Building an LED light/flash

mjkzz · Post by **mjkzz** » Thu Jan 30, 2020 2:40 am

Testing the Zhongyi 5X Mk II lens, as well as the CD-ROM drive focus stacking rail and the latest LED flash/lights.

Camera is Sony A7III in electronic shutter mode. Because of this, Sony A7III does not fire flash, so I have to fire the LED flash via controller and syncing the camera and LED flashes means longer LED duration even though shutter speed is 1/80s. This turns out that the LED flash is rather flexible -- adjustable duration

The Zhongyi 5X Mk II lens is set at f/4 and 2.5X magnficiation. Somehow I think the aperture is smaller (higher f number) than f/4.

Camera settings are 1/80s and ISO 100.

Here is the video

mjkzz · Post by **mjkzz** » Sun Feb 02, 2020 1:09 am

Added some feature requested by many. Continuous power level can be saved by push and hold the button for over 2 seconds, this value will be restored even after power off (it is saved to EEPROM).

Now, 3 levels, 1/4, 1/8, and 1/16 of full power, meaning if full power is 20W, output will be 5W, 2.5W, and 1.25W. The observed flickering will be fixed using a new microprocessor that is 8 times faster, hence increased PWM frequency.

Here Is the Video

viktor j nilsson · Post by **viktor j nilsson** » Fri Mar 06, 2020 4:53 am

mjkzz wrote:ah, you did :-)

I highly doubt XM-L2 would be any different from other LEDs, that said, LEDs are current device, as long as you do not push the current over too much and for prolonged period of time, it should work.

Older LED drivers are designed having normal lighting in mind and dimmable drivers use PWM. On the other hand, ZK-4KX adjusts output voltage with current limiting, no PWM.

Hello again,
I recieved the two ZK-4KX and hooked one up to my microscope illumination, which is a Cree XM-L2 attached to a aluminum rod with Arctic silver in-between.

Unfortunately, the ZK-4KX didn't solve all my problems.

Problem 1: too bright and flickering at lowest setting.

This problem was the same with the ZK-4KX and the BuckPuck. Below 0.6A or so, the LED started to fluctuate in intensity. Further reduction in current caused an abrupt switch to being very dim, and then go all black. Not the behavior I was hoping for! I guess this behavior is either inherent to the XM-L2, or I had a bad batch. I tried with three different LEDs (20mm star PCB) and all behaved this way.

Problem 2: not bright enough at highest setting.
Well, this was improved with the ZK-4KX. My old Buckpuck is the 1A version, so being able to crank up the power to 2A or 3A definitely helped. So that could be useful in some cases.

However, what I didn't like about the ZK-4KX was the interface. Changing the current requires, what, three clicks on two different buttons before you can even start twisting the pot. Actually, it's even more. One to switch to "change power mode", two clicks to get from Watt to Ampere, and two more clicks to go from changing in 0.001A increments to 0.1A increments. THEN twist the pot. I'm definitely not going to enjoy doing that every time I switch from 10x to 40x magnification, while looking through the eyeoieces.!

Actually, I realized I'm quite happy with the 1A Buckpuck in comparison, with it's single pot.

But I do think I need to get on with my project to build a beam splitter flash system.

Still useful things to have, these mini lab supplies!

Pau · Post by **Pau** » Fri Mar 06, 2020 5:33 am

Hi Viktor,

viktor j nilsson wrote:I recieved the two ZK-4KX and hooked one up to my microscope illumination, which is a Cree XM-L2 attached to a aluminum rod with Arctic silver in-between.

Unfortunately, the ZK-4KX didn't solve all my problems.

Problem 1: too bright and flickering at lowest setting.

This problem was the same with the ZK-4KX and the BuckPuck. Below 0.6A or so, the LED started to fluctuate in intensity. Further reduction in current caused an abrupt switch to being very dim, and then go all black. Not the behavior I was hoping for! I guess this behavior is either inherent to the XM-L2, or I had a bad batch. I tried with three different LEDs (20mm star PCB) and all behaved this way.

