Linear Dropout Regulators and LiFePO4

Chrontius · Sep 26, 2009

It occurs to me that some of the excess heat produced by LDO regulators on LiCo cells is from dropping the voltage - the rest is from the LED as expected. What if we fed them something a little less crunchy? Instead of 3.7v, 3.3v from a LiFe. I'd like to see some numbers on this, and while I'll look into getting a LiFe cell/charger (and lightmeter x_x ) to try this out, I don't have any good use for them after one test. Anybody else got a suitable light/cell combination?

Hypothesis: it'll depend very strongly on the Vf of the emitter, but could result in full-power output with astoundingly high efficiency - for such an old, "inefficient" technology.

TorchBoy · Sep 27, 2009

I'll bite. What are you calling 'such an old, "inefficient" technology'? In the situation you're talking about, linear regulators have always depended on Vf of the LED and the supply voltage of the regulator. That's not going to change; LiFePO4 cells simply give another option for powering them, but it's an option that may limit you to lower drive currents because of the Vf of presently available LEDs.

What do you consider "astoundingly high efficiency"? Linear regulators can easily get to 97% even without using those cells, and it's also possible with good selection of components to maintain over 90% for almost the entire battery life.

Experiment away!

Justin Case · Sep 27, 2009

Not only do you need low Vf, you need enough voltage headroom to run the LDO regulator in regulation. So that means a Vf of at most ~3.2V (if the LiFePO4 cell can really hold 3.3V). And you'd probably prefer even lower Vf since running right at the limit of Vbatt=Vf+dropout voltage isn't going to give you a long runtime in regulation. So let's say you'd prefer Vf(max) ~ 3.1V. Well, that is probably rather limiting in terms of corresponding drive current.

Chrontius · Sep 28, 2009

TorchBoy said:
I'll bite. What are you calling 'such an old, "inefficient" technology'? In the situation you're talking about, linear regulators have always depended on Vf of the LED and the supply voltage of the regulator. That's not going to change; LiFePO4 cells simply give another option for powering them, but it's an option that may limit you to lower drive currents because of the Vf of presently available LEDs.

What do you consider "astoundingly high efficiency"? Linear regulators can easily get to 97% even without using those cells, and it's also possible with good selection of components to maintain over 90% for almost the entire battery life.

Experiment away!

I was referring (mostly) to an early AWR LDO regulater, with a "Version One" tag etched on the PCB... though as I understand it, by using a transistor in a partially-on state, you're dissipating whatever excess voltage as heat. Don't take my word for it...

Wikipedia said:
It is very inefficient compared to a switched-mode power supply, since regulates the voltage by burning off "unwanted" voltage as heat.

Seems like the Rebel would be ideally suited for LiFe technology, based on typical Vfs from the datasheet...

uk_caver · Sep 28, 2009

TorchBoy said:
What do you consider "astoundingly high efficiency"? Linear regulators can easily get to 97% even without using those cells, and it's also possible with good selection of components to maintain over 90% for almost the entire battery life.

'Regulator efficiency' is a debatable goal with linear regulators.

Assuming LEDs were from the same power bin, I could easily make the lights I build (running via linear regulation off 3xNiMH or 3xalkaline) more 'efficient' in regulator terms by using higher Vf LEDs, but that would seem to be a dubious exercise.

Since the overall efficiency (in lumens per mA) is the same, unless it is harder to deal with the heat in the linear regulator than the LED, it seems to make general sense to use the lowest Vf LED, to gain maximum full-power runtime.
All that having a higher Vf would do is shift inefficiency in power use from the electronics to the LED.

rushnrockt · Sep 28, 2009

uk_caver said:
'Regulator efficiency' is a debatable goal with linear regulators.

Assuming LEDs were from the same power bin, I could easily make the lights I build (running via linear regulation off 3xNiMH or 3xalkaline) more 'efficient' in regulator terms by using higher Vf LEDs, but that would seem to be a dubious exercise.

Since the overall efficiency (in lumens per mA) is the same, unless it is harder to deal with the heat in the linear regulator than the LED, it seems to make general sense to use the lowest Vf LED, to gain maximum full-power runtime.
All that having a higher Vf would do is shift inefficiency in power use from the electronics to the LED.

The efficiency gain that is targeted here is that of lumens/total system power consumption. Having a battery close to the Vf of the LED and assuming the dropout of the LDO can handle it, will of course give you gains from that perspective.

