LED Vf testing - estimate method

[Blindasabat]

I finally tried a few LEDs direct drive off a 17670 (4.05V no load) all at once to try to estimate the Vf of a few emitters.

First I measured the current off a known Rebel of D bin (3-3.25V) and got 2.2Amp. Same I measured off my TVOD K2 several times. Good so far.
Then I did it to a K2 F bin. 0.9-1A. I had my highest and lowest brackets. So I thought.
Unknown bin K2 = 1.3Amp. An E bin?
Unknown Rebel = 2.2Amp. D bin.
Then I tried an XP-E Q2-5C. 3Amps solid. Woah! It's only rated for 1A. What bin is that? It must be under 3V. I tried it a couple of times for just a second each. Same reading. For a Q2 flux bin, it should be efficient.

Interesting results if you plan to direct drive any LEDs. I should try a CR123 next.

All LEDs were on stars and only for 2-3 seconds except less for the XP-E at 3A.

 

[uplite]

BaaB...this tells you nothing about the Vf of the LEDs, because you don't know how far the cell voltage was sagging under each load.

Just connect a voltmeter across the LED leads, and you can read the actual Vf.

-Jeff

 

[Blindasabat]

It's an estimate method. Comparing to known Vf Leds does tell me something. I used the same cell (same droop) and compared to the current running through two known Vf Leds.
The drive current under one Li-Ion is what I would use the Vf to estimate, so the current is actually perhaps more important than Vf. I guess that is the reason to test this way. I need to know what current I will get, and so doing, I get an estimate of the Vf.
I have labelled a K2 and a Rebel as E & D bins respectively because of this and am very confident in those.

 

[uplite]

"Droop", aka voltage sag, varies with the current draw, cell temperature, cell cycle count, and state of charge. A one-off current measurement does not tell you much about the forward voltage of the LED.

But it sounds like you have a multimeter? Just touch the leads across the emitter and you can read the actual Vf. Super easy. :thumbsup:

-Jeff

 

[Blindasabat]

If I do it as you suggest, I will NOT get Vf. I will get some other (mostly much higher) Voltage because the Rebel, XP-E, & XR-E LEDs are being driven far higher than 350mA spec for Vf. Even the K2 are not being driven at spec. Since I can not drive them at spec current to get an accurate Vf reading, this is a decent way to compare to known LEDs.

I used the SAME Li-Ion on all the LEDs at the same time, so differing sag is not an issue. Notice that I said I only tested for a few seconds each. That was to limit the voltage change of the Li-Ion. It worked - testing at 4.05V before and after testing. I used the one cell specifically to eliminate any sag issues. I never said anyone else can use my numbers, just the method (see the title of the thread).

 

[uplite]

If I do it as you suggest, I will NOT get Vf.

Yes you will. You will get the Vf at a specific current, which you are already measuring. Then you can work backwards from the LED specs (Vf vs If graph in the datasheet) to estimate the Vf at 350mA.

When you measure the actual Vf, you will also have a better idea of how accurate that estimate could be. If the Vf was around 4V, the numbers would be all over the place, because the LEDs would be seriously overdriven and other factors would come into play. But voltage sag on that 17670 cell will make the Vf much lower than 4V under load.

Trying to "guess" the Vf from current draw on direct drive, will not get you anywhere. Even if you charged your 17670 test cell identically for each test (which you clearly didn't, since it was only 4.05V), and let the internal cell temperature stabilize for several hours before each test...it doesn't matter. Very small changes in Vf can produce huge changes in current draw, even in different samples from the same bin. That's why we use a current-reference feedback loop to keep constant current on these high-power LEDs.

-Jeff

 

[Blindasabat]

So with your 'more exact' method, I have to
1) measure Voltage across the LED (that I often can not reach in a light),
2) measure the current, then
3) consult a chart that will stack up these three sources of error.

My Li-ion was 4.05V before and after the testing. MEANING: the same cell, same temp, same charge condition since I tested for only a couple of seconds per LED. It has been eliminated as a source of error.

I did some measurements in a simple way that some people might get a better understanding of LED behavior from, felt like posting it.

 

[uplite]

Yikes! I'm sorry. I was just explaining how you could get more meaningful results if you want to.

