12V to 40V Boost using an LT1070

[made2hack]

Hi all,
In a previous thread, Driving 3 LEDs in Series @ 1000mA. Is this circuit ok?, I had a circuit with 12V in to power 3 led's in series and have constant current of approximately 1A. (Using an Mosfet + NPN and resistors).

What I want to do now is use the same 12V in, boost it to 40V and drive 14 led's in series (each with Vf of 2.7, or around 37.8V and around 1A, but probably 0.8A). To do this, I first need to boost my voltage from 12V to 40V and I found the LT1070 to use in boost mode. LT1070 Data Sheet

So I went through Linear Technology's Application Note 19 and went through the whole math and configuration and I came up with the following schematic.



C1 = Electrolytic cap 1uF
C2 = PANASONIC - EEUFR1J561S - CAP, ALU ELEC, 560UF, 63V, RAD .029Ohm ESR. Needs to be low ESR and large enough capacitance. My calculations came up with 265uF and .03 Ohm
C3 = Electrolytic cap 100uF - The design Note specified "Required if Input Leads ? 2" " Which I guess mine are? Length of wire going in to circuit?
D1 = FAIRCHILD SEMICONDUCTOR - SB550 - DIODE, SCHOTTKY, 5A, 50V, DO-201. Can I use a 30V / 3A Schottky???
L1 = As per my calculations, I need a 400uH Cap, but I guess I can go higher? Using 400uH means a current ripple of 0.5A. How much is that a problem? I can use larger cap for smaller ripple I guess?
L2 + C4 is an optional filter as opposed to using C2 (as per design note).
R1 -> R3 = 600mW 1% Vishay Thin Film.

So, will this circuit work? Will I get 40V / 1A out to my older circuit? And will that older circuit handle the 40V In to power my LED's @ constant current of 0.8A ?

But, I noticed a problem AFTER ALL THE WORK!!! The LT1070s are 14 - 20 EUROS  . I didn't check the price BEFORE starting to design using this component!!! 

So, all this work to figure things out (5 Pin T0-220) package (still new to me) and for naught!!!  . I then find the ON SEMICONDUCTOR - MC34167TVG , which at first glance seems to be able to do the same thing for only 3 Euros 

Anyway, if anyone has other suggestions for a Boost converter that doesn't have too many pins (so that I can understand the schematics), or any idea how I can boost from 12V to 40V using the On Semi?
Thanks all

[made2hack]

Oh, and I forgot to mention I don't know the first thing about ordering an inductor. If anyone can pitch in. I'm lost at the Current Saturation specs, I don't even know where to start

[mariush]

Boosting 12v to 40v will always be hard... you have to keep in mind that you'll get 40v @ 0.8A  or about 35w at the output with about 80-85% efficiency, so that means at the input you'll have  about 45w / 12v = 3.75A .. let's say 4A. That's a lot, large inductors, large capacitors, the works.

Why not make two strips of 7 leds and use a cheaper led driver for each of those two strips?

For example, get a cheap LM25** (or equivalent chinese clone) based boost regulator from eBay to boost the 12v to 20-24v:

http://www.ebay.com/itm/XL6009-DC-DC-Boost-Buck-Converter-Step-Up-Down-Power-Supply-Regulator-Module-/121291847302?pt=LH_DefaultDomain_0&hash=item1c3d8eb686
http://www.ebay.com/itm/DC-DC-Adjustable-Step-Up-Power-Converter-XL6009-Module-Board-4A-Maximum-Current-/201036396439?pt=LH_DefaultDomain_2&hash=item2eceb3f797
http://www.ebay.com/itm/DC-DC-Adjustable-Step-Up-Power-Converter-XL6009-Module-Board-4A-Maximum-Current-/111276706604?pt=LH_DefaultDomain_0&hash=item19e89bcb2c
http://www.ebay.com/itm/NEW-DC-DC-Adjustable-Step-up-Power-supply-voltage-Converter-Module-Output-5V-35V-/141048598227?pt=LH_DefaultDomain_0&hash=item20d726ced3
http://www.ebay.com/itm/LM2577-Adjustable-Step-up-Power-Converter-Module-DC-DC-/121290311689?pt=LH_DefaultDomain_0&hash=item1c3d774809
http://www.ebay.com/itm/DC-DC-Adjustable-Step-up-boost-Power-Converter-Module-XL6009-Replace-LM2577-/310717070508?pt=LH_DefaultDomain_0&hash=item48582e3cac

then use two buck led drivers to do 19v @ ~ 0.8a to drive 7 leds in series.

