OPEN SOURCE: "CC-Boost" - 2.4 Amp boost driver - RC1

[rhd]

OSH stencils is fast. I think they cut them in the USA.

Okay, now that we have RC1 behind us, and there's a working, clean output, inexpensive 2.4A boost. It's worth discussing objectives for the future.

I would propose as a worthy next goal, the ability to reliably output a clean 4A at 5V, from a single lithium ion, without substantially increasing the size of the driver, and keeping the part costs under $10.

That's not an easy target.

 

[IWIRE]

That's not an easy target.

You seem to enjoy a challenge. If it would be possible to bump it 10%, to 4.4a, then it would be able to power the NDB7A75 to about it's max. That's the highest current draw "common" diode that's widely used and available at the present. Beggars can't be choosey though. I'd be thankful for anything close.

 

[rhd]

You seem to enjoy a challenge. If it would be possible to bump it 10%, to 4.4a, then it would be able to power the NDB7A75 to about it's max. That's the highest current draw "common" diode that's widely used and available at the present. Beggars can't be choosey though. I'd be thankful for anything close.

Working backwards:
- 4.4 Amp, 5V output
- Assume we want the driver to run on an input down to 3.2V.
- Assume a roughly 200 mV drop across the sense resistor.
- Assume a saturation current of at least 1.4 x the input current.

((4.4 * (5 + 0.2)) / 3.2) * 1.4 = 10.01 Amps​

We'd need a indcutor with a saturation current of 10A+.

This won't be an easy task. I'm not sure if it's doable within the constraints above (ie, keeping size under 13x13, and part cost under $10). A couple things are clear at this stage:
- To keep size in check, with that saturation current, we'll have to accept lower inductance.
- To keep ripple in check, with the lower inductance, we'll need a higher frequency IC.
- To handle this much current, we're going to need an IC that uses an external MOSFET.
- If the above is even possible, the input cell would need to be capable of very high current output (think Panasonics if 18650, or LiPos, etc)

 

[ARG]

OSH stencils is fast. I think they cut them in the USA.

Okay, now that we have RC1 behind us, and there's a working, clean output, inexpensive 2.4A boost. It's worth discussing objectives for the future.

I would propose as a worthy next goal, the ability to reliably output a clean 4A at 5V, from a single lithium ion, without substantially increasing the size of the driver, and keeping the part costs under $10.

That's not an easy target.

Why not just make a PCB that is two of these in parallel?

I don't see a purpose of having a tiny driver for such a high current as any diode being driven at 4A will produce a lot of heat which will need a large heatsink, a large host and presumably have more room for the driver.

 

[crazyspaz]

Why not just make a PCB that is two of these in parallel?

I don't see a purpose of having a tiny driver for such a high current as any diode being driven at 4A will produce a lot of heat which will need a large heatsink, a large host and presumably have more room for the driver.

Theres a thought. How much bigger would it make the whole thing if you has essentially one driver per side? Surely not that much bigger, and even if it IS, like ARG said- any diode running at that current is bound to need a large host-heatsink. If you think of the SXB-2 boost drive by Lazeerer that can do 5A, its 26mm in diameter- not small by any means. I say if you can get it to that size or smaller, you could count it as a success.

 

[rhd]

Why not just make a PCB that is two of these in parallel?

I don't see a purpose of having a tiny driver for such a high current as any diode being driven at 4A will produce a lot of heat which will need a large heatsink, a large host and presumably have more room for the driver.

I have that perspective regarding boost drivers in general. I don't particularly like them, and hardly ever use them for anything.

In theory, I could see this being useful in a 26650 or 32650 host. I'd actually like to be able to run MTG2 LEDs in single cell flashlights.

Regarding paralleling though, I am not entirely confident in the cleanliness of the output. I've seen some strange behavior where it seems like the regulators take some time to get "in sync". It may actually just be a matter of beefing up the INPUT caps to solve a lot of paralleling issues with switching regs. But at any rate, one driver instead of two would be nice.

 

[ARG]

I agree - I much prefer buck drivers.

Are there any open source buck drivers out there? I could see a high current buck driver being much more useful than high current boost driver.

 

[IWIRE]

I'm sure your right because this isn't the first time I've heard 10a at that current level with boost drives.
I built one using two sxb`s in parallel because that was available. When the cell is fully charged it draws 6a on start up and 5.5a during normal operation.
Does the input current go up as the cell depletes ? I guess I could just go run a cell down and find out.

I agree - I much prefer buck drivers.

Are there any open source buck drivers out there? I could see a high current buck driver being much more useful than high current boost driver.

That would be sweet. I've heard rumors of them. Think they are related to mythical Unicorns

 

[The Lightning Stalker]

Does the input current go up as the cell depletes ?

Yes. If output power must remain the same, then it
follows that the input power must also remain the
same. If input voltage is dropping, then where must
the input current go in order to maintain the same
power level?

In the real world, more current means more power
dissipated as heat through intrinsic resistances.
Therefore, as input voltage goes down, real input
power actually must INCREASE to compensate for
this lost efficiency.

 

[rhd]

Not only must input power stay the same as cell voltage drops, but it probably actually requires bit more wattage due to efficiency losses increasing as the in/out voltage difference grows larger.

The PCB size would double... any efficiency gains made up by combining two per board would be lost due to the need for larger traces and more input capacitance to compensate for the paralleling.

ARG: Re open bucks. Not sure if there are open designs. I've got many good bucks though, if you ever need something for a specific task. I have up to 4A in sub-1cm^2 that are super straightforward (just a pain to reflow, but fine for a driver or two at a time). 4A to 10A gets larger.

 

[IWIRE]

Yes. If output power must remain the same, then it
follows that the input power must also remain the
same. If input voltage is dropping, then where must
the input current go in order to maintain the same
power level?

In the real world, more current means more power
dissipated as heat through intrinsic resistances.
Therefore, as input voltage goes down, real input
power actually must INCREASE to compensate for
this lost efficiency.

Thanks. I thought that is where the discrepancy must be. I'm just learning about drives and how they actually operate. I can get a design into Eagle from Webbench and I can almost get a board set up. If I could just stop forgetting how I did something the day before

 

[rhd]

I think this (datasheet attached) is the winner for an IC to transition to for use in the next generation.

- Simple IC, and comes in a smaller package
- 1.2 Mhz, so almost 2.5x the switching frequency of the IC used in RC1 (smaller inductors needed)
- External MOSFET, so we can increase the current

 

[The Lightning Stalker]

That does indeed look very nice. I don't know how
you find these things. 1.2Mhz is pretty fast to be
switching a FET. Any idea of the switching losses?

 

[rhd]

That does indeed look very nice. I don't know how
you find these things. 1.2Mhz is pretty fast to be
switching a FET. Any idea of the switching losses?

This might work for just a bit above 4A of boosted output.... shot in the dark though - this is getting up into some pretty heavy input currents.

Maybe this fall I'll test the design. I'll grab some of the chips to have on hand.

 

[IWIRE]

this is getting up into some pretty heavy input currents.

It really is. I couldn't believe how much more 4.4a put a load on host components than 3a does. I had switches melting, voltage drop and heat at threaded connections. I finally got it all working but your right, it's asking a lot from components.
Thanks for working on this. Sounds like your already planning your Fall/Winter projects too. :beer:

 

[rhd]

Just ordered 30 eup2573. Will test the next iteration of this driver in the fall.