Problem 2: not bright enough at highest setting.
Well, this was improved with the ZK-4KX. My old Buckpuck is the 1A version, so being able to crank up the power to 2A or 3A definitely helped. So that could be useful in some cases.

...

But I do think I need to get on with my project to build a beam splitter flash system...

I also use the XM-L and XM-L2 in three of my microscope illumination setups but with the 2100mA BuckBlock driver. I have plenty of light even with a 50/50 beamspliltter (for flash) and DIC at 100X on sensor.
I also suffer your same first problem. It's pretty annoying and my trick to live with is a strong ND filter at the tray under the condenser and often for low power to maintain the lever of the trinoc to 70% of light to the camera all the time and the polarizer on the light path. Once used to it I find it not problematic and I prefer it over incandescent.
A log pot in place of a linear one improves a bit the light intensity regulation with the BuckBlock drivers.

Of course I would love that one of our Electronics savants could find and post a good and easy solution

viktor j nilsson · Post by **viktor j nilsson** » Fri Mar 06, 2020 5:48 am

Pau wrote:Hi Viktor,

I also use the XM-L and XM-L2 in three of my microscope illumination setups but with the 2100mA BuckBlock driver. I have plenty of light even with a 50/50 beamspliltter (for flash) and DIC at 100X on sensor.
I also suffer your same first problem. It's pretty annoying and my trick to live with is a strong ND filter at the tray under the condenser and often for low power to maintain the lever of the trinoc to 70% of light to the camera all the time and the polarizer on the light path. Once used to it I find it not problematic and I prefer it over incandescent.
A log pot in place of a linear one improves a bit the light intensity regulation with the BuckBlock drivers.

Of course I would love that one of our Electronics savants could find and post a good and easy solution

Yep, I have been thinking about getting the 2100mA BuckBlock. I got the BuckPuck when I was completely new to electronics and it seemed a good idea to get one with a suitable pot already in place.

I'm actually going to sell one of my scopes to a friend soon. I might take it as an opportunity to throw in the BuckPuck for him, and get me a BuckBlock instead (and a ND filter, I actually don't even own one).

How strong is your ND filter?

Pau · Post by **Pau** » Fri Mar 06, 2020 6:04 am

My ND filter has a 1/32 factor (sometimes labeled 0.03)

mawyatt · Post by **mawyatt** » Fri Mar 06, 2020 6:50 am

Pau wrote:Hi Viktor,
viktor j nilsson wrote:I recieved the two ZK-4KX and hooked one up to my microscope illumination, which is a Cree XM-L2 attached to a aluminum rod with Arctic silver in-between.

Unfortunately, the ZK-4KX didn't solve all my problems.

Problem 1: too bright and flickering at lowest setting.

This problem was the same with the ZK-4KX and the BuckPuck. Below 0.6A or so, the LED started to fluctuate in intensity. Further reduction in current caused an abrupt switch to being very dim, and then go all black. Not the behavior I was hoping for! I guess this behavior is either inherent to the XM-L2, or I had a bad batch. I tried with three different LEDs (20mm star PCB) and all behaved this way.

Problem 2: not bright enough at highest setting.
Well, this was improved with the ZK-4KX. My old Buckpuck is the 1A version, so being able to crank up the power to 2A or 3A definitely helped. So that could be useful in some cases.

...

But I do think I need to get on with my project to build a beam splitter flash system...
I also use the XM-L and XM-L2 in three of my microscope illumination setups but with the 2100mA BuckBlock driver. I have plenty of light even with a 50/50 beamspliltter (for flash) and DIC at 100X on sensor.
I also suffer your same first problem. It's pretty annoying and my trick to live with is a strong ND filter at the tray under the condenser and often for low power to maintain the lever of the trinoc to 70% of light to the camera all the time and the polarizer on the light path. Once used to it I find it not problematic and I prefer it over incandescent.
A log pot in place of a linear one improves a bit the light intensity regulation with the BuckBlock drivers.

