Building an LED light/flash

[mawyatt]

Interesting videos!!

The Fast LED Strobe design we presented should have no trouble keeping up and synching with a 1000 frames per second (1ms) camera.

https://www.photomacrography.net/forum/ ... hp?t=40999

Best,

[mjkzz]

I am a little math phobia, I'd rather deal with physical things than those symbols And maybe, just maybe it is beneficial for you to get a power LED and experiment with it, it actually is a lot of fun and since you like to tinker.

Anyways, in certain operating regions, I think the LED behave "normally", ie, a little change in voltage does not make power consumption jump, you can experiment with one. Therefore, those small ripples would not cause sudden change in power output, significantly enough to cause flickering. For example, if I crank up voltage to 9.5V from 8.85V, current goes up from 0.1A to about 0.3A. Those ripples looks like less than 10mV.

Of course, these are probably AVERAGE current (read from that unit), actual current could be much higher for those ripples, thus flickering. But since I do not observe any flickering (it does not mean it is not there) You (Mike) have a high end scope, so it is a good idea to experiment it. My scope is a portable, battery operated one, only capable of about 1Mhz

I will shoot a video as soon as I finish some coding.

[mawyatt]

Therefore, those small ripples would not cause sudden change in power output, significantly enough to cause flickering. For example, if I crank up voltage to 9.5V from 8.85V, current goes up from 0.1A to about 0.3A. Those ripples looks like less than 10mV.

This equates to ~2 ohms total series LED resistance.

Of course, these are probably AVERAGE current (read from that unit), actual current could be much higher for those ripples, thus flickering. But since I do not observe any flickering (it does not mean it is not there) You (Mike) have a high end scope, so it is a good idea to experiment it. My scope is a portable, battery operated one, only capable of about 1Mhz.

Your scope should be fine at resolving the LED ripple current. As mentioned, just insert a small resistor in series with the LED and use the scope AC input mode which will display the ripple voltage across the resistor (ripple current = ripple voltage /resistance). Since the LED optical power output is somewhat proportional to LED current, you can surmise the effects of the ripple current on the LED output.

A good "seat of the pants" estimate of LED optical output variation in % due to ripple would be ~ 100((peak to peak LED ripple current)/(average LED current)).

Best,

[mawyatt]

I am a little math phobia, I'd rather deal with physical things than those symbols And maybe, just maybe it is beneficial for you to get a power LED and experiment with it, it actually is a lot of fun and since you like to tinker.

Agree it would be fun to play with the power LEDs when/if time permits. Developing the Fast LED Strobe provided an opportunity to study these devices and found that the most efficient devices produce 100~150 lumens per watt. The higher power COB modules support up to 10,000~15,000 lumens per module and require much higher voltage than the usual ~12 volts. This makes sense in that these modules have many LED chip clusters in series to help equalize the distribution of LED chip currents, so they are a bunch of LED chips in parallel and series to achieve the higher output power.

Anyway, the current mode control concepts presented would have no trouble controlling these high power COB LED modules in either continuous or pulsed flash modes, the problem ends up being how many COB modules & cost to duplicate a typical speedlight performance. Strobes are completely out of the question as they are ~10X more powerful than speedlights.

Producing an LED based Flash setup where the output requires ~25 10,000 lumen COB modules at $16 each for a 10ms flash pulse with the equivalent output of a typical speedlight just didn't seem worthwhile pursuing.

Best,

[mjkzz]

I am a little math phobia, I'd rather deal with physical things than those symbols And maybe, just maybe it is beneficial for you to get a power LED and experiment with it, it actually is a lot of fun and since you like to tinker.

Agree it would be fun to play with the power LEDs when/if time permits. Developing the Fast LED Strobe provided an opportunity to study these devices and found that the most efficient devices produce 100~150 lumens per watt. The higher power COB modules support up to 10,000~15,000 lumens per module and require much higher voltage than the usual ~12 volts. This makes sense in that these modules have many LED chip clusters in series to help equalize the distribution of LED chip currents, so they are a bunch of LED chips in parallel and series to achieve the higher output power.