For example, something like this:

http://uk.farnell.com/zmdi/zled7020-zi1r/led-driver-buck-1-2a-sot89-5/dp/1898428
http://uk.farnell.com/diodes-inc/ap8803wtg-7/ic-led-driver-buck-1a-5tsot23/dp/1825359RL

Use a Rsense to tweak the output current, and optionally you can do 0-5v pwm into the adjust pin or use a potential divider to adjust brightness by feeding 0.3v-1.2v into the adjust/CTRL pin.

[made2hack]

So splitting the load in two will be a lot better? Just concentrate on boosting 12 - 20 and using an LM25** ?

I've already had a chinese clone 10V - 40V boost unit that has already blown a cap and blown 5 LEDs. That's why I wanted to make my own.

Plus, one day, I hope to go from 220VAC -> to my LEDs, by building the entire circuit. But all in due time.

[made2hack]

A problem I'm also having is that I keep finding TSOP16 packages and other type of ICs with multiple pins.

I'm not sure I can handle understanding 16 pins, it seems complex.

What's the "minimum" amount of pins I should worry about and how many, in these type of situations, seem to be just "Extra" ?

I know it's kind of a stupid question 

[fcb]

LT1070 - great part, an oldie but goody.

Also highly recommend the LT1270 or LT1270A if you need gobs of power.

[Hideki]

TI's webench suggests LM3478 to meet your requirements at the lowest price. It only has 8 pins but some additional components are needed.
If that is still too hard, there are more integrated (and more expensive) solutions like LM2587/LM2588.

[made2hack]

I'm looking @ LMR62014 right now, http://www.ti.com/product/lmr62014. Any thoughts?

[mariush]

Looks like a nice chip, but..  switching current 1.4a ... when boosting to 19-20v 0.8a you may exceed that 1.4a current.. I also doubt it can thermally keep itself cool, you may find out you need a large pcb area to maintain it cool enough, or that it's not really possible to get both 20v out and high current without tripping its thermal protection.

I'm not a fan of boost dc-dc converters modified to work as led drivers, though they can work just fine.  A LM25** can do the "brute force" boosting without caring about the ripple voltage and current and then proper led drivers can control the current and brightness with very high efficiency (up to 95-98%). In addition, the LM** can be bought in TO-220 or DPAK packages which can also be cooled much easier... and those ready made step-up boards are cheap as chips on eBay. As the led drivers control the current there would also be no scenario like you described when such a board damaged 5 leds. 

[made2hack]

@mariush, so what you're saying is that I should keep my 10V / 40v boost unit, and just add a led driver afterwards?

Like the onsemi CAT 4101?

or one of the others you suggested?

[made2hack]

OK, so a new schematic  .

I am keeping my chinese clone boost source, which goes up to 46V (actually around 90% of that so 41.4V).

I will be configuring it to give 40V as supply and use a ZLED7020 driver to power my LEDs.



The ZLED7020 data sheet: ZLED7020-ZI1R

C1 = 1uF MULTICOMP MCR30W105K1HL5L-RH - X7R 50V
C2 = 0.1uF MULTICOMP MCR20W104K1HL5L-RH - X7R 100V
C3 = 22uF Electrolytic cap ESR 0.035ohm PANASONIC 32SEPF22M 32V
D1 = 20V/5A Schottky Diode FAIRCHILD SEMICONDUCTOR - SB520
L1 = 100uH 10% 1.2A MURATA 18104C Axial
Rs = 0.13ohn 1% 1W - MULTICOMP MC14694

C1 & C2 = As per datasheet, X7R with sufficient capacitance for C1 for duty cycle.
C3 = Datasheet recommends low ESR without too high a start-up delay. Specifies to start with a few uF and work your way up (I can't test for ripple or frequency). I don't know, I selected a Panasonic cap that seems to work, but I'm out of my depth.
D1 = Datasheet recommends Total voltage Vlx (including ripple) to be smaller than <50V. So, I chose a 20V diode. But I can put in a 50V diode, same specs from Fairchild. Not sure here either.

