Boost Circuit Schematic

lava

Newly Enlightened
Joined
Aug 29, 2011
Messages
30
I would like to build my own boost, something that runs off a single 18650 and can accomodate an LED with 8V Vf and up to 700mA current. I have a BB Nexgen from the Sandwich Shoppe that I'm attempting to reverse engineer, but there are a couple of things I haven't been able to figure out yet. Does anyone have a boost schematic that they would be willing to share that meets my specifications?

Additionally, can someone generally explain to me how one takes a constant voltage boost chip and make it function as a constant current source?
 
Last edited:

hiuintahs

Flashlight Enthusiast
Joined
Sep 12, 2006
Messages
1,840
Location
Utah
.................. can someone generally explain to me how one takes a constant voltage boost chip and make it function as a constant current source?
In short you use a small resistor to translate the LED current to a voltage. Here is an example of one I did years ago when a lot of us were into modifying Maglites.

5LjPua4.jpg


Hope you can read it. I had to downsize it to the CPF 800 pixel max size. You start with a boost controller IC. I think there are a lot of choices. The one I picked was a Linear Technology LT1619 and I did this 10 years ago so there may be better choices. This part did not have an internal switching mosfet so I had to add one (Q1).

The output voltage of the boost switching regulator is controlled by the feedback pin (FB). It will drive the output voltage so as to maintain a set FB voltage. In the case of the LT1619 it's 1.24v. So the output (which is to the LED) will got to whatever it has to, to maintain 1.24v at the FB pin. And it gets to that voltage via the current through R2. Thus the voltage will go to whatever it has to across the LED to maintain a constant 1.24v at the FB pin. And a constant 1.24v at the FB pin is derived via the current through R2. This would be constant current through the LED.

Therefore you can set the current via the size of this resistor. 1.24v/R2 = LED current. However I took it one step further so as to not lose much power across R2. I added an op-amp (U2) to bump up the FB voltage so I could keep R2 very small.

There is also a 12v zener diode with an accompanying transistor that is part of a shutdown of the boost regulator. This in the event that the LED opens up such that I now have an open loop (ie: FB goes to zero) which will cause the boost circuit to ramp up to a high voltage and possibly damage things. So those two components keep the voltage from ever going above 12v.

The rest of the components around the LT1619 are required to make a boost topology along with making the boost regulator stable and some capacitors for filtering.

This circuit would take the voltage of two or three AA batteries and bump it up to 9 volts where I had three LED's in series. What you are trying to do is pretty much exactly the same. The board I made wasn't the smallest thing, but then I had lots of room in a Maglite. Hope this helps.

kRhkKJM.jpg
 
Last edited:

DIWdiver

Flashlight Enthusiast
Joined
Jan 27, 2010
Messages
2,725
Location
Connecticut, USA
Making a voltage regulator into a current regulator may or may not be fairly simple, depending on the design. A chip that uses a resistor divider on the output to generate a feedback voltage may be able to be converted by simply replacing the resistor divider with a current sense resistor.

This often involves a significant hit on efficiency, because you lose the output current times the feedback voltage in the sense resistor. In some cases you can use a lower value sense resistor and an amplifier to generate the feedback voltage with substantially lower power loss. However, the more you mess with the feedback loop, the more challenging it may be to keep it stable.

But why would you want to do that when there are chips specifically designed for what you want to do? I would probably start with the Linear Technology LT1618. Note that LT was recently acquired by Analog Devices. It looks fairly simple to build what you need. If that doesn't work out, googling "LED boost driver" should give you some more options.

Edit: I guess I wasn't the only one responding to this question. Hope my input is helpful too.
 
Last edited:

hiuintahs

Flashlight Enthusiast
Joined
Sep 12, 2006
Messages
1,840
Location
Utah
.........Edit: I guess I wasn't the only one responding to this question. Hope my input is helpful too.
Wow, we said pretty much the same thing! You'll notice the through hole capacitor on my pcb board. There was a little goof with not enough bypass capacitance on the VIN of the LT1619. That was due to the way I layed out the board. If doing it again, I'd lay it out better.
 

lava

Newly Enlightened
Joined
Aug 29, 2011
Messages
30
Thanks guys, very helpful info. The reason I asked about using a voltage boost IC to drive constant current is that the BB Nexgen from the Sandwich Shoppe uses a TPS61030 which appears to be a voltage boost on the datasheet, and yet the BB Nexgen board has very few components on it and strikes me as very elegant.