Anyway, the current mode control concepts presented would have no trouble controlling these high power COB LED modules in either continuous or pulsed flash modes, the problem ends up being how many COB modules & cost to duplicate a typical speedlight performance. Strobes are completely out of the question as they are ~10X more powerful than speedlights.

Producing an LED based Flash setup where the output requires ~25 10,000 lumen COB modules at $16 each for a 10ms flash pulse with the equivalent output of a typical speedlight just didn't seem worthwhile pursuing.

Best,

I think we have different approaches toward things.

If not discussing theoretical thing, I like to get my hands on stuff, use it, and solve problems when I see them (like I smoothed out flickering with large caps to shoot those high frame rate videos)

On the other hands, Mike, you like to take things from simulation or theory side. Designing fascinating circuits, with amazing scope images captures, etc, etc. But, this is a big but, in the end, you do not even own a power LED, the device your circuit is designed for.

Even with this approach, do you have a model for LED? If you do, can you simulate the condition of 9V power supply with 10mV ripple and see if current changes dramatically, thus test your 59mV theory you had?

This is why I like to reply back to viktor, not because he got different results, but because he actually bought the thing, played with it, and drew a conclusion based on his experience with it, instead of speculating in theory. Even though he has different results, his words carry a lot of weight and respect from me.

As far as how useful LED flash is, I think it is up to people to decide. Why do people still use continuous light in photography? People see the needs of LED flash because they encountered, ie, hands on experienced, problems. Without hands on experience, sure, you can theorized whatever you want.

[mawyatt]

I am a little math phobia, I'd rather deal with physical things than those symbols And maybe, just maybe it is beneficial for you to get a power LED and experiment with it, it actually is a lot of fun and since you like to tinker.

Agree it would be fun to play with the power LEDs when/if time permits. Developing the Fast LED Strobe provided an opportunity to study these devices and found that the most efficient devices produce 100~150 lumens per watt. The higher power COB modules support up to 10,000~15,000 lumens per module and require much higher voltage than the usual ~12 volts. This makes sense in that these modules have many LED chip clusters in series to help equalize the distribution of LED chip currents, so they are a bunch of LED chips in parallel and series to achieve the higher output power.

Anyway, the current mode control concepts presented would have no trouble controlling these high power COB LED modules in either continuous or pulsed flash modes, the problem ends up being how many COB modules & cost to duplicate a typical speedlight performance. Strobes are completely out of the question as they are ~10X more powerful than speedlights.

Producing an LED based Flash setup where the output requires ~25 10,000 lumen COB modules at $16 each for a 10ms flash pulse with the equivalent output of a typical speedlight just didn't seem worthwhile pursuing.

Best,

I think we have different approaches toward things.

If not discussing theoretical thing, I like to get my hands on stuff, use it, and solve problems when I see them (like I smoothed out flickering with large caps to shoot those high frame rate videos)

On the other hands, Mike, you like to take things from simulation or theory side. Designing fascinating circuits, with amazing scope images captures, etc, etc. But, this is a big but, in the end, you do not even own a power LED, the device your circuit is designed for.

Even with this approach, do you have a model for LED? If you do, can you simulate the condition of 9V power supply with 10mV ripple and see if current changes dramatically, thus test your 59mV theory you had?

This is why I like to reply back to viktor, not because he got different results, but because he actually bought the thing, played with it, and drew a conclusion based on his experience with it, instead of speculating in theory. Even though he has different results, his words carry a lot of weight and respect from me.

As far as how useful LED flash is, I think it is up to people to decide. Why do people still use continuous light in photography? People see the needs of LED flash because they encountered, ie, hands on experienced, problems. Without hands on experience, sure, you can theorized whatever you want.

Certainly glad that Newton, Maxwell, Einstien, and NASA believe in a little theory! Heck how was NASA supposed to get "hands on" experience sending (gravational slingshot) Voyager billions of miles away and communicating (takes radio signals 17 hrs to reach earth):? Lots of theory involved with these folks and NASA missions.

I do have a few years of "hands on" experience tho, I'm old and started in electronics at the age of 9~10, had my 1st Heathkit Oscilloscope at 12 when I began playing with transistors. Learned a lot, but mostly learned there are many things that folks don't need to build and test for "hands on" experience to find the answers. Sometimes you can't get "hands on" like NASA Voyager milestones, with integrated circuit design, and I'm sure many others.