Calling LDO "inefficient" can be wrong based on the application and the type of switcher it is competing against. Wikipedia can't address all the possible applications. Assuming Vf of 2.7V and LiFePO4 going from 3.2V to 2.9V, you are getting an efficiency of 84 to 93%. Granted, you might not be getting the most out of the battery either.

TorchBoy · Sep 28, 2009

I'm not quite sure what what ball we're throwing around any more. I think we're all agreed that linear regulators "burn off" excess voltage as heat (I don't think there was ever any doubt about that). I think we're all agreed that they can be very efficient (whatever Wikipedia might incorrectly or misleadingly say) or very inefficient depending on how many volts are being dissipated. In the situation we're talking about, where better linear regs tend to be quite efficient, I think you're quite right, uk_caver, there's really no point in going for those few extra percent.

I guess that only leaves the question of whether LiFePO4 cells would work well with any particular linear regulators. If you're right about the 2.9 V end voltage rushnrockt then I suspect not. You're not going to get much light out of any white LED feeding it only 2.7-2.8 V.

rushnrockt · Sep 28, 2009

TorchBoy said:
I'm not quite sure what what ball we're throwing around any more. I think we're all agreed that linear regulators "burn off" excess voltage as heat (I don't think there was ever any doubt about that). I think we're all agreed that they can be very efficient (whatever Wikipedia might incorrectly or misleadingly say) or very inefficient depending on how many volts are being dissipated. In the situation we're talking about, where better linear regs tend to be quite efficient, I think you're quite right, uk_caver, there's really no point in going for those few extra percent.

I guess that only leaves the question of whether LiFePO4 cells would work well with any particular linear regulators. If you're right about the 2.9 V end voltage rushnrockt then I suspect not. You're not going to get much light out of any white LED feeding it only 2.7-2.8 V.

I used numbers those numbers for simplicity. I think we need to have OP provide us with the desired LED/Vf and we can go on calculating from there. I won't speculate any more until then

I can see how there can be a considerable improvement in total system efficiency with a particular setup, but that does not necessarily translate into longer run time either. Anyhoo, waiting!

uk_caver · Sep 28, 2009

In the end a user is only really likely to be interested in how bright a light is, and what the runtime is on a given power source, and in a linear light, up to the point of regulator dropout, that won't depend on the LED Vf.

I suppose it's possible to try and get better efficiency by switching to a slightly lower voltage battery chemistry, but if efficiency was the only reason for making that switch, that seems a strange thing to do if the other consequences (capacity, runtime) are negative.

TorchBoy · Sep 28, 2009

Ah, but think how much safer you'll feel while caving with LiFePO4 cells at the back of your head instead of those nasty explosive Li-ions.

(I use NiMH.)

Chrontius · Sep 28, 2009

rushnrockt said:
I used numbers those numbers for simplicity. I think we need to have OP provide us with the desired LED/Vf and we can go on calculating from there. I won't speculate any more until then

I can see how there can be a considerable improvement in total system efficiency with a particular setup, but that does not necessarily translate into longer run time either. Anyhoo, waiting!

... I've created a monster. :crackup:

What I have is a Groovy with an AWR LDO sandwich and a Luxeon 3. (Note, this is older technology than what was originally included with the Groovy when it was in the concept stage!) While pursuing the expensive options, (a new light engine starts at around $60, and my awesome custom idea is also hella expensive) I'm also looking into cheaper ways of powering the little bugger, as while it'll run on primaries, it'll sit up and shout on 3.7v IMRs. I like that, but runtime is short and it runs hot... and the driver has no low mode, making any use with 3.7v run hot and short.

I'm not ruling out the idea of an emitter swap, if I don't have to find a compatible 17mm Fraen reflector/optic... or matching the two is easy. What I'd really like is a cooler, longer-running, but very nearly as bright light... by swapping my red-label AWs for blue-labels.

uk_caver said:
I suppose it's possible to try and get better efficiency by switching to a slightly lower voltage battery chemistry, but if efficiency was the only reason for making that switch, that seems a strange thing to do if the other consequences (capacity, runtime) are negative.

Well, having only played with IMR batteries, the LiFe batteries are about the same capacity and better efficiency, it should run cooler (and thus, more efficiently).

Quoting Wikipedia again, LiFe has very nearly the same energy/gram that LiCo has, in theory, just at a lower voltage.

