Variable SMPS

[Overclocked]

This is bugging me a bit, Instead of a LM338K Im going to use a variable SMPS IC to replace it, and to give me more efficiency (along with size). The Chip I will be using is a MAX1745 buck converter. Its not the chip thats bugging me, its the design.
**broken link removed**

Iripple is Given 2 ways;

Iripple = (Vin-Vout)*1uS/L
Iripple = Vout*2uS/L

Which do I use? One gives a higher Ripple than the other. Using the 2nd formula I get a ripple of 16A!

Oh, here are my specs and such..

Vin (min)=11.45
Vin (Max)= 16.46

Vout(min)=2V
Vout (Max) =15V

Iout Min =0A
Iout Max = 4A

 

[mneary]

The inductor ripple (peak to peak) is very high for discontinuous mode operation. Ironically this occurs at low output currents. See figure 3, trace C.

Just choose your MOSFET and inductor accordingly.

 

[Overclocked]

The inductor ripple (peak to peak) is very high for discontinuous mode operation. Ironically this occurs at low output currents. See figure 3, trace C.

Just choose your MOSFET and inductor accordingly.

The Min Inductor (according to my calcs) is 1.82uH. According to this table,http://www.interfacebus.com/Inductor_SMD_Values.html such a value exists. Peak Current is 5.2 A.

I think I know what each one is for, The first Iripple (1uS On time) is for DCM Operation, while the 2uS Off Time is for CCM Operation.

Wait, How can there be more than 16A of Ripple current if the Max Peak current is only 5.2A? HA! Didnt see that one there...I just violated the 2nd Law of Thermodynamics...

 

[mneary]

Design guides advise against using the min. inductor.

P-P ripple current is much different from rms or average ripple current.

 

[Overclocked]

Design guides advise against using the min. inductor.

P-P ripple current is much different from rms or average ripple current.

Your right again, I rounded it up to 4.7uH. In the datasheet, it says to use a inductor value 2X to 10X greater than Lmin.

But Would I still have to get the appropriately sized inductor for ripple? Or Do I just need one with a Saturation Current greater than 5.2A?

 

[mneary]

Would I still have to get the appropriately sized inductor for ripple?

Yes, but increasing the inductance decreases the ripple.

 

[indulis]

Yes, but increasing the inductance decreases the ripple.

But there are limits as to how "BIG" you can make the inductor in relation to the switching frequency. Make it to big and you won't be able store enough energy in the inductor during the "on time".

 

[mneary]

Increasing the inductance to 10x is within the vendor's recommendation. The inductance is limited by the transient response needed.

If there's no transient response requirement, then the inductance is unlimited - - except it must always carry the (average+peak) output current. (This gets difficult with large inductances and high current.)

 

[Gordz]

Check out this link. The unit works great as I have built and used one and I think it will fit your bill nicely.
**broken link removed**

 

[Overclocked]

Check out this link. The unit works great as I have built and used one and I think it will fit your bill nicely.
**broken link removed**

Thats looks even Better. Hmm now the question is, Should I go SMD or Through Hole since the IC is available in both? I defiantly know Through-hole parts are easier to come by, and are a bit cheaper. Would I still use Tantalums to Filter the output? Or would I use electrolytic if I chose Through hole?

Size is not a issue either, but there is a Upper Limit.

 

[dknguyen]

I don't think the decoupling capacitor type would change if you chose a different IC Package.

 

[Gordz]

The tantalums have a very low ESR which is why they are specified. You can also get SMD tantalum anyway so either methodology will do. The inductor is fairly large anyway so you may as well go through hole.

 

[Overclocked]

The tantalums have a very low ESR which is why they are specified. You can also get SMD tantalum anyway so either methodology will do. The inductor is fairly large anyway so you may as well go through hole.

How come The Datasheet for Tant's Doesnt Specify How much current they can handle?

Also, According to National Semiconductors WEBBENCH, My "Output high" (15V) is impossible and it suggests another IC...But I Know its Not impossible since My output is below 40V

 

[Overclocked]

Is ESR In Capacitors really that important? I cant seem find any low ESR Capacitors.

ADD: The required Capacitance is 330uF with a ESR of 14.25 mOhms. The Best I can find is 1,000uF @ 30mOhms (2 in parallel would give 15 mOhms, which is close enough..) Can I use these as a substitute for the required value?

 

[Nigel Goodwin]

Is ESR In Capacitors really that important? I cant seem find any low ESR Capacitors.

Don't get too carried away - make sure you use 105 degree ones and not 85 degree ones - but any decent supplier should stock low ESR capacitors anyway, checK RS Components or Farnell.

HIgh ESR is the most common failure mode of capacitors, and the most common fault in domestic electronics.

 

[Hero999]

I don't see anything wrong with two 1000:mu:F capacitors.

You could use four 100:mu:F tantalums if you like.

 

[Overclocked]

Don't get too carried away - make sure you use 105 degree ones and not 85 degree ones - but any decent supplier should stock low ESR capacitors anyway, checK RS Components or Farnell.

HIgh ESR is the most common failure mode of capacitors, and the most common fault in domestic electronics.

Yeap! They are. https://www.mouser.com/search/produ...virtualkey66100000virtualkey661-LXY35VB102MLL

Wont the pot connected to Feedback pick up noise from the Inductor? I have a Toroid Inductor picked out, so should I be worried about noise?

 

[Hero999]

Toroid inductors have a pretty low field leakage so you shouldn't be too worried about it.

 

[Gordz]

According to National Semiconductors WEBBENCH, My "Output high" (15V) is impossible and it suggests another IC...But I Know its Not impossible since My output is below 40V

I used a slightly higher Vin max on my unit, perhaps that is why. WEBBENCH is pretty accurate so I would go with what ever it suggests as a better part.

 

[Overclocked]

I used a slightly higher Vin max on my unit, perhaps that is why. WEBBENCH is pretty accurate so I would go with what ever it suggests as a better part.

Ive figured out why too. The calculation used to calculate L puts the Inductor Off the Graph. For my lowest output Ive calculated a 10uH, but the Datasheet suggests a 33uH Inductor, which I went with. That link to national semi you posted early are pretty much what I chose for values (with the exception of Radj and the feedback resistors)

Also, Since this is being powered from a transformer, and capacitors are a short when they first charge up, WIll I need a Thermistor to limit Inrush current so it wont damage the IC?