For example, if we put 10 volts across a 10000 ohm resistor then 1 millamp of current will flow, if I add a conventional silicon diode (1N4148) in series then the current will be ~0.9375ma at ~300C (BTW this is transcendental, no closed form solution exist for a diode in series with a resistor ). I don't need to build and test either of these I already know what the answer is. Same goes for the LED, I know that a small voltage variation (ripple) across an LED without any series resistance will cause the output to fluctuate because the current will fluctuate much more than you might think with a small voltage ripple based upon Ohms Law. This is called the dynamic conductance of a pn junction and is the derivative of junction current with respect to the applied junction voltage, simply Id/Vt. Another way of thinking about this is 1/conductance is impedance, and thus the dynamic impedance of the diode is Vt/Id, or just ~0.25 ohms at 0.1a and ~0.025 ohms at 1a

Adding single LEDs in series as in the COB devices helps, since you now have more junctions in series.

So the dynamic impedance (resistance) is what's at play here, and it's quite small for an LED and why you need to add series resistance to help "swamp" it out and make the LED more linear like a regular resistor when operated voltage mode where a small voltage change can cause a large current change.

For efficiency many LED drivers utilize a switch mode type control (DC to DC converter for example), which alternates energy storage between an inductance and capacitance. This mode of operation naturally creates an output voltage ripple between when the inductance is charging and when the capacitance is charging. So without additional filtering (like your added 4000uF caps) this ripple will modulate the LED current, and thus modulate the LED optical output. This modulation current will be higher than simple Ohms Law predicts, because of the dynamic impedance mentioned. With the added capacitance for filtering the LED is severely hampered for speed (you have to charge and discharge this enormous capacitance to turn on and off the LED), so not very good for flash use, but you need lots of capacitance to reduce the LED current ripple. All this is the very reason for the selection of linear current mode control without switch-mode or PWM for the Fast Flash LED Controller design mentioned.

Anyway, not all things need to be built and tested to figure out certain performance metrics. Physics, simulation and some math can be very beneficial at times, and save lots of time and $!!

Best,

[mjkzz]

I am not going to quote what you said, it is too long.

Please do not take my previous comment negatively, a lot of your talk are sound, professional, reflecting your experience. But sometimes, I feel like your comments are a bit far fetched, but backed with heavy reading references to make it look like legit.

Take the 59mV theory for example, do you know what are the conditions for that to occur? Can it occur in the operating region a LED is being driven normally? Since you do not have an LED to do hands on experiment, could you simulate that in your software? If it is the reason for flickering, I think you have solved an important issue in the industry.

But somehow, from my experiments, it is not the case, and in theory, your 59mV comment does not apply in this operating region, hence, I think it is far fetched. If you had more experience with LEDs, I think you probably would not threw the 59mV theory out.

Anyways, I will make a video in a few weeks, though I got extended holidays, I still have other stuff to do.

[mawyatt]

I am not going to quote what you said, it is too long.

Please do not take my previous comment negatively, a lot of your talk are sound, professional, reflecting your experience. But sometimes, I feel like your comments are a bit far fetched, but backed with heavy reading references to make it look like legit.

Actually my comments are related to experience, but more importantly directly related to the solid state physics involved, not some hearsay representations I or others might create. They may seem "far fetched" to someone not well versed in electronics, and more specifically a lack of detailed understanding of fundamental semiconductors. All this is taught in undergraduate electronics at the introductory level, so certainly not that "far fetched"

These discussions regarding the behavior of semiconductor pn junctions, be they silicon, germanium, or InGaN/GaN (LED) are directly related to the fundamental basics of most semiconductor devices that include pn junctions, which pretty much covers almost every semiconductor device! Your phone, computer, camera, speedlight, stepper motor controller, TV, radio, just about everything that has electronics rely on this behavior.

Take the 59mV theory for example, do you know what are the conditions for that to occur? Can it occur in the operating region a LED is being driven normally? Since you do not have an LED to do hands on experiment, could you simulate that in your software? If it is the reason for flickering, I think you have solved an important issue in the industry.