Of course I would love that one of our Electronics savants could find and post a good and easy solution

Pau, Viktor,

These are some of the reasons we ventured into developing a full custom LED Current Mode Controller for Continuous and Flash for macro use.

https://www.photomacrography.net/forum/ ... sc&start=0

Unfortunately the LED output flash power, roughly equivalent to a typical speedlight, at reasonably fast speeds required too many LED modules that are somewhat costly. For continuous use these did not use pulse width modulation to control the LED current, they utilized a continuous current mode control loop, so no flickering. Also, they could use a multi-turn potentiometer with dial for precise stable light control since the LEDs are under constant current rather than voltage control. Even a RPi controlled version was under development for those wanting to integrate with a previously developed Trinamic Stepper Motor Focus Rail Controller.

With enough interest, we might reconsider reviving this abandoned development effort at Mike's Labs.

Best,

mjkzz · Post by **mjkzz** » Sat Mar 07, 2020 3:49 am

viktor j nilsson wrote:
mjkzz wrote:ah, you did :-)

I highly doubt XM-L2 would be any different from other LEDs, that said, LEDs are current device, as long as you do not push the current over too much and for prolonged period of time, it should work.

Older LED drivers are designed having normal lighting in mind and dimmable drivers use PWM. On the other hand, ZK-4KX adjusts output voltage with current limiting, no PWM.
Hello again,
I recieved the two ZK-4KX and hooked one up to my microscope illumination, which is a Cree XM-L2 attached to a aluminum rod with Arctic silver in-between.

Unfortunately, the ZK-4KX didn't solve all my problems.

Problem 1: too bright and flickering at lowest setting.

This problem was the same with the ZK-4KX and the BuckPuck. Below 0.6A or so, the LED started to fluctuate in intensity. Further reduction in current caused an abrupt switch to being very dim, and then go all black. Not the behavior I was hoping for! I guess this behavior is either inherent to the XM-L2, or I had a bad batch. I tried with three different LEDs (20mm star PCB) and all behaved this way.

Problem 2: not bright enough at highest setting.
Well, this was improved with the ZK-4KX. My old Buckpuck is the 1A version, so being able to crank up the power to 2A or 3A definitely helped. So that could be useful in some cases.

However, what I didn't like about the ZK-4KX was the interface. Changing the current requires, what, three clicks on two different buttons before you can even start twisting the pot. Actually, it's even more. One to switch to "change power mode", two clicks to get from Watt to Ampere, and two more clicks to go from changing in 0.001A increments to 0.1A increments. THEN twist the pot. I'm definitely not going to enjoy doing that every time I switch from 10x to 40x magnification, while looking through the eyeoieces.!

Actually, I realized I'm quite happy with the 1A Buckpuck in comparison, with it's single pot.

But I do think I need to get on with my project to build a beam splitter flash system.

Still useful things to have, these mini lab supplies!

Oh, wow. I do not remember having flicker issues with that thing. I am a bit busy now, but I will check my unit. Just got a camera capable of 1000 fps, maybe I will use that to check it.

Usually a large capacitor can smooth out flickers, I do not know what value to choose, I usually just pick one in my drawer. Do this only with a power supply with constant current regulation as a large cap will be seen as a short circuit for a short period of time. Make sure you get polarity right (if that type) and voltage spec right.

I also have a circuit some where on this forum, it was about voice coil thing, it is simple and easy to construct, and you can use a POT to adjust power output, that will solve your 2nd problem

mjkzz · Post by **mjkzz** » Sat Mar 07, 2020 4:49 am

actually, just did a quick test, even set at 0.1A.

I do not see any flickering, used high frame rate camera set at 1000fps, a Canon 6D II, a Sony A7 III, set shutter speed to different values. I do see flickering on my monitors.

Also changed to battery so that we do not have interference from AC main outlet (50 Hz in China, I think), same result. My power supply is rated 12V 5A, so it should have enough juice to power a 0.9W (9V at 0.1A) device.