You said your LEDs are on stars, so it is super easy to measure their Vf. If you were testing a flashlight, yeah, you'd have to access the emitter board/pill to read the actual Vf.

Testing the current like you did, with all other conditions held constant, is a good way to gauge the efficiency of LEDs from the same make & flux bin. For example, it's a good way to test different XP-E emitters and bin them yourself, since Cree does not bin by Vf. Still a useful technique. :thumbsup:

-Jeff

 

[Blindasabat]

Forgive me, It's just a little frustrating today.

Those Vf charts are not very useful unless your Vf is the exact average of your chart. I measured one K2 at 2.25A DD of an IMR123 (minimal droop and the current was consistant for quite some time). If I took that V and I to the K2 Vf chart, I would have to do some REAL estimation and basically have to slide the chart over to fit my measurements, which would be incorrect in itself. This is a chart with no zero to scale to. The K2 chart is for the average Vf of 3.65V at 1A. The chart only goes up to 1.6A and 3.9V. By the chart, my 4V should only give me somewhere around 1.8A (roughly extrapolated), but I got 2.25A. The chart does't jive there. If I assume there is significant droop, the chart actually makes less and less sense.

Since I have a labeled D-bin K2 to compare to that also gets 2.25A off the same cell at the same time, I conclude that my other K2 must be a D-bin too.

 

[uk_caver]

What is the basic equivalent circuit for an LED?
Ideal diode in series with a resistor and some NTC element?

If Vf varies between devices, is that down largely to the resistive element varying, meaning that if a device Vf is significantly different from the 'standard' Vf, then the I/V curve for the device will have a significantly different slope.
Or is it more complicated than that?

 

[MrGman]

Forgive me, It's just a little frustrating today.

Those Vf charts are not very useful unless your Vf is the exact average of your chart. I measured one K2 at 2.25A DD of an IMR123 (minimal droop and the current was consistant for quite some time). If I took that V and I to the K2 Vf chart, I would have to do some REAL estimation and basically have to slide the chart over to fit my measurements, which would be incorrect in itself. This is a chart with no zero to scale to. The K2 chart is for the average Vf of 3.65V at 1A. The chart only goes up to 1.6A and 3.9V. By the chart, my 4V should only give me somewhere around 1.8A (roughly extrapolated), but I got 2.25A. The chart does't jive there. If I assume there is significant droop, the chart actually makes less and less sense.

Since I have a labeled D-bin K2 to compare to that also gets 2.25A off the same cell at the same time, I conclude that my other K2 must be a D-bin too.

Its not the chart that doesn't make sense its your inaccurate method of testing. You don't know what you are doing and have no way to extrapolate your measured current under load to a Vf of any of those LEDs.

The unloaded voltage of the battery means nothing and doesn't help in your process because you don't know the Vf of the LED or the resistances of the completed circuit.

You were told the real world forward voltage drop of the LED could be measured and would be meaningful at the same real world current draw but you refused to listen and said that it would introduce more errors.

If you have the known current draw and the known Vf at the current draw you can determine a lot about an LED.

Further you stated you were running these things direct drive in the very first sentence of your original post, and then you are saying that the "drive" circuitry will prevent you from measuring Vf of the LEDs that you said you were "direct driving" so you aren't making any sense.

If you have LEDs that are drawing far more current then they normally would off of a single lithium battery then it sounds like direct drive to me. If they are regulated and drawing a different current then what the battery current is, then you can't use the battery current draw at all as a way to interpret the LED current and Vf in the first place.

So first you must know for each flashlight, if the LED has a driver or not. If its direct driven you can first measure the current to the LED and then the voltage drop of the LED directly across the leds to the LED and then you will have something useful. If a particular model light has a regulating driver circuit, knowing the battery current, isn't nearly enough information to figure out LED forward voltage. You would need to know what the driver regulated current is. If you say there is no regulator current because they are all direct driven, then you are back to square one. There is nothing from stopping you to measure the actual Vf across the LED at a known current level and that will be a better starting place then what you have now. IF you had a 0.2 ohm 5 watt resistor to put in series with the battery to these so called direct drive LEDs and reduce the current then you could learn a lot more as well.

