Just how much power from a single 123 cell?

[McGizmo]

Hi guys,

I'm not sure which forum this thread should be in but since it relates to those of us modifying or building our own lights, I figured I'd place it here.

Quite some time ago I was messing with a BB750 and one watt LED in a build and I noticed that the current going to the LED was in the low 600 mA's when the circuit was hooked up to a single 123 cell. Since then I have shied away from trying to use a constant current driver above 600 mA with a single 123. The other issue at the time was the fact that 600 mA is almost double the spec on the 1W and a typical 1x123 light isn't boasting massive amounts of metal for thermal relief.

Well now we have the Luxeon III's and they can take currents of up to an amp. Just how much current can we provide to the LED using a single 123 cell? Tonight, I built a 667 mA BB and went to the trouble of putting one meter on the voltage of the 123 cell and another on the current to the LED. I used a fresh SF 123 cell. Before the first minute was up, the current was down at 600 mA and the voltage drop on the battery was at 2.05 volts. My test LED has a Vf of 3.3 at 350 mA so it's really on the friendly side. I don't trust my equipment let alone myself 100% on tests like this but it sure didn't seem that the 667 drover was able to run at spec! For the heck of it, I hooked the driver and LED up to my bench supply and set the voltage at 2.05 volts. The current to the LED was 667 mA and I noticed the current being drawn by the circuit was 1.8 amps!

I then changed the sense resistors to make the driver a 611 mA driver. I hooked up the equipment and again saw within less than a minute about 600 mA going to the LED and the voltage drop across the battery was just barely above 2 volts. I touched the IC and noticed it was definitely hot.

Frankly guys, I would like to see some of you electronic experts do some thinking and perhaps emphirical study on the capacities of some of these systems we are building! I hope I am wrong but I sense a ceiling of somewhere around 600 mA current to the LED on a single 123. I think using higher current drivers is not bringing more light out the front end?!?!?

I think one could do the math and figure out how much power is required but the voltage drop on the battery dictates how much current it needs to get you this power. What are realistic max targets for the 1x123 power pak?

I should mention that initially, the current and voltage measurements I saw looked good but you could watch the numbers just plumet in no time. I imagine if a lux meter had been mesuring the light output that it too would have started out high and then dropped.

Forgive my rambling and hopefully someone can set me straight here.

 

[paulr]

I think something is wrong. Surefire HOLA bulbs pull over 2 amps from those batteries. I wonder if you could measure the voltage drop across the cells while running a P61 or something.

 

[G Pilot]

So what the heck do I do with my brand new BB750? I was going to use it on a 3 watt and a single 123. Maybe I can find a 5 watt /ubbthreads/images/graemlins/confused.gif Anyone out there have a spare 5 watt?

 

[hotbeam]

G Pilot, the BB is great when your battery source voltage is close to the voltage of the LED, eg. 6v for a 5w and 3v for a 1w/3w (3w is a different case). If you are trying to run a 5w from 1 x 123, you'll be disappointed. Of course, you will get light. You will also cook the driver chip (on prolonged run) and your runtime will be disappointing.

 

[G Pilot]

If I am understanding McGizmo's post right, basically my BB750 is only good for 600ma from a singel 123? So driving it with two 123's to a 5 watt would be a better choice, right?

 

[Miciobigio]

Don,

your experimentation is a big help for all of us "lovers of little 1x123 led-flashlight" but it's not a good news /ubbthreads/images/graemlins/mecry.gif.
Just one question , did you provide some sort of heat dissipation while you were doing the test ?
It is very important to know if BB > 600mA do not provide more than 600mA to the led if driven by a single 123 !

 

[MR Bulk]

Hmmm, I guess I'm forced to chime in here for fear of mass desertion from the 100+ people waiting in line for the new Super Baby Pin (Thanks, Don!).

When we did the McModule run using a BB750 and two 123s, the circuit quickly overheated and the light began to flicker. I had to "pot" the circuit in thermal goo to keep it running correctly.