Anyways, will present a video on various settings. Maybe it is just good for video work, ie, adjustable power levels for video.

mjkzz · Post by **mjkzz** » Sat Mar 07, 2020 5:33 am

this is set at 8.86V, about 0.1A, driving a 10W CREE LED . . . some small ripple exist, but I really do not see that will cause visible flickering . . . and this is powered by AC adapter rated 12V 5A (60W)

So really not sure what is going on here.

mawyatt · Post by **mawyatt** » Sat Mar 07, 2020 3:37 pm

LEDs are fundamentally semiconductor junctions and have a exponential relationship between current & voltage. You may need to measure/see the LED current, not the LED voltage, which might be contributing to the flicker. Place a small (~1 ohm ~1 watt) resistor in series with the LED and monitor the voltage drop across the resistor. You will need to increase the LED supply voltage to compensate for the resistor voltage drop to keep the same average LED current, and use no capacitor for averaging.

For higher resolution use the AC coupled input on the scope, this removes the DC or average current component and shows the dynamic or changing current value which is what may be causing the mentioned flicker.

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.

Best,

mjkzz · Post by **mjkzz** » Sat Mar 07, 2020 4:15 pm

However, what I didn't like about the ZK-4KX was the interface. Changing the current requires, what, three clicks on two different buttons before you can even start twisting the pot. Actually, it's even more. One to switch to "change power mode", two clicks to get from Watt to Ampere, and two more clicks to go from changing in 0.001A increments to 0.1A increments. THEN twist the pot. I'm definitely not going to enjoy doing that every time I switch from 10x to 40x magnification, while looking through the eyeoieces.!

Not sure why you need to change CC limit, you can just set it once, then just use the dial to change voltage, probably not proportional to power level, but you do have the ability to change it by simply turning the knob.

Anyways, glad you find it at least useful of some sort

mjkzz · Post by **mjkzz** » Sat Mar 07, 2020 4:41 pm

mawyatt wrote:LEDs are fundamentally semiconductor junctions and have a exponential relationship between current & voltage. You may need to measure/see the LED current, not the LED voltage, which might be contributing to the flicker. Place a small (~1 ohm ~1 watt) resistor in series with the LED and monitor the voltage drop across the resistor. You will need to increase the LED supply voltage to compensate for the resistor voltage drop to keep the same average LED current, and use no capacitor for averaging.

For higher resolution use the AC coupled input on the scope, this removes the DC or average current component and shows the dynamic or changing current value which is what may be causing the mentioned flicker.

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.

Best,

Yeah, I do not have a through hole 1 ohm resistor, only surface mount ones . . . it is hard to solder that thing. I used DC coupling because I want to show voltage level.

Not sure about 59mV can cause 10X current flow for LEDs . . ., in your Wyatt lab, use a lab power supply, start at maybe 8.5V on a 10W LED (I think any should do), slowly crank it up to 10V, do you get (1.5/0,059)*10x = 254X of current change?

I think it is probably true at some critical point, ie, at that point, a 59mV change can cause 10X change in current, but afterwards, it does not. And this might be what viktor observed (as well as I do) that at some point, it is very dim, but brighten up abruptly.

LEDs are current device for sure, however, I think it more or less behaves like a non-linear resistor after some critical point. Measuring voltage across it should be enough to gauge it. This is what I have observed, I am not in the field of LED developments.

mjkzz · Post by **mjkzz** » Sat Mar 07, 2020 5:02 pm

mawyatt wrote:LEDs are fundamentally semiconductor junctions and have a exponential relationship between current & voltage. You may need to measure/see the LED current, not the LED voltage, which might be contributing to the flicker. Place a small (~1 ohm ~1 watt) resistor in series with the LED and monitor the voltage drop across the resistor. You will need to increase the LED supply voltage to compensate for the resistor voltage drop to keep the same average LED current, and use no capacitor for averaging.

For higher resolution use the AC coupled input on the scope, this removes the DC or average current component and shows the dynamic or changing current value which is what may be causing the mentioned flicker.

