Maker Pro
Maker Pro

Best LED driver for bright, fast strobe

mvo

May 13, 2020
5
Joined
May 13, 2020
Messages
5
I am a retired scientist doing photophysics experiments in a home lab. I can do the basic stuff required to customize instruments, program microcontrollers, and build digital circuits but lack the EE skills to design anything complicated. I want to build a one-off device that will allow me to pulse high-current LEDs (for example, those in the Luminus Devices Phlatlight-121 series). These LEDs operate at a few volts and are rated for continuous currents up to 30A. I want to create ~100 microsecond pulses at maximum current. I would like a reasonably square wave form. Pulse frequency does not have to be particularly high: 100 Hz would be okay. I am building my own strobe rather than buying one partly to save money but also because I want complete, software-level control over the pulse train.

I am wary of using a switching power supply because of concerns about power-quality and not knowing what will happen if I put a bunch of MOSFETs downstream from a device that is already full of transistors, inductors, capacitors, and feedback loops. Hence, my plan is to use an old-fashioned, heavy-duty linear power supply. For example, I have one that can produce 37A continuous/50A peak at 15 V, although the voltage-regulation range is narrow (11-15 V) so I will need to step-down the DC voltage if I use it.

Now to the question. I need to put something I can regulate with TTL output from a microcontroller between the power supply and the LED. I find the world of LED drivers bewildering. What kind of driver do I need for my application? Please be as specific as possible: companies like Mouser sell dozens of different ones with high current ratings, but none seems a natural fit for my project since they are mostly designed to implement complicated dimming schemes. Or, is there a better, relatively simple way to achieve my goal? Thanks in advance.
 

Harald Kapp

Moderator
Moderator
Nov 17, 2011
13,700
Joined
Nov 17, 2011
Messages
13,700
I will need to step-down the DC voltage if I use it.
But: that will introduce the switching regulator you want to avoid.
need to put something I can regulate with TTL output from a microcontroller between the power supply and the LED.
Regulate in which way? TTL is on/off, no analog control. Are you thinking of using pwm to control the brightness of the LEDs? Or simply turn the LEDs on and off?
none seems a natural fit for my project since they are mostly designed to implement complicated dimming schemes.
A more detailed description of what you want to achieve will help us help you.
 

mvo

May 13, 2020
5
Joined
May 13, 2020
Messages
5
Thanks for the quick reply. It has already helped me think more clearly about this project. I will reply to your comments and questions in reverse order:

3. What do I want to achieve?

I want to do strobe photography of moving objects: dark space, open shutter, series of flashes on the time scale specified. My inspiration for the approach I outlined came from looking at an MIT Master’s thesis posted online (https://dspace.mit.edu/bitstream/handle/1721.1/85515/871037927-MIT.pdf?sequence=2). The student did just what I want to do. I attach a block diagram of his setup as a jpg file, but it is easy enough to describe. He used a power supply of the sort I described. Its output was attached in parallel with three “V-Tech Power Boards.” Each power board was attached to one LED (one red, one green, one blue). When a simple microcontroller provided a TTL signal to one of the power boards, it drove the LED attached to it as long as the TTL signal stayed high. The student reported beautiful, square, 100-microsecond waveforms, in contrast to the really ugly ones produced by arc-lamp flashes. These waveforms appeal to me because they will make it easier to model the systems I plan to study, particular if I am illuminating fluorescent or phosphorescent objects. I do not really need the three-color capability, but it would be useful in some circumstances.

The V-Tech Power Boards are not described in any detail. V-Tech seems only briefly to have been in the LED-driver business a long time ago. The student does mention that the boards support flash frequencies as high as 4 kHz, which is much higher than needed for either his applications or mine.

2. Do I need brightness control?

No. I am going to want all the light I can get.

3. DC step-down issues.

You are right, of course. I can find an old-fashioned linear power supply that can provide the needed voltage and current. Or perhaps I can get by with a switching power supply. The role of the power boards must be to respond to a sudden change in load faster than the power supply can. From the simple physics of the situation, I presume these boards contain capacitors that can deliver a lot of current quickly without discharging appreciably. If I have that right, perhaps the quality of the power provided to them has little effect on the LED waveforms.