In the "regular" Baby Pin, it was driven off a BB750 using a single CR123 and output was spectacular (about 1000 lux output average from all 60+ Baby Pins) with no overheating, and although output did eventually decline to the point of 85 lux, this was after an hour of being left alone, steadily running with no "human heatsinking" from any hands. And when the light was turned off for five minutes and switched back on, it made and continued to make 260 lux for several more minutes. Also for more details from an unbiased source take a look at Bernhard Kiessling's run time and beamshot test of a variety of little lights including the BP, most if not all running on single CR123s.

Now with the SUPER Baby Pin a completely new format of converter was invented, one that allows near 1500 lux of initial output and we expect a similar run time but with the output proportionally scaled upward to reflect the higher initial brightness readings.

As well one must consider what we are actually going for when building a little light with big brightness -- for me personally it is not necessarily long run times (I as well as nearly all True Flashoholics have a multitude of other lights for that) but something you will always have in your pocket or purse for those times, perhaps safety- or even life-threatening times, when huge light output is a Godsend for seeing what that shadowy figure is across the parking lot next to your car or, as SureFire so aptly advertises, "stunning assailants by temporarily blinding them", something your average Opalec/Arc LS/Inova/whatever simply will not do to the degree of a seriously bright yet pocketable little light.

So there -- thass muh story an' ah'm stickin' to it!

 

[kj]

It's very interesting. Thank you for sharing these data with us.

Anyway, like paulr, I've seen over 2A current on SF incan so I'm thinking we can pull more current from the cell and that may be a BB's limitation. Wayne wrote, in his FAQ page, there was a critical fixed parameter in BB.

[ QUOTE ]
Q: How much current could Badboy be set to?
A: This depends on what battery voltage you are using. The critical fixed factor is the input current which is around 1.5A. This limits the input power (Vin * Iin).

[/ QUOTE ]
How do you think? I'm not an electronics expert so I would like to hear opinions from gurus. Actually, I still feel strange since BB pulled 1.8A from the bench supply.

BTW, maybe the converter board that now Mr BULK is working on can handle more input current. I don't know any detail, though.

***EDIT***
Oops, MR Bulk just chimed in!

 

[MR Bulk]

A very astute observation, kj...

 

[PaulW]

[ QUOTE ]
paulr said:
I think something is wrong. Surefire HOLA bulbs pull over 2 amps from those batteries. I wonder if you could measure the voltage drop across the cells while running a P61 or something.

[/ QUOTE ]

In fact, I did just that. When I ran a SureFire MN16 (250 lumen) bulb on three fresh 123s, I measured 6.8 volts at 2.48 amps. That's 2.26 volts per cell. Each cell is putting out 5.6 watts.

The maximum power output will be achieved when the load resistance is equal to the battery's internal resistance. I have measured the internal resistance at about 0.3 ohms per cell when fresh. So, if a fresh 123 were connected to 0.3 ohms, the total resistance in the circuit would be 0.6 ohms, and the current drawn would be 3.0 / 0.6 = 5 amps. The voltage across the load resistance would be 1.5 volts. The power would then be 5 * 1.5 = 7.5 watts, the maximum such a cell could ever put out. (Of course, an additional 7.5 watts is being consumed by the cell's internal resistance, causing it to get very hot.)

As an interesting side note, the maximum current available from a battery is approximately its open circuit voltage divided by its internal resistance. That would be 3.0 / 0.3 = 10 amps. But that level won't last for long. To do this you'd have to short the cell, and all the power would be dissipated within the cell; none would get out except as heat.

Paul

 

[evan9162]

Paul,

Your measurement of cell internal resistance for 123s is actually very good. Duracell rates their cells for 0.25 ohms internal resistance (Energizer does not specify this value).

 

[McGizmo]

Hey guys,

I am certainly not trying to cause any problems here and I openly admit that there may be all kinds of flaws in my bench testing! I am using leads with clips to make the meter connections. There are voltage drops across the meter that will effect the current to the LED. My meters used are Fluke true RMS but with small gauge leads, perhaps this is causing some error. It is interesting though that the power supply when substituted for the battery does show a different picture. This is with the rest of the cobbled circuit being the same.