Looking at the LT1618 datasheet, I see the Li Ion White LED Driver circuit on page 11. It appears to be the closest thing to my application in the document. I guess I can simulate that in LTSpice until I get it working how I want. Any pointers using that device?
 

HarryN

Flashlight Enthusiast
Joined
Jan 22, 2004
Messages
3,976
Location
Pleasanton (Bay Area), CA, USA
It might be easier to just copy one of the open source schematics from the chip suppliers. (edit - I see that this was already suggested)

Linear, TI, etc. all post engineered solutions for LED current drivers with spice simulations on their web sites.

As a possible alternative, you could also take 3 each 123s in series with a very simple buck converter (possibly even just a resistor) and make it work vs designing a boost converter. I have only been involved in 3 LED board driver designs, but in all of those cases a buck converter was a heck of a lot simpler / smaller to implement than a boost.

Supporting 2 amps on the input side is usually the main challenge.

Drivers are cheap so I guess you must be doing this for self satisfaction. Makes sense from that perspective.
 
Last edited:

DIWdiver

Flashlight Enthusiast
Joined
Jan 27, 2010
Messages
2,725
Location
Connecticut, USA
The TPS61030 has a 0.5V feedback voltage, so running it with a current sense resistor replacing the voltage divider feedback would not be outrageous. At 700 mA you'd loose 350 mW in the sense resistor.

Driving an LED with a 3.0V Vf, this would mean the driver efficiency could never reach 85%, probably landing around 80%, maybe 82% if it's really well designed, maybe less if not. It should be possible to do substantially better with modern, purpose-built chips and good design.

I suspect the Sandwich Shoppe design is a bit long in the tooth. I didn't look carefully, but it looks like that driver has been around for quite a few years.
 

lava

Newly Enlightened
Joined
Aug 29, 2011
Messages
30
Drivers are cheap so I guess you must be doing this for self satisfaction. Makes sense from that perspective.

I'd much rather buy a driver, but there are no 17mm (or any other size) drivers that can support a Vf of 8V up to 700mA. Unless you know of one. I'm working with high Vf UV LEDs here.
 

DIWdiver

Flashlight Enthusiast
Joined
Jan 27, 2010
Messages
2,725
Location
Connecticut, USA
I just noticed the absolute max input voltage for the TPS61030 is 3.6V. So that rules it out with a LiIon cell, unless you are talking about LiFePO4.
 

lava

Newly Enlightened
Joined
Aug 29, 2011
Messages
30
I just noticed the absolute max input voltage for the TPS61030 is 3.6V. So that rules it out with a LiIon cell, unless you are talking about LiFePO4.

The BB Nexgen boost, that uses that part, can take an input voltage up to 6V according to the Sandwich Shoppe site.
 

DIWdiver

Flashlight Enthusiast
Joined
Jan 27, 2010
Messages
2,725
Location
Connecticut, USA
Oops, read a little too fast. That's for the LBI pin. All other pins it's 7V. Still, with a max of 7V output, this isn't suitable for the task at hand.

So here's what I was working on when I noticed that:

Running 7V out (max Vf of 6.5V on the LED), the efficiency hit for using 0.5V on the sense resistor is much lower, only 7%. You might reasonably expect to get an overall efficiency in the mid 80s. The LT1618 uses only 0.05V on the sense resistor, but has a schottky diode in there that is almost as bad as the 0.5V sense.

If I were in the mood for experimenting, I would start with the basic schematic on page 1 of the TPS61030 datasheet. Remove all the resistors and tie LBI to gnd. Connect the LED and sense resistor from output to PGND. Put a lowpass RC filter across the sense resistor and connect its output to the FB pin. You end up with a total of 8 components on your driver board.

Given that the device is always operating at maximum load, the RC filter may not be necessary, and you might connect FB directly to the sense resistor, reducing parts count to 6.

For an 8V LED, I think I'd still be looking at the LT1618. The efficiency might actually be lower in this application, but at least it can handle the voltages.
 

lava

Newly Enlightened
Joined
Aug 29, 2011
Messages
30
Oops, read a little too fast. That's for the LBI pin. All other pins it's 7V. Still, with a max of 7V output, this isn't suitable for the task at hand.