It sounds like to me that you should not be the one doing these measurements.

 

[HarryN]

Well - kind of harsh on a hobbyist, aren't we?

I can of course understand the benefit of taking measurements "properly" as well as the "proper" way to drive LEDs, but not all experiments and learning come from doing things "properly".

I learned a great deal about LEDs by doing things wrong, not listening to the information in the data sheets, and burning some out. Knowledge costs time and money, and I see nothing wrong with playing around with LEDs, burning some out, using them improperly, and generally having fun.

Most of the lights that are pushing the limits are built using LEDs that are run outside of the "proper" data sheet operation. Do you really think a "DEFT" is built by the book?

IMHO, what he is doing is in fact giving him information which is useful for the level of sophistication of many hobbyist at a low cost. Not everyone has a power supply or multiple DMMs to take measurements, so this fills a gap in learning about the behaviour of LEDs. Certainly the post is a lot more sophisticated than the 100's of threads in the electronics and battery section asking about what is the cheapest / best battery when they barely know what an AA cell is.

I will admit that the information gathered would be better and easier to interpret if he hooked a 350 or 700ma driver to the led and measured across the leads, as this would make charting it easier.

This is a hobby forum, not an EE class.

 

[uplite]

Well - kind of harsh on a hobbyist, aren't we?

HarryN...I think MrGman was responding in kind to post #7, which was "kind of harsh" before BaaB went back and edited it.

My core advice was "Just connect a voltmeter across the LED leads, and you can read the actual Vf." That is definitely hobby-level. Actually it is more "hobbyist appropriate" than reading the current, which (besides not giving the desired result) is quite likely to blow the fuse on a new hobbyist's multimeter.

The problem was not the level of the advice, it was the response against it. If we aren't here to listen and learn as well as share, what's the point of a public forum? :candle:

:grouphug:
-Jeff

 

[MrGman]

HarryN...I think MrGman was responding in kind to post #7, which was "kind of harsh" before BaaB went back and edited it.

My core advice was "Just connect a voltmeter across the LED leads, and you can read the actual Vf." That is definitely hobby-level. Actually it is more "hobbyist appropriate" than reading the current, which (besides not giving the desired result) is quite likely to blow the fuse on a new hobbyist's multimeter.

The problem was not the level of the advice, it was the response against it. If we aren't here to listen and learn as well as share, what's the point of a public forum? :candle:

:grouphug:


-Jeff

You pretty much nailed it.

And Harry N I pretty much have to disagree with your entire premise. He published data, was given very simple advice on how to take better data that would help him to understand better but refused to consider it. If a person can take current readings without blowing the fuse on the meter than they can surely take voltage readings without have to be an EE. He chose to play technician and help was offered. I consider your advice to the same type of logic for the dumbing down of the public school system. Lets just make it easier and accept mediocre results cause its just for fun, instead of actually learning something useful in a "teachable moment".

 

[Blindasabat]

I think the point is that I was posing an alternate method to the industry standard method, and that I found it to have some accuracy. I don't have a power supply to drive an LED at a spec current, as I suspect many others don't. I was't refusing to do it the right way, I was experimenting and showing what I saw. Harry, thanks for understanding that. Nice accurate post.
But then it came out in the discussion that I actually see more error in trying to account for a non-average Vf (one not shown on the spec sheet chart (see my K2 3.65V example) without a power supply than comparing to a known LED Vf as I did. It is an "estimate method" but I am actually seeing it as possibly a more accurate method if all you have is a cell, two wires and a DMM. If that came out as argumentative, then I apologize. That is what I peceived from my viewpoint.

I don't recall mentioning any circuitry. I may have referred to LED's being 'driven' by a cell or voltage. I also mentioned that sometimes LEDs are not accessible while in lights. I have had that problem with several of mine. I had driverless SF L1 heads for a long time that I could not get open for any effort. That was when I first posted about the measured current on different cells in different heads/LEDs. At the time, I thought it was something else about the different LEDs, but somebody mentioned that it was just different Vf's. I listened, looked into it, and discovered that was the answer. I learned a lot about Vf as a result. I don't see very constructive criticism here, just misunderstanding of my point.