Bottom line: My test is suspect as hell! BUT has anyone done a proper test or are we all going to continue on throwing numbers around that we like and figure that our circuits are performing as we expect and more significant, want them to perform?

I understand that the 123 cell can deliver multiple amps. What is the voltage drop though to get to those amps? Don't forget that power is volts x amps. I suspect that if you take a 2x123 HOLA incandescent and put it to a bench supply at 6 volts, it will not last nearly as long as it would in a flashlight. Same thing with DD of a 5W Luxeon in 3x123. We know this works but put the 5W on a bench supply of 9 volts and have an envelope ready to send the carcass to Doug S. Why? because in the cases of batteries, we have the internal resistances and voltage drops under loads which limit the actual power coming from the batteries.

Let me put it another way, with a single 123 cell, what is the sustainable (good for a run time of 1 minute, 2 minutes?) power that it can provide? What is the voltage and what is the current? As I understand it, the BB has an internal limit of about 1.5 amps and I have seen as have others that this driver will shut down after a period of time due to the heat and or current?

I just want to know what the realistic limits are with the components we are using are. I have built 700 mA driver lights hosted on 1x123 and they DO work, no doubt about it. However, if, as I expect, a driver set for a lower current output will provide as much actual power to the system and possibly with less strain, shouldn't we know this?

I think everyone agrees that we can't get 5 watts out of a single 123 cell. How many watts can we get? Will the BadBoy or some other type of boost driver allow us to effectively pull this power out of the battery and use it to drive the Luxeon III's?

Let me throw some simple math at you and PLEASE correct me if I am looking at this wrong!

Lets take a Luxeon III that has a Vf of 3.9 V at 1 amp. That gives us 3.9 watts of power. Now lets assume that a driver is 90% efficient ( we wish). 3.9/.90 = 4.33 so the driver needs to output 4.33 watts, agreed?

Now let's take a 3 volt battery that's on PCP and it can provide whatever power asked without any voltage drop. Well for 4.33 watts, at 3 volts, it will be providing 4.33/3 = 1.44 amps. OK cool, the driver will see 1.44 amps and in the case of the BB, this is below the limit. Now lets assume that the battery can provide the current but it will have a voltage drop. Say the drop is down to 2.5 volts. Now 4.33/2.5 = 1.73 amps. I think we may have a problem here, certainly with the BB driver! Now if the voltage drop is as bad as I saw in my suspect testing (this was at a current demand considerably below Io of 1 amp), we have 4.33/2.05 = 2.11 amps. Nope, we can't use a BB1000 on a single 123 cell and hope for success here.

So, at least with the BB drivers, there is no point in attempting to drive the Luxeon III at full power, even if we have the mass and surface area to accommodate the heat. Simple question; how much current can we drive the Luxeon III at? This is a rephrase of the subject title.

I am going to be building some single 123 cell powered Luxeon III heads and I want to know what a realistic expectation and target output current should be; runtime be damned. On the other hand, I see no reason to push the circuit or system beyond that which it is capable of. /ubbthreads/images/graemlins/thinking.gif /ubbthreads/images/graemlins/help.gif

 

[evan9162]

Don,

One thing to consider also - are you measuring the average current from the battery? I'm assuming so, since you're using RMS meters. If that's the case, then things are even worse. It's averaging the current from when the power is being switched through the inductor with the current from when the switch is off. I'll bet that the switch current is even higher than what you estimate from your power equations, and peak currents from the battery will be higher than that as well.

Also, are you putting your current meters in series between the LED and driver board (to measure LED current)? If so, you may be introducing inductance that may be throwing off the BadBoy converter. I seem to remember Wayne (dat2zip) saying something about using a current meter screws up the way a BB operates, causing it to pull more power from the battery and deliver less power to the load...though I can't be 100% positive about this.

We really need Wayne to chime in on this. However, I definitely believe you when you say it's impractical to drive a L3 from a single 123 cell. The power requirements vs. delivery capablity of a 123 cell just don't make sense in these conditions, especially since a switching converter is involved, especially if there are some limits in the converter its self that we are hitting.