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.

Best,

And the problem is, I do not see flickering at all . . . even at 1000 frames per second where it is very sensitive to flickering. A few years back, I was shooting 480fps video, I used 2 commercial 100W LEDs lights, the flickering was so bad, I had to put 4 4000uf capacitors across the LEDs, 2 each. At that time, these 100W leds do not have good CRI

Here is a video of it: https://youtu.be/V8X4h6jN-mQ

And here is a video using 4 50W LED lights: https://youtu.be/FVizxuqPkOU

Anyways, I will shoot a video about this device at different levels when I get time. I am puzzled

mawyatt · Post by **mawyatt** » Sat Mar 07, 2020 6:33 pm

mjkzz wrote:
mawyatt wrote:LEDs are fundamentally semiconductor junctions and have a exponential relationship between current & voltage. You may need to measure/see the LED current, not the LED voltage, which might be contributing to the flicker. Place a small (~1 ohm ~1 watt) resistor in series with the LED and monitor the voltage drop across the resistor. You will need to increase the LED supply voltage to compensate for the resistor voltage drop to keep the same average LED current, and use no capacitor for averaging.

For higher resolution use the AC coupled input on the scope, this removes the DC or average current component and shows the dynamic or changing current value which is what may be causing the mentioned flicker.

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.

Best,
Yeah, I do not have a through hole 1 ohm resistor, only surface mount ones . . . it is hard to solder that thing. I used DC coupling because I want to show voltage level.

Not sure about 59mV can cause 10X current flow for LEDs . . ., in your Wyatt lab, use a lab power supply, start at maybe 8.5V on a 10W LED (I think any should do), slowly crank it up to 10V, do you get (1.5/0,059)*10x = 254X of current change?

I think it is probably true at some critical point, ie, at that point, a 59mV change can cause 10X change in current, but afterwards, it does not. And this might be what viktor observed (as well as I do) that at some point, it is very dim, but brighten up abruptly.

LEDs are current device for sure, however, I think it more or less behaves like a non-linear resistor after some critical point. Measuring voltage across it should be enough to gauge it. This is what I have observed, I am not in the field of LED developments.

Doesn't have to be 1 ohm, just any small value should work. Simply place it in series with the LED wires or otherwise and use Ohm's law to calculate the current from the voltage drop. BTW I never purchased any high power LEDs since, as noted, it wasn't going to be practical or cost effective to create an effective LED Flash with similar Speed-light capability, so the effort was abandoned.

See reference below for notes on semiconductor junction physics. The standard pn junction (diode) equation states Id = Is(exp(Vd/Vt) -1), since Is is very small then Id ~ Is(exp(Vd/Vt)), where Vt = kT/q and k is Boltzmanns constant 1.38E-23 J/K, T is temp in Kelvin, and q is electron charge 1.9E-19 Coulombs, and Vt ~ 0.025875 Volts @ 300K

Id1 = Is(exp(Vd1/Vt) and Id2 =Is(exp(Vd2/Vt)

Id1/Id2 = exp((Vd1-Vd2)/Vt), Vd1-Vd2 = Vt(ln(Id1/Id2)

So for Id1/Id2 = 10 then Vd1-Vd2 ~ 59.579mv, for Id1/Id2 = 2 then Vd1-Vd2 - 17.935mv

Please note that this is independent of any specific diode parameters and temperature, thus my believe that an LED should show similar results. However (there's always a "however") this assumes an ideal pn junction diode and no series diode resistance. The ideality factor (see reference) likely is not unity for an LED, the temperature won't remain the same between the currents and there will always be some series resistance, in fact the series diode resistance is how they parallel LEDs to help equalize the individual LED currents.

https://ecee.colorado.edu/~bart/book/bo ... /ch4_4.htm

Anyway enough of the Solid State Physics, if you aren't seeing any flickering with your fast camera, power supply and LEDs, that's a good indication that your LED setup has a uniform output, at least short term.

Best,