So here's what I was working on when I noticed that:

Running 7V out (max Vf of 6.5V on the LED), the efficiency hit for using 0.5V on the sense resistor is much lower, only 7%. You might reasonably expect to get an overall efficiency in the mid 80s. The LT1618 uses only 0.05V on the sense resistor, but has a schottky diode in there that is almost as bad as the 0.5V sense.

If I were in the mood for experimenting, I would start with the basic schematic on page 1 of the TPS61030 datasheet. Remove all the resistors and tie LBI to gnd. Connect the LED and sense resistor from output to PGND. Put a lowpass RC filter across the sense resistor and connect its output to the FB pin. You end up with a total of 8 components on your driver board.

Given that the device is always operating at maximum load, the RC filter may not be necessary, and you might connect FB directly to the sense resistor, reducing parts count to 6.

For an 8V LED, I think I'd still be looking at the LT1618. The efficiency might actually be lower in this application, but at least it can handle the voltages.

Thank you!

And what topology would you use with the LT1618?
 

HarryN

Flashlight Enthusiast
Joined
Jan 22, 2004
Messages
3,976
Location
Pleasanton (Bay Area), CA, USA
I'd much rather buy a driver, but there are no 17mm (or any other size) drivers that can support a Vf of 8V up to 700mA. Unless you know of one. I'm working with high Vf UV LEDs here.

I understand. Sometimes there just aren't any ideal choices. Not an expert but I looked around and didn't see any drivers from the usual suspects that would work - taskled.com, ledsupply.com, etc.

Sometimes that is a useful indicator that the challenges of building something with a nearly 2 amp input boosted to 8 volts output isn't so trivial to accomplish in that size package. (found some that are larger at taskled.com)

Boost drivers can be tricky to manage - and if there is not a load can go into runaway and destroy the output components.

Sometimes there are applications limitations - for instance if the chip company doesn't think that a Vf of 8 volts is very likely, they just won't design for it. The chip designers are typically going for the volume applications, such as USB, computer chip power supplies, and conventional LED drivers.

Unless you go to fairly high frequencies, getting the inductor much smaller than George did on that taskled board will be tough.

If somehow you could come up with a battery configuration so that Vbat > Vf, life will be much easier. A single 18650 isn't always the only possible solution. 3 x R123s or 3 x 18650s ?

It might sound ancient, but for a (3 x R123) + (your UV LED), the efficiency of a well chosen, simple resistor circuit might be (efficiency) competitive with a high tech boost circuit.

Resistor based circuits are also quite robust. From a past project I think I still have some 2 and 10 ohm high wattage resistors built on a high conductivity substrate - think it was AlN but need to check. Let me know if that would be useful. It has been a while, but I think that they are 1206 size but can dissipate a ton of heat.

Looking at your past posts I realize that your electronics background is way past mine but for fun:

Assuming a nominal V bat of 3.8 volts x 3 = around 11.4 volts V bat

Vf = 8 volts

11.4 - 8 = 3.4 volts

RI = V

3.4 volt / .7 amps = 5 ish ohms

So a pair of 10 ohm resistors in parallel will be pretty close if you are willing to be slightly imperfect.

8 volts x 0.7 amps going to the LED
3.4 volts x 0.7 amps going to the resistor

I guess not so ideal but simple.

Harry
 
Last edited:

DIWdiver

Flashlight Enthusiast
Joined
Jan 27, 2010
Messages
2,725
Location
Connecticut, USA
HarryN makes some good comments. If you can go any of those ways, it would be a lot less aggravation and a lot more certainty of success.

Still, I'm having fun here, so I'm going to keep going.

You asked which topology I'd use, and I think the 2-cell Luxeon driver circuit looks the most suitable.

Let's see what the highest peak current in a boost regulator for this design (one LiIon cell, 8V, 700 mA output) would be. The duty cycle (DC) is determined by the input and output voltages, and is worst at low input voltage. So we don't have to go crazy, let's assume cutoff at 3.0V.

Ideally, the output DC is Vi/Vo. In this case that's 3/8 or 0.375. Realistically it will be worse than this, so let's guess around 1/3. Assuming continuous conduction, the average inductor current needs to be 1/DC times the overall average output current, or 1/(1/3) * 0.7, or 3*0.7 = 2.1A.

The inductor current rises during the input phase and falls during the output phase. Let's call this maximum peak-to-peak change dI. The larger our inductance, the lower dI. Conversely, the larger we allow dI to be, the lower the required inducance. Since we like smaller inductors, let's say we will allow dI to be 300 mA. The '1618 runs at 1.4 MHz, giving it a 714nS period. Since our output duty cycle is 1/3, the output period is 1/3 of 714nS, or 238nS.