 

[Doug S]

[ QUOTE ]
McGizmo said:
Hi guys,

I'm not sure which forum this thread should be in but since it relates to those of us modifying or building our own lights, I figured I'd place it here.

Quite some time ago I was messing with a BB750 and one watt LED in a build and I noticed that the current going to the LED was in the low 600 mA's when the circuit was hooked up to a single 123 cell. Since then I have shied away from trying to use a constant current driver above 600 mA with a single 123. The other issue at the time was the fact that 600 mA is almost double the spec on the 1W and a typical 1x123 light isn't boasting massive amounts of metal for thermal relief.

Well now we have the Luxeon III's and they can take currents of up to an amp. Just how much current can we provide to the LED using a single 123 cell? Tonight, I built a 667 mA BB and went to the trouble of putting one meter on the voltage of the 123 cell and another on the current to the LED. I used a fresh SF 123 cell. Before the first minute was up, the current was down at 600 mA and the voltage drop on the battery was at 2.05 volts. My test LED has a Vf of 3.3 at 350 mA so it's really on the friendly side. I don't trust my equipment let alone myself 100% on tests like this but it sure didn't seem that the 667 drover was able to run at spec! For the heck of it, I hooked the driver and LED up to my bench supply and set the voltage at 2.05 volts. The current to the LED was 667 mA and I noticed the current being drawn by the circuit was 1.8 amps!

I then changed the sense resistors to make the driver a 611 mA driver. I hooked up the equipment and again saw within less than a minute about 600 mA going to the LED and the voltage drop across the battery was just barely above 2 volts. I touched the IC and noticed it was definitely hot.

Frankly guys, I would like to see some of you electronic experts do some thinking and perhaps emphirical study on the capacities of some of these systems we are building! I hope I am wrong but I sense a ceiling of somewhere around 600 mA current to the LED on a single 123. I think using higher current drivers is not bringing more light out the front end?!?!?

I think one could do the math and figure out how much power is required but the voltage drop on the battery dictates how much current it needs to get you this power. What are realistic max targets for the 1x123 power pak?

I should mention that initially, the current and voltage measurements I saw looked good but you could watch the numbers just plumet in no time. I imagine if a lux meter had been mesuring the light output that it too would have started out high and then dropped.

Forgive my rambling and hopefully someone can set me straight here.

[/ QUOTE ]