If you use Google a number of examples will show up on LED behavior & models. Here's a link to a simple LED SPICE model that includes the series resistance and non-ideality factor for an LED. The graphs revel ~4 orders of magnitude current change for a ~1 volt LED change in the middle operating current ranges with series resistance.

https://arxiv.org/pdf/1903.07538.pdf

But somehow, from my experiments, it is not the case, and in theory, your 59mV comment does not apply in this operating region, hence, I think it is far fetched. If you had more experience with LEDs, I think you probably would not threw the 59mV theory out.

BTW this isn't "my theory" as you seem to continuously misrepresent, but IN FACT the very theory of pn junctions attributed to Dr. William Shockley, the co-inventor of the bipolar transistor!!

https://en.wikipedia.org/wiki/Shockley_diode_equation

I don't know about the specifics of the LED COB modules you are using, but suspect that they are comprised from 3 or 4 LEDs in series with a resistor, then a group of these series LED resistors strings in parallel to form the complete LED COB Module. I recall an estimated equivalent series resistance of ~2 ohms from your measurements of the LED COB you are using, which seems reasonable.

Anyway, I would recommend you measure the LED COB Module current rather than voltage as mentioned, to get an idea of the optical output modulation due to ripple.

Edit: Just did a quick test of a single clear LED (COB modules are made up of many of these single LED die) which has a voltage drop of 2.6435 volts at 1ma with a 100.2mv/decade slope and series resistance of 41.6 ohms @ 1ma. This LED has a significant series resistance and still a large current change with a small voltage change, so confirms that a small voltage change can cause a significant current change in an Light Emitting DIODE (LED)!!

Best,

[mjkzz]

Mike, as we all know, you need certain voltage potential for a diode to conduct. At that critical point, typically 0.7V, from non-conducting to conducting, you have current change of infinity.

"your theory" means you threw out that thing in the discussion. I really doubt you are the inventor/discoverer of it. Even that, I do not think you know how to apply that, ie, under what conditions it occurs, when to expect to observe it. You just threw it out for no reason.

You keep throwing out stuff that is common knowledge, yet, you make them sound like something serious. On the other hand, you are "suspecting", "speculating" instead of contributing some facts. So far, what I get from you are empty words, far fetched theories.

I now have serious doubt that you used an LED model as load for the circuit you "designed" under simulation since you kept on evading my question about LED models.

This thread is about concrete stuff, with specific model number of devices, let me be polite and suggest you to get one of those devices, plot some chart with voltage and current, use your scope (if you really have one) and do some hands on experiment, make some concrete contribution. PLEASE.

So, let me be polite and stop responding to you.

[mawyatt]

Mike, as we all know, you need certain voltage potential for a diode to conduct. At that critical point, typically 0.7V, from non-conducting to conducting, you have current change of infinity.

"your theory" means you threw out that thing in the discussion. I really doubt you are the inventor/discoverer of it. Even that, I do not think you know how to apply that, ie, under what conditions it occurs, when to expect to observe it. You just threw it out for no reason.

You keep throwing out stuff that is common knowledge, yet, you make them sound like something serious. On the other hand, you are "suspecting", "speculating" instead of contributing some facts. So far, what I get from you are empty words, far fetched theories.

I now have serious doubt that you used an LED model as load for the circuit you "designed" under simulation since you kept on evading my question about LED models.

This thread is about concrete stuff, with specific model number of devices, let me be polite and suggest you to get one of those devices, plot some chart with voltage and current, use your scope (if you really have one) and do some hands on experiment, make some concrete contribution. PLEASE.

So, let me be polite and stop responding to you.

Peter,

Do you even read what you post!!

Lets start here.

"Mike, as we all know, you need certain voltage potential for a diode to conduct. At that critical point, typically 0.7V, from non-conducting to conducting, you have current change of infinity."

