an auxiliary winding. In order for the transformer to convey any power, M1 needs to be turned on by U1. Can U1 do this before it is powered up?

Click to expand...

Yes,U1 is AP3101 and the other U1 is ice3bs03ljg.They have been designed for this purpose but they are surface mount(SOP) package also not common. I think that i will use uc3842 or Ka3842b are the same. The question is why we need auxilary winding to feed these control IC's ? Can I use sg3525 with no auxilary winding or something like this as used for push-pull smps actually ?

If i use uc3842 can i feed it from 48V lithium-ion with a resistor as you said ?

BradtheRad said:

2.

Sudden load changes at Vo1 may affect output at Vo2 (and vice-versa.) It is because they draw from the same transformer. I suppose the solution is to raise both their output voltage and current.

Click to expand...

Actually, i will feed 12V output to my gate drivers(tlp250) and with 5V to my microcontroller. Therefore, my mosfets are IXTQ170n10P(100V-170A Rds<9mohm) my driving frequency is 20KHz. If required gate current calculated based on gate total charges,each N-mos need 3mA and i have 6 mos so total gate current is very low.Also, maximum current drawn of tlp is 11mA from supply. My question is should my output current rating level is really low like 1A for 12V 0.5A for 5V (microcontroller consumption when i measured 60-70mA) in fact these values maybe higher than expected.

BradtheRad said:

3.

The TL431 is frequently used as an adjustable zener. Therefore you may be able to omit Z2.

Click to expand...

I don't know exactly TL431 my drawings also automating creations so that i don't examine them component by component. I actually don't know how is working principle of adjustable zener. Instead, can i use pc817 optoisolator or if Tl431 is as same as this it is not problem ?

Thanks for your interest BradtheRad

I've worked pretty hard.Eventually, i've designed my circuit using LTspice and LT3748. I am sending to you my schematic also waveforms in pdf format but, i am confused about this IC. That does not require auxilary winding,opto-isolator :shock: even power switch for some of LT series :shock:

How is it possible ? Simulation results said bro you've designed it well but it did not evoke trust.

What is your comments about it BradtheRad Here is the schematic and simulation results :

As far as I can tell, your simulation looks normal. (I can't be sure since I have not used control IC's for switched-coil converters.)

A boost converter generally surges on powerup. This is not serious since you can put simple voltage regulators on your two outputs.

By trimming values, you got 12V and 5V, without regulation. Of course it is theoretical. If you wind your own transformer you will get different values. Your control IC is supposed to compensate.

Try making sudden changes of Rload. See how it affects both output voltages.

Did you find an optocoupler in LTspice? You'll want to do simulations with that by itself, to get familiar with its behavior.

Instead of using the TL431, you can experiment with zener values. The goal is to obtain a certain amount of expanded scale effect from the optocoupler's led brightness.

actor23 said:

If i use uc3842 can i feed it from 48V lithium-ion with a resistor as you said ?

Click to expand...

Typical method using a zener to reduce and regulate a power supply.



The datasheet lists several possibilities in this family of control IC. uc3842 / 3843 / 3844 / 3845

- - - Updated - - -

actor23 said:

If i use uc3842 can i feed it from 48V lithium-ion with a resistor as you said ?

Click to expand...

Typical method using a zener to reduce and regulate a power supply.



The datasheet lists several possibilities in this family of control IC. uc3842 / 3843 / 3844 / 3845

Thanks for all BradtheRad

I got your zener design to feed IC. Due to the fact that i cannot simulate uc3842 in proteus or Ltspice i want to design new circuit using LTxx series which is replacable with uc3842. I don't know what it is but the one told about that in this forum. Simulation results seem correct but in real life i don't think it is work as expecteed,too.

Another problem,LT3748 is surface mount(Sop) IC.Therefore, i will not plug it into breadboard or something and really don't want to deal with compatibility problem.

I'll try and see

I hope i solved my problem using another chip of Linear family namely LT1245. This is like UC3842 but does not require opto isolator indeed. Here is the new design which seems more applicable for real.

But new concern is : in the datasheet of LT1245 there is a warning : "Do not
attempt to use Proto-boards or wire-wrap techniques to
breadboard high speed switching regulator circuits.
They will not work properly." WHY ??

Should i build it onto my breadboard or directly mount in pcb ? Confused

actor23 said:

But new concern is : in the datasheet of LT1245 there is a warning : "Do not
attempt to use Proto-boards or wire-wrap techniques to
breadboard high speed switching regulator circuits.
They will not work properly." WHY ??

Click to expand...

Often we see advice for constructing circuits: 'component leads should be as short as possible.'

The wire-wrap, or the spring contacts of breadboards, create a roundabout route. It is prone to generate and pick up interference.

I think it is saying that high frequencies, and high speed switching, is best done with secure soldered connections.

But, if we can propery cabled all components as short as possible onto breadboard, i think there is no problem with that. I'm hasty for building this setup to see whether it works. If i draw pcb then mount it if there is a problem, i must change maybe the whole circuit.

Let's cut it short, i want to implement this circuit and get the same results as simulated.

actor23 said:

But, if we can propery cabled all components as short as possible onto breadboard, i think there is no problem with that.

Click to expand...

Your design frequency is comparatively low, 20 kHz. It should be feasible.

I've not found anything applicable yet :sad: Can anyone show me design path related to my requirements ? Especially, transformer design process has not been understood yet

As an aid to comprehension, here is my simulation of a self-regulating flyback converter. Its control network is not isolated. It will require more effort to make it practical.

The control network consists of a 555 timer IC, and an op amp (because I like the idea of doing without those special-purpose control IC's).

A 555's duty cycle can be lengthened or shortened by varying the voltage to the control terminal. The op amp provides this by comparing a reference voltage to the load voltage (divided down).

Actually BradtheRad we share same opinion while designing any circuits. I don't want to use special purpose IC,as well.In fact, while designing my motor controller i've used tlp250 optocoupler with bootstrap tecnique :lol: I am strange of designing flyback smps so that i am searching the aided ICs firstly,after understand it very well i can do it by myself.

I liked your last design however first question i want to ask is that how to get +12V to feed 555 also +6V to opamp' non-inverting input ? Secondly, im confused with D2 and nameless diode, 500ohm resistors connected at the secondary side of transformer ?

Thanks for all

actor23 said:

I liked your last design however first question i want to ask is that how to get +12V to feed 555 also +6V to opamp' non-inverting input ?

Click to expand...

The same type of zener arrangement in post #6. That could give you 12V. Then a resistive divider (or potentiometer) to obtain a midpoint voltage.

Secondly, im confused with D2 and nameless diode, 500ohm resistors connected at the secondary side of transformer ?

Thanks for all

Click to expand...

A real flyback is supposed to have only D2.

My schematic is experimental. I included the full diode bridge to examine behavior at various times. (If D1 were conducting it would imply things were not adjusted properly.)

The 500 ohm resistors keep the simulation running smoothly, by reducing convergence hangups.