Don, I don't think that anyone needs to set you straight. Your empirical evidence is in close agreement with what the theoretical approach to the problem would predict. I will rise to the bait you have offered but first I will ramble a bit as an aside. As you know, my resources and skills for mechanical fabrication of mods are rather limited. As a result, my role here on the CPF has mostly been relegated to that of chirping on the sidelines about technical issues while you and others are building the neat stuff. Lately, I have been watching with interest and admiration while you, Larry Black, Dan C., Katokichi, Chief Wiggum, and others that I am probably forgetting right now, have raised [or is it shrunk] the bar on bright and small CR2 based lights. It has got me to thinking about the theoretical limitations [with current technology] on the size of a light meeting the arbitrary criteria of driving a 1W Luxeon at max rated for 1 hour minimum regulated output. I have researched the subject a bit and hope to post on the subject eventually. In a nutshell, the CR2 appears to be about the optimum for this set of criteria. Anyway, my research on this question has yielded info relevant to the issues that you raise in this thread. Since this thread has a wider audience, I will be telling you some stuff that you likely already know.
Your observed limitations on the drive current into a Lux load from a single CR123 driving a BB results from the interaction of the charactistics of CR123 cells and the Badboy converter board and specifically the Linear Technologies LT1618 stepup switching regulator IC which is the heart of the BB. Here are the key considerations:
1. The terminal voltage and total extractable energy of a battery decreases with increasing power demand from the battery. The graphical data on the Duracell 123 datasheet illustrates this well. Open the following link in another window: Duracell 123 datasheet
2. The efficiency of a stepup switching regulator decreases with decreasing input voltage.
3. The efficiency of a stepup switching regulator decreases with increasing [beyond a certain low level] load.
4. The LT1618 switch current limit has a typical 25C value of 2.1A with datasheet min/max limits of 1.5/2.8A.
5. Whatever the actual switch current limit of the particular IC, it decreases by about 10% as the IC temp increases from 25C to 125C.
6. Whenever the load/voltage source [assuming Vin>than the IC minimum of 1.6V] prevents the IC from delivering the setpoint current, the IC current will ramp up to the switch current limit on each cycle. Note that the average current into the IC is less than the switch current limit since the current varies in a sawtooth pattern each cycle between the current limit value and a lesser value. This is the *least* efficient possible operating condition for the IC. My testing on the LT1618 configured as in the BB shows that driving a single luxeon load with a Vin of 2V the efficiency is *at best* around 0.6 [60%] when the IC is operating in current limited mode.
So the general scenario of your testing is that the heavy demands of the high current BB draws high currents from the battery which in turn causes its voltage to sag, thus requiring more current for the same power input and even extra additional current due to the declining efficiency of the IC at higher current/lower input voltage thus causing the battery voltage to sag a little more, etc., etc. Meanwhile, the IC is heating up which causes its current limit value to drop. Before long, the maximum input power the IC can draw, reduced by the lowered efficiency is less output power than the load requires at the BB setpoint current. The system runs "balls to the wall" with slowly declining output as the battery voltage declines with discharge. Under these conditions, about 80% [!!!] of the battery energy that would be available under very low drains is dissapated in the battery and the converter as heat. While I acknowledge Mr Bulk's position that "when you drive a Ferrari you don't care about gas mileage" some of us do care about gas mileage. I don't think we'll ever see Darell driving a Hummer /ubbthreads/images/graemlins/tongue.gif
Now let's look at Don's specific test data:
With his bench supply at 2.05V the IC draws 1.8A. This gives an input power of (2.05V)(1.8A)=3.69W. Multiplying by an efficiency of 60% gives an output power of (0.6)(3.69)=2.21W. Now if we assume an LED Vf of 3.5V, we should be able to get an output current of 2.21W/3.5V=0.633A. Don actually got 0.667A which isn't to far off.
Don speculates that the ceiling of a Lux/BB/1X123 combo is 600mA and this appears about right to me. Whether this is a practical limit is a more subjective judgement. If you look at the Duracell datasheet, at 4W load, a cell can deliver about 1.4Whr but at 1.5W it can deliver 3Whr [remember these are log-log graphs, trust me if you are unfamiliar with interpreting these]. When you factor in the efficiency of 80%+ vs 60% of the BB at 350mA vs 700mA [123 cell power source] you are looking at 25% the runtime at 700 vs 350mA to get perhaps 60% increase in light.
Now to the broader question of the limit to drive current from a single 123 cell, Don's 600mA figure is with a 60% efficient converter. It is certainly possible to design a converter that will deliver 900mA from a 2V source at an efficiency of 90% which is exactly the same input power. The problem is that to achieve the higher efficiency larger/more components are required which would make it very difficult using the same board area/volume of the BB. Personally, if I have the 90% efficient convertor, I would opt for double the runtime at 600mA rather than the small increase in light obtained at 900. I'll leave the Ferrari to others.
EDIT: All this written without seeing Don's post that followed the one I am responding to. /ubbthreads/images/graemlins/grin.gif

 

[PaulW]

Don,

My apologies. I didn't thoroughly read your original post and answered a question you weren't asking. I can ramble on at length with theoretical discussion. I love it. But that's not much help in answering a practical question.

The numbers I was throwing around are for fresh cells only. I suppose that for a constant load they could be extended to take into account a cell's degradation as it is drained. But for a varying load, such as presented by a regulator, things get murky. The fact that more current is drawn as the cell discharges is a consideration beyond what I can accommodate.

You called for "some of you electronic experts do some thinking and perhaps emphirical study on the capacities of some of these systems we are building!" If I had read this carefully, I would have kept quiet and waited. Maybe I ought to do that now. /ubbthreads/images/graemlins/smile.gif

Paul

 

[LEDmodMan]

Don,
I don't see anything wrong with your math. /ubbthreads/images/graemlins/grin.gif

Something does seem strange about the initial test though, now that I look back at it. I would expect better performance from the 123 battery and the BB.