L1 = COMPLETELY OUT OF MY DEPTH. They recommend 33uH to 220uH to optimize "switch on/off times". My Ton calculations gave me 14.12uS and my Toff calculations 0.63uS.
Frequency of Lx, or Flx is 67.8kHz and duty cycle, Dlx is 0.957 is that 95.7% duty cycle? 

So yeah, I don't know if my chosen inductor is ok. It doesn't mention anything in terms of Saturation current or peak output current. All it says is DC Amp max 1.2A, so if I use 0.8, 1.2A represents 50% more, which is ok. I just don't know if 96% duty cycle is ok.

If I change from 100uH to 47uH, I get Ton = 6.14uS, Toff = 0.293uS, Freq = 155.45kHz and Duty cycle of 0.9544. Maybe I'm reading too much into this, or maybe it is vitally important, I just don't know. So, if anyone has suggestions or if you guys think the circuit is fine, please let me know!

[made2hack]

It turns out that after my Ton calculations, C1 should be 1.67uF, so I chose to put 2 1uF caps in parallel, maybe I will put a 2.2uF instead, but the 1uF x 2 were cheaper.

[mariush]

I'm not comfortable to say absolutely sure that chinese boost regulator will be able to do 40v 0.8a continuously without crapping itself.  Cheaper LM25** usually go only up to 35v, not sure what your boost converter uses to do 40v+.

The maximum voltage is 40v... well, 50v but really shouldn't be more than 40v.
That makes running all those leds a bit tricky.. 2.7v x 14 = 37.8v which is a bit too close to 40v but it could work.

Capacitor derating... A 1uF 50v ceramic capacitor won't be anywhere close to 1uF at 40v on it. You'll probably have to go with a 6.8-10uF 100v x7r if you want to have 2.2-3uF on the input.

However, note that if your 12v to 40v converter is very close to this led driver, the output capacitance of that converter may be enough and would make this c1 not needed.
 
The output capacitor C3 is not that critical, it's just a minimum capacitance to minimize output ripple.
You could easily go with a 100uF Panasonic FM or FR capacitor rated for at least 50v which basically costs 0.20-0.30 uk pounds: http://uk.farnell.com/panasonic/eeufr1j101/cap-alu-elec-100uf-63v-rad/dp/2079295
Yes, at 0.06-0.08 ohm esr it has higher esr but it's really not that high, but in return you have a 8000h lifetime rating and it's small.
The C3 you chose is only rated for 32v and was much more expensive.

I'm not so sure that you understood correctly what should be used for diode.
It only says :

The flyback diode D1 must have a continuous current rating greater than the maximum output load current and a peak current rating higher than the peak L1 coil current. Important: Use a low-capacitance, fast Schottky diode that has low reverse leakage at the maximum operating temperature and maximum operating voltage for the application to avoid excess power dissipation and optimize performance and efficiency. For silicon diodes, there is a concern that the higher forward voltage and increased overshoot from reverse recovery time could increase the peak LX pin voltage (VLX). The total voltage VLX (including ripple voltage) must not be >50V.

That's all you care about...  your operating voltage is around 40v so the diode should handle a reverse voltage higher than this... I would maybe go for something around 60v.

The inductor choice seems reasonable but I admit I'm no expert here.. for small difference between input and output voltages, there's no need for large inductors.  A 68uH would probably be enough, and they're cheap, but 100uH is probably also good enough.

[made2hack]

@mariush.

Thanks, I'm gonna modify and finally order the parts. Hope everything works when it's connected.

Can I prototype on breadboard? Or is there too much current in the circuit?

[made2hack]

Does anyone know how much current is going through my Rsense resistor? The 0.13 Ohm ?

I chose a 1W rated, 1% Resistor, but I want to know if this is enough and can't seem to figure it out.

[mariush]

Current will be as much as you use on the output.

P = IxIxR =>I max = sqrt(~ 0.9w / 0.13) = sqrt(6.92) =2.6A

[made2hack]

Ok,

Sorry, what I meant to say, is what voltage / power is going through?

It's only a 1W resistor, is that enough?

I couldn't figure out the voltage drop across it.