Nope, you got that totally wrong!! This is just a very old and very crude rule of thumb approximation for a silicon diode, for very simple "hands on" use. A diode conducts below 0.7 volts, has an exponential voltage current relationship, is temperature dependent and even conducts with reverse current with reverse voltage. Diodes made with different materials like silicon, germanium, LED (GaN and InGaN) for example, all have different characteristics, including forward voltage drop vs diode current. Did you even read the link to Shockley Diode Equation?

""your theory" means you threw out that thing in the discussion. I really doubt you are the inventor/discoverer of it.

Nope, you also got that totally wrong! It's not "my theory" but the fundament semiconductor diode equation from 1949, and I certainly didn't invent it! Again did you even read the link to the Shockley Diode Equation??

"Even that, I do not think you know how to apply that, ie, under what conditions it occurs, when to expect to observe it. You just threw it out for no reason."

Nope, you got that wrong as well!! I certainly know how to use, apply and observe this diode effect. It's baseline knowledge in elementary electronics!! Get an Introduction to Electronics text and do some reading, you might learn a heck of a lot, or maybe not since you seem to have difficulty understating some simple fundamental semiconductor basics!

"This thread is about concrete stuff, with specific model number of devices, let me be polite and suggest you to get one of those devices, plot some chart with voltage and current, use your scope (if you really have one) and do some hands on experiment, make some concrete contribution. PLEASE."

Now this is funny, so the diode equation isn't concrete enough for you I guess you are smarter that all of us, including Shockley since you know the diode equation isn't concrete!! BTW if you had read my post you might have noticed I included some data just measured on a LED, which is a Light Emitting DIODE. The LED current changed by 10X with a 100mv voltage change and the LED has an effective series resistance of 41.6 ohms. BTW how does your diode model of no conduction below 0.7 volts and fully conducting above 0.7 volts explain this LED with a 1ma forward voltage of 2.6435 volts!! My guess is you'll conjure up something like an LED isn't a real diode

"use your scope (if you really have one) and do some hands on experiment, make some concrete contribution. PLEASE."

BTW I have a pair of Tektronix 2465 scopes that were repaired and put into service, and just did a simple "hands on" experiment if you bother to read my previous post.

Think you've made the most important concrete contribution in this thread by reveling the true depth of your knowledge in electronics and semiconductors

"So, let me be polite and stop responding to you."

Please continue this is the most entertainment I've had in awhile.



Best,

Edit: Just did another quick "hands on" test of another single clear LED (COB modules are made up of many of these single LED die) which has a voltage drop of 2.6503 volts at 1ma with a 104.6mv/decade slope and series resistance of 45 ohms @ 1ma. This LED also has a significant series resistance and still a large current change with a small voltage change, again confirming that a small diode voltage change can cause a significant diode current change in an LED, just as Shockley predicted in 1949!!

[mjkzz]

OK, I am going to try to be peaceful :-)

Please do the following:

Put your COB in 200mA state, take a note on voltage, then increase it by 59mV, see what you got on current.

Maybe do a plot for every 50mA from start to, maybe 1000mA, since LEDs are current device.

Do you see any reason for those ripples to have significant effects?

[mjkzz]

Mike,

You wrote,

BTW an ideal forward biased silicon diode (or junction) has a 10X current change for a ~59mv voltage change @ 300K, suspect an ideal LED (no series resistance) has similar behavior.

OK, just re-read this, it looks like you are NOT one of those stubborn people who kept arguing an LED behaves just like a diode and kept applying theories about diode on LEDs. Sorry about stressful discussion (but you were entertained anyway) .

See, there are a lot people who kept arguing an LED is a diode just because of the D in the LED. Some even used similar arguement like the 59mV (yes, diodes, but not all LEDs, have that behavior) without knowing why it does not apply to an LED (but I think you know why, after re-read your comments). They pin-point things like flickering on it, too, just like you SUSPECT. However, when I ask them to do the similar exercise, plotting a graph, they do not know why, they just insist on the initial response of the LEDs (like what you observed).

Now a days, LEDs are not just purely current device anymore, as matter of fact, I just got some sample of 100W, high CRI leds that do not need CC driver to work and they are more or less like a voltage device -- you adjust output power by adjusting voltage across it, of course, not exceeding some limit. I can drive these LEDs with some bad switching power supply without flickering and very consistent output level.