The IC needs to boost the voltage to 3.3v and deliver 350mA at spec, right? That's only 1.2 Watts. If your LS Vf rose to about 3.7v (probably less but we'll take the worst case scenario here) at 667mA, that is only 2.47 Watts. Taking the converter efficiencies into account (say %80), you should be pulling about 3.1 Watts from the battery at this level. At 3v nominal, that's about an Amp. With voltage drop, I would expect it to be right about 1.15A give or take a bit. Hmm... /ubbthreads/images/graemlins/thinking.gif

At the 2.05 input voltage from your bench supply, I would have expected around 1.5A going to the driver assuming %80 efficiency. Still only 1.7A at %70 efficiency. So that means you're only getting about 67% efficiency from the BB, and I don't recall it ever being that bad. /ubbthreads/images/graemlins/confused.gif Something definitely seems wrong here, but I am at a complete loss to figure out what that may be. Bad driver? /ubbthreads/images/graemlins/icon3.gif

/ubbthreads/images/graemlins/popcorn.gif

***edit***
Wow, three posts while typing this... /ubbthreads/images/graemlins/ohgeez.gif
Including that monster post by Doug that supports what I said about the converter efficiency. I never realized it got down to %60 at 2v, OUCH! I defer to Doug's expertise on this matter, as they are greater than mine. He seems to support the experimental results, too bad...

 

[McGizmo]

Doug,

Thanks for taking the time to respond and I was able to understand most of what you said! /ubbthreads/images/graemlins/smile.gif I currently have BB's available to me for my builds and until or unless some other information surfaces, I will be using or at least suggesting a BB 611 or possibly a BB556 as the "high power" driver for the 1x123 builds. I will stick with the Luxeon III since at these levels, the III is being underdriven whereas the 1W is being overdriven; at least by Lumileds data sheets ( what do they know? /ubbthreads/images/graemlins/tongue.gif )

- Don

 

[Doug S]

[ QUOTE ]
evan9162 said:
Don,

One thing to consider also - are you measuring the average current from the battery? I'm assuming so, since you're using RMS meters. If that's the case, then things are even worse. It's averaging the current from when the power is being switched through the inductor with the current from when the switch is off. I'll bet that the switch current is even higher than what you estimate from your power equations, and peak currents from the battery will be higher than that as well.



[/ QUOTE ]
Darin, I think that you are confused here. The the input current waveform is a continuous sawtooth, not an ON/OFF rectangular waveform. You are thinking of a stepdown switcher. RMS reading meters do not measure average current. In Don's application, he really should be measuring with an average reading meter, not an RMS reading meter as he is. His method will be overstating the average current but the error is not large enough to be significant unless he was computing I/O efficiencies in which case I would have to /ubbthreads/images/graemlins/twakfl.gif him a time or two.

 

[LEDmodMan]

This really turned into one heck of a good discussion quickly! /ubbthreads/images/graemlins/thumbsup.gif

 

[Doug S]

[ QUOTE ]
McGizmo said:
Same thing with DD of a 5W Luxeon in 3x123. We know this works but put the 5W on a bench supply of 9 volts and have an envelope ready to send the carcass to Doug S.

[/ QUOTE ]

Standing by! /ubbthreads/images/graemlins/wink.gif

[ QUOTE ]
McGizmo said:

I am going to be building some single 123 cell powered Luxeon III heads and I want to know what a realistic expectation and target output current should be; runtime be damned. On the other hand, I see no reason to push the circuit or system beyond that which it is capable of. /ubbthreads/images/graemlins/thinking.gif /ubbthreads/images/graemlins/help.gif

[/ QUOTE ]


It depends a bit on the intended operating scenario. That Duracell datasheet provides a bit of guidence. Look at that 1.8A 3 seconds on/7 seconds off graph. In short bursts and room temperature the voltage holds up to 2.1V [required for your 600mA version] for over half the discharge, on the other hand at -10C it is a loser from the git go. For continuous operation, at room temperatures and above, I don't think that I would go above a BB500.