My complaint in further posts was that I was then repeatedly told to just do it the real way and quit my nonsense. The world is flat, let it go. I was told this or that was wrong (namely different cell conditions which I had already accounted for) and causing error, when I believe it was not and tried to re-explain why that was so. That was summarily ignored again.

I still believe that without a settable power supply, you are just estimating off the Vf charts IF your Voltage measurement doesn't luckily end up on the curve of the chart provided with average Vf for that LED. My K2 was like this (not matching the line on the published 3.65V Vf chart), but I was getting the exact same current from the same cell as another K2 measured seconds earlier and seconds later than the unknown K2. And my cell measured the same resting voltage after all the eight seconds or so of testing - showing it had not changed charge during the testing, and therefore should sag the same under the same current. Are there some different droop conditions and/or causes I am not aware of? Is there any reason the same current (in short pulses) at the same voltage would be coming from different binned LED's of the same design & manufacture?

I posit the same as post #1: That this is a valid estimation method. I know I'm stubborn, but I come by it honestly from my O'Brien and Murphy ancestors. And a little bit of scientific method. Emphasis on little bit.

I used to estimate answers to numerical methods and physics lab calculations before my lab partners & I got nearly the same exact answer the long way. I learned that even in rigorous science, there is more than one way to get the right answer. And different people's mids think in different ways, so sometimes you need to explain things a different way.

 

[Blindasabat]

Nice... keep swinging.
If I had a regulated power supply, I'd do it the "right" way. But I'd likely still do it this way as an experiment. Harry was spot on. Read his post again. Especially experimentation part. We learn from experimenting. If we all just do it the same exact way, many people will never learn. Thanks for proving my point for me.
I got criticism that was itself off the mark or maybe not explained well. That method has it's own flaws. See my problems with the chart in my previous post. If it appeared I 'refused' to consider it, then that was a misunderstanding on your part. I did what few here seem to do: I read it, considered, it, then posted my thoughts and critique on it in light of the still seemingly misunderstood topic and reason for posting in the first place: the *estimation* method. I concede there is another estimation method, but I think less accurate and more difficult to do.

And Harry N I pretty much have to disagree with your entire premise. ...I consider your advice to the same type of logic for the dumbing down of the public school system. Lets just make it easier and accept mediocre results cause its just for fun, instead of actually learning something useful in a "teachable moment".

 

[znomit]

You could knock up an LM317 constant current circuit for a few $ and test the LEDs at 1A. It doesn't add a great deal of cost or complexity though you'll need a few more volts from somewhere.
Then you could compare results with the rest of us, and with LEDs you buy next week/month/year.

 

[uplite]

You could knock up an LM317 constant current circuit for a few $ and test the LEDs at 1A. It doesn't add a great deal of cost or complexity though you'll need a few more volts from somewhere.
The you could compare results with the rest of us, and with LEDs you buy next week/month/year.

+1 :twothumbs

BaaB...please understand...no one is saying that you must perform any test that it is beyond your capabilities.

We are just saying that the high-current measurement which you proposed is basically useless for estimating Vf (see title of this thread).

Please don't take that as an insult. It is objective technical fact. I'm just trying to help here. :wave:

-Jeff

 

[uk_caver]

You could knock up an LM317 constant current circuit for a few $ and test the LEDs at 1A. It doesn't add a great deal of cost or complexity though you'll need a few more volts from somewhere.
Then you could compare results with the rest of us, and with LEDs you buy next week/month/year.

For quick Vf testing of LEDs bought in, I made up a unit with a AMC7135 driver, a clip to attach to a 3xAA holder, and output via a pair of croc clips - I can clip the output onto my meter tips, and then quickly dab my way across a load of LEDs.
That also means if I'm playing with LED/optic combinations, I have a simple CC source that just involves soldering a couple of wires to the LED to clip into, and which is portable for outdoor comparison shots, etc.

If I wanted to do measurements at more than just 350mA, I'd probably stick a few 7135 driver boards to a bit of stripboard, then wire them up so I could get 350mA/700mA/1050mA/2.8A or whatever by switching a jumper or two around on pins on the board, or if I wanted something slicker, maybe box it up and have a rotary switch. I'd run off a 4xAA holder for the higher currents.