Code:

Cathode Material	Average Voltage	Gravimetric Capacity	Gravimetric Energy
LiCoO2	3.7 V	140 mAh/g	[B]0.518[/B] kW·h/kg
LiMn2O4	4.0 V	100 mAh/g	0.400 kW·h/kg
LiFePO4	3.3 V	150 mAh/g	[B]0.495[/B] kW·h/kg
Li2FePO4F	3.6 V	115 mAh/g	0.414 kW·h/kg

TorchBoy · Sep 28, 2009

If you use CODE tags instead of QUOTE you can get the columns to line up, and that data is just the cathodes. Li-ion batteries have more to them than just cathode. 100-160 Wh/kg is the actual range for Li-ion cells and 90-110 Wh/kg for LiFePO4.

rushnrockt · Sep 30, 2009

What is the Vf for the Rebel 3? Pardon my laziness

uk_caver · Sep 30, 2009

TorchBoy said:
Ah, but think how much safer you'll feel while caving with LiFePO4 cells at the back of your head instead of those nasty explosive Li-ions. (I use NiMH.)

So do I.

Chrontius · Sep 30, 2009

rushnrockt said:
What is the Vf for the Rebel 3? Pardon my laziness

About 3.15v

TorchBoy · Sep 30, 2009

I'm no LiFePO4 expert so I went and found some LiFePO4 cell discharge graphs:
http://static.rcgroups.com/forums/attachments/4/4/9/7/4/a1690927-226-LiFePo4-1500-versus-A123-11.gif
http://www.yesa.com.hk/uppic/200710/2007101113592887292.jpg
http://maximusrc.free.fr/images/p009_1_00.jpg

There's not much voltage there for white LEDs, but if you were happy with 350 mA they would probably work with linear regulators like the AMC7135.

BillyNoMates · Oct 1, 2009

I've measured Vf of the Rebels I have been using to be around 2.90 - 3.10v for currents between 300mA and 850mA, so these would probably just about work with a simple linear regulator and the battery characteristics you have been looking at (providing you stick to one LED per battery to keep the load down).

Isn't the problem more to do with not being able to guarantee the forward voltage of the LED one buys? To be >90% efficient with a linear voltage, you need the LED voltage to be within 10% of the battery voltage, when we are talking about 3v devices (or thereabouts) there is very little margin for error, you could end up with very few LEDs that work in your application. As I mentioned, I have measured some low Vfs with Rebels, but they all came from the same bit of tape - I have no way of knowing if I went out and bought some more they would be similarly low.

My preference is always to go for design flexibility/robustness. When driving a white led from a single Li cell, I'll choose a buck/boost every time. It may not be the most electrically efficient solution possible for all possible applications, but I don't need to worry about LED Vf or battery voltage characteristics, it just works with no fuss. Same driver works for White or Red LEDs, LiPo (3.7V) LiFePO4 (3.2V) 3xNiMH, 4xNiMH etc etc

uk_caver · Oct 1, 2009

Could you chose the Vf bin of the Rebels you bought (and if so, was it G bin), or was voltage a matter of luck?

BillyNoMates · Oct 1, 2009

uk_caver said:
Could you chose the Vf bin of the Rebels you bought (and if so, was it G bin), or was voltage a matter of luck?

No, I bought them from Farnell they are 80lm neutral whites, but I couldn't choose a Vf bin - it was pure chance.

HarryN · Oct 1, 2009

One approach is to use K2s instead. Example - Photon Fanatic sells TVOD which are pretty low Vf, and measured at 1000 ma, so it is even lower when used at lower currents.

Since the K2 can operate at such a wide current range (even up to 3 amps by the looks of the JRT1962 charts in the LED section), your can just "go for it".

The rebels are really nice, but the K2s are more forgiving, at least according to my buddies at PL.

You are almost within range of just adding a diode in series instead of the LDO. I have been toying around with using that in some "nearly at the right voltage" setups. As the current goes up, the Vf of both the LED and diode also rise, sort of keeping the current under control. Of course, there is that rather nasty item of the Vf of both falling as they get hot. :thinking:

Linear Dropout Regulators and LiFePO4

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Newly Enlightened

Flashlight Enthusiast

Newly Enlightened

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Newly Enlightened

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Newly Enlightened

Flashlight Enthusiast

Newly Enlightened

Flashlight Enthusiast

Similar threads