Anyways, I think you know why viktor got flickering issue and I do not.

[mawyatt]

OK, I am going to try to be peaceful :-)

Please do the following:

Put your COB in 200mA state, take a note on voltage, then increase it by 59mV, see what you got on current.

Maybe do a plot for every 50mA from start to, maybe 1000mA, since LEDs are current device.

Do you see any reason for those ripples to have significant effects?

Peter,

Lets get some proper definitions straight.

First, a COB LED and a LED are not the same thing. Here's a quote from Silicon Lightworks:

What are Chip-on-Board (“COB”) LEDs?
Chip-on-Board or "COB" refers to the mounting of a bare LED chip in direct contact with a substrate (such as silicon carbide or sapphire) to produce LED arrays.

Source:
https://siliconlightworks.com/resoures/ ... e-cob-leds

So a COB LED is a Chip-on-Board with many LEDs, it's basically a module, not a single LED chip device, and why I often refer to them as COB LED Modules, not LEDs!

Second, an LED is a DIODE. No matter how to try and evade this, it's a DIODE is called a DIODE and behaves as a DIODE based upon the Shockley DIODE Equation. It has a series resistance and a ideality factor just like any diode, be it a silicon diode 1N4148, Schottky Diode 1N5818, Germanium diode 1N34, or LED Cree C503 Red, Blue, Amber for example.

Here's an exact quote from Wiki, only color, italic and bold added:

Power sources[edit]
Main article: LED power sources

The current in an LED or other diodes rises exponentially with the applied voltage (see Shockley diode equation), so a small change in voltage can cause a large change in current. Current through the LED must be regulated by an external circuit such as a constant current source to prevent damage. Since most common power supplies are (nearly) constant-voltage sources, LED fixtures must include a power converter, or at least a current-limiting resistor. In some applications, the internal resistance of small batteries is sufficient to keep current within the LED rating.


Source:
https://en.wikipedia.org/wiki/Light-emitting_diode[u][/u]

How can one even remotely dispute that an LED is not a DIODE. Please provide us with a reputable source that says an LED is NOT a DIODE and doesn't behave like a DIODE, we are all waiting!!

As I've stated earlier I don't have a COB LED Module (yet), I have the type LED device chips that make up the COB LED Module.

Here's what you quoted from my earlier post.

"BTW an ideal forward biased silicon diode (or junction) has a 10X current change for a ~59mv voltage change @ 300K, suspect an ideal LED (no series resistance) has similar behavior."

Please reread this very carefully, as you have gone off on a tangent and completely misrepresented this statement, as you have claiming it's "my theory". This refers to an ideal LED with no series resistance, nowhere does this refer to a COB LED Module, nor a module with series resistance. As you've just discovered from above a COB (Chip-on-Board) is an array of many LED chips arranged to produce a higher output, this arrangement involves parallel and series connections with resistors to help equalize the LED chip currents, thus certainly not a single LED device.


Anyway, we still strongly suggest that you measure the COB LED Module current rather than voltage to determine if ripple is causing COB Module optical output variation. As previously mentioned, hands on measuring the COB Module current only requires a single resistor of small value, surely you can accommodate this hands on measurement with your portable scope.

Best,

[mawyatt]

This thread is about concrete stuff, with specific model number of devices, let me be polite and suggest you to get one of those devices, plot some chart with voltage and current, use your scope (if you really have one) and do some hands on experiment, make some concrete contribution. PLEASE.

BTW, we do have a few scopes and 6 1/2 digit DVMs here at Mike's Labs, so lots of "hands on" here

Best,

[chris_ma]

I guess it‘s fair to say that any complex engineering task need lots of theory on and lots of hands on.

Simple problems can be solved with theory alone, or hands on alone, but as soon as it gets really complex either approach alone will usually fail and only the combination of both will work.

I totally agree that voyager or the moon landing were incredibly well calculated and I‘m still amazed that they managed to pull it off, but they also did lots and lots of testing.

Even today with an extremely deep understanding of physics and massive computing power, space one etc still needs to do launch tests.

On the other hand, without the necessary theory don‘t even bother to try to launch a rocket :)