To calculate the inductance, we use the inductor equation, V = L(dI/dT). In this case, the voltage across the inductor is Vo-Vi, or 5V. We just calculated dI, and dT (the output period). Solving this for L gives L = 5(dT/dI) = 4 uH. The next largest commonly available size is 4.7uH. While none of the datasheet recommendations is rated to meet our current, several from DigiKey seem to be suitable. Note that because of the high frequency, I've ruled out anything but ferrite core materials (nickle-zinc is high-frequency variety of ferrite). Sizes range from 4.8mm square and up. I'm pretty confident that the Abracon ASPIAIG-S6055-4R7 at 6mm square should work well.

You also need a schottky diode. The average current through it will be output DC * Iavg = 1/3 * 2.1 = 0.7A. You'll probably need to select a 1A part.

So between the regulator IC, the inductor, and the schottky, you have several good sized parts to fit on the board. It might still be doable, but it will be cramped.
 

HarryN

Flashlight Enthusiast
Joined
Jan 22, 2004
Messages
3,976
Location
Pleasanton (Bay Area), CA, USA
DIWdiver - thanks for the analysis. That is really helpful.

Perhaps incorrectly, the way I was looking at it is the way that I have built some setups

2 each copper core boards sandwiched back to back

- Board 1 is the board with the LED and optics mounted facing "foreward"
- Board 2 is has the driver and contact to the battery (+)

The rim of both boards act as the ground and thermal path to the flashlight body.

A hard wired connection for the LED contacts that passes through the boards.

In order for that to work and have the inductor small enough, it would be operating more in the 10 - 50 MHz range, so probably would have to use SiC power components to pull it off.

http://www.genesicsemi.com/

https://www.wolfspeed.com/

SiC requires specialized driver chips though, so still not sure if that helps or just makes life more complex.
 

DIWdiver

Flashlight Enthusiast
Joined
Jan 27, 2010
Messages
2,725
Location
Connecticut, USA
HarryN,

I've never seen a copper (or aluminum) core board used for the driver. That offers considerable advantages, but the cost/benefit trade-off generally prohibits it. And while copper is indisputably better, aluminum is far more common for the LED board.

With that in mind, what you described is far and away the most common construction, and if it works it's almost certainly the best way. But if it doesn't work for some reason, it's not the only way. I built one light with part of the driver in the head like that, but the rest was in the tail. I used a piezo switch to turn the thing on and off, so there were no moving parts to leak or jam when diving at 100+ ft. That scheme worked, though the light was not a success for other reasons. This mod I made offers lots more space: http://www.candlepowerforums.com/vb/showthread.php?369404-MJ-852-mod-with-XM-L2, though the construction would need to be different to make optimal use of the space.

I don't know that you have to be that high in frequency, as my analysis was assuming 1.4 MHz, and the inductor is not huge. There are some 4.8mm inductors with similar ratings, but I couldn't be sure the core materials are suitable. OP did mention 17mm diameter for the driver, but didn't say he was married to that. At 20 mm, I'd bet it's doable with the parts I outlined.

With SiC, specialized drivers would be the least of your worries. Once you get away from a highly-integrated monolithic IC, the parts count goes up quite a bit, which is death in really tight spaces. I only found one SiC part that is both available and might take up less board space than a 6mm inductor (the rest are way bigger). It's a 3-pin through-hole part in a metal can (way less than optimum). Oh, and it costs over $60 each.

At this point, I don't believe SiC is a viable option. I am glad you mentioned it though, as previously I wasn't aware SiC was making inroads at voltages below 600V. Maybe better parts will be available soon. Check back every six months or so.
 

Tedy321

Newly Enlightened
Joined
Nov 16, 2020
Messages
1
I would like to build my own boost, something that runs off a single 18650 and can accomodate an LED with 8V Vf and up to 700mA current. I have a BB Nexgen from the Sandwich Shoppe that I'm attempting to reverse engineer, but there are a couple of things I haven't been able to figure out yet. Does anyone have a boost schematic that they would be willing to share that meets my specifications?

Additionally, can someone generally explain to me how one takes a constant voltage boost chip and make it function as a constant current source?

It is difficult for me to say anything about your situation, but I had experience working with schemes, and specialists helped me. I advise you to contact them.
 
Top