UC3843 30W Boost problems

[ali_asadzadeh]

Hi,
I have designed a boost based on the UC3843, it's the first time I'm using this chip, I have separated the power input which is 15V and the aux power supply which is 12, to make things easier for now, every thing works as expected, the oscillator runs, the ref output is stable, and the output would be regulated to 51V,But when I try to get  some load from the circuit, the voltage drops completely! Do you have any Idea?

[strawberry]

How much is it some load
That MOSFET is 3ohm, too high for some load
15V is missing capacitor

[ali_asadzadeh]

How much is it some load

I have tried a 120 ohm resistor

That MOSFET is 3ohm, too high for some load

I know and I have used it deliberately, because it's super cheap and I want it to be able to handle 400V at least.

15V is missing capacitor

How much cap should I use? the input comes from a DC power supply, also I wanted to use a 400V input cap too, so the input cap can not got any higher than 47uF because of it's size!

[SiliconWizard]

I'd take a look at the ISENSE pin while the circuit is running under load.

[spec]

HI ali_asadzadeh,

A few obvious points:

Decouple the 15V supply line with two 100nF X7R disc ceramic capacitors in parallel.

What is the part number of D1? It needs to be a high-speed, low-loss, Schottky type, of at least 2A

What is the part number of L1. It needs to be an HF high-current type, of at least 1A.

What type of capacitor is C1. It needs to be an HF, low ESR type.

Decouple the 50V supply line with two 100nF X7R disc ceramic capacitors in parallel.

Change R9 to 4K7

Change C6 to 4n7 (HF low ESR type)

(the new values for R9/C6 are intended to give 50KHz oscillator frequency)

Remove R8 and its wiring.

https://www.onsemi.com/pub/Collateral/UC3842B-D.PDF

https://www.vishay.com/docs/91059/91059.pdf

[strawberry]

diode could be fast/super fast (reverse recovery)
SMPS cant be developed on breadboard due to weak konnections

[ali_asadzadeh]

What is the part number of D1? It needs to be a high-speed, low-loss, Schottky type, of at least 2A

It's  a SCHOTTKY  diode ,CDBC5100, 5A 100V.

What is the part number of L1. It needs to be an HF high-current type, of at least 1A.

it has no part number, I grab it from an old PC motherboard, it's wire size is around 0.5mm and it's core size is around 14mm

(the new values for R9/C6 are intended to give 50KHz oscillator frequency)

I have placed the pot to adjust the ferequency, now I have set it to about 155kHz

Now I have added another 0.5 Ohm in-series with R5, now the total is about 0.6 ohm, It works better now, now when I connect the load, it would drop from 51V to 42V,

And here is the oscillator waveform,note that the probe is set to 10X in all pictures.



The current sense Pin waveform



Also I have added a 1600uF input cap parallel to the 15V input, the input voltage waveform



The output waveform



And the circuit construction on a vera-board



I have removed the 1600uF cap for now, and the circuit works as before, also the MOSFET gets really hot, when it's under the load! I think it's for it's Rds on of 3Ohm, so lowering it would solve this problem.

when I increase the input voltage to 24V the voltage would not drop and it can fully charge the load, I want it to be able to deliver 51V to the load at least @15V, Also if it's possible I want to increase it's range to 11V input and 60V output, Do you have any Idea about how I can improve it? Also because of the space and size, I can not add any larger than 47uF input cap

[T3sl4co1l]

Besides the Rds(on), the inductance is too high.

That core material is also unsuitable for a peak current mode controller.  Try finding a ferrite cored inductor with a lower value.

Tim

[ali_asadzadeh]

Thanks Tim for the feedback

That core material is also unsuitable for a peak current mode controller.

You mean the core color being yellow? which color should I choose? also for the improved production one, I should choose based on the yellow core, because they are super cheap in here, maybe increasing it's size.

Try finding a ferrite cored inductor with a lower value.

Thanks for the hint How did you calculate it?
I have used excelentit  for The L calculation, I have mentioned it in this thread (the booster app)

https://www.eevblog.com/forum/projects/do-we-have-an-english-version-of-this-very-cool-calculator-(excellentit)/msg2104366/#msg2104366


Also How much in frequency I can go up? since UC38xx can go till 500KHz, is it safe to go to 500KHz? is it do able?

[T3sl4co1l]

If you're stuck on using that material, 50kHz is about the most it's good for.  You still need a low ripple fraction to keep from melting it.

You should use a TL494 or the like, in average current mode control.  A peak current mode controller like 3842 typically operates in DCM (>100% ripple), and just can't operate stable at such a low ripple fraction, even with slope compensation.

So, using 3842: use ferrite core, high ripple fraction, smaller inductance.
Using yellow core: low ripple fraction, lower switching frequency, average current mode control.

Tim

[ali_asadzadeh]

Thanks Tim for the feedback, certainly I want to use smaller Core,because I have limited Space, the yellow core is not ferrite!? I thought It was a ferrite core

Also I want my design to self power it self from the Inductor,by using a secondary winding and I don't know how to calculate the secondary winding? So Maybe I go with an EE16 core on a transformer, is this the appropriate size core or should I go higher, Also I want to increase the frequency to 400-450KHz, is this UC3843 Reliable in this frequency?

There is one more thing though, After reaching my design goals, I want to over-derive the input till  350V, so essentially the boost would be bypassed, so the question is that, would the design survive there? I wanted it to not work at all when the input is higher than 50V

[strawberry]

http://www.comtrafo.com/img/cms/noticias/catalogo_foto3_nuevos_productos.pdf
oscillator dead time vs frequncy , could give some limitations at 400...450kHz

[T3sl4co1l]

So, like an automotive application?  You'll need a lot more engineering to solve that reliably -- handling high input voltages can be done by adding a protection or cutoff circuit, and there are a few options depending on source, load and operating requirements.

If the input is 15V nominal, there's no need for aux power, the 3843 will run on 15V just fine.

Yellow-white is usually the color code for a mix #26 powdered iron, see www.micrometals.com.

Tim

[ali_asadzadeh]

So, like an automotive application?  You'll need a lot more engineering to solve that reliably -- handling high input voltages can be done by adding a protection or cutoff circuit, and there are a few options depending on source, load and operating requirements.

If the input is 15V nominal, there's no need for aux power, the 3843 will run on 15V just fine.

Yellow-white is usually the color code for a mix #26 powdered iron, see www.micrometals.com.

Tim

Really thanks for the help Tim, But it's not an automotive app It's either a brilliant Idea or a poor's man way of making ultra wide input power supply