Again, thanks.
 

Attachments

  • 2020-05-14 08_41_03-Window.jpg
    2020-05-14 08_41_03-Window.jpg
    24.4 KB · Views: 11

Bluejets

Oct 5, 2014
6,901
Joined
Oct 5, 2014
Messages
6,901
Perhaps think along the lines of using Lipo cells.
3 cell pack will fall within your required voltage level.
The higher C rating designed for high current discharge.
 
  • Like
Reactions: mvo

Harald Kapp

Moderator
Moderator
Nov 17, 2011
13,700
Joined
Nov 17, 2011
Messages
13,700
For this application there is no need to worry about switching regulators. They will be perfect for the job.
You'll find that there are many ways to realize circuits that do what you want.
Before you start with a highly engineered, possibly over-engineered, current controlled circuit with all the bells and whistles one can imagine, why not start with a rather simple setup, see how it works and improve if necessary?
A very simple LED driver can look like this circuit:
upload_2020-5-15_7-23-2.png
The elements here are:
V1 = your power source
C1 = a buffer capacitor for high pulse current
R1 = current limiting resistor
D1... Dn = a number of LEDs of your choice
M1 = a power MOSFET with logic level gate drive (low threshold voltage) to be controlled by your microcontroller
R2 = a current limiting resistor to protect the output of the microcontroller from too high current.

You need V1 to be greater than the sum of the pass voltages of LEDs 1...n.
Calculate R as R = V1 - (n × LED pass voltage) / Iled
where Iled is the peak LED current you are going to use.
Select M1 such that Vds max >= V1 and Ids max >= Iled(peak). Note that you will need a logic level MOSFET if you want to drive it directly from a microcontroller.

If your power source V1 does not supply enough voltage to drive a string of LEDs (V1 < n × LED pass voltage), you can split the string into shorter strings and parallel them like so (just an example, you get the gist):
upload_2020-5-15_7-33-19.png
Note that each LED string will require its own resistor. You cannnot simply connect LEDs in parallel, that will give you uneven lighting at best, destroy the LEDs at worst.

If the light output from this circuit should not be crisp enough for your purposes, there are ways to accelerate the edges of the current signal using additional circuitry.
 
  • Like
Reactions: mvo

mvo

May 13, 2020
5
Joined
May 13, 2020
Messages
5
Perhaps think along the lines of using Lipo cells.
3 cell pack will fall within your required voltage level.
The higher C rating designed for high current discharge.
Thanks for the tip. I had no idea that there were lithium batteries that can deliver this kind of current. They seem mostly to be used in drones, electric cars, and so forth--not an area I have any experience with. I will buy some along with a suitable charger and try them out. I can think of a lot of potential applications beyond the one discussed in this post.
 

mvo

May 13, 2020
5
Joined
May 13, 2020
Messages
5
For this application there is no need to worry about switching regulators. They will be perfect for the job.
You'll find that there are many ways to realize circuits that do what you want.
Before you start with a highly engineered, possibly over-engineered, current controlled circuit with all the bells and whistles one can imagine, why not start with a rather simple setup, see how it works and improve if necessary?
A very simple LED river can look like this circuit:
View attachment 48213
The elements here are:
V1 = your power source
C1 = a buffer capacitor for high puls ecurrent
R1 = current limiting resistor
D1... Dn = a number of LEDs of your choice
M1 = a power MOSFET with logic level gate drive (low threshold voltage) to be controlled by your microcontroller
R2 = a current limiting resistor to protect the output of the microcontroller from too high current.

You need V1 to be greater than the sum of the pass voltages of LEDs 1...n.
Calculate R as R = V1 - (n × LED pass voltage) / Iled
where Iled is the peak LED current you are going to use.
Select M1 such that Vds max >= V1 and Ids max >= Iled(peak). Note that you will need a logic level MOSFET if you want to drive it directly from a microcontroller.

If your power source V1 does not supply nough voltage to drive a string of LEDs (V1 < n × LED pass voltage), you can split the string into shorter strings and parallel them like so (just an example, you get the gist):
View attachment 48216
Note that each LED string will require its own resistor. You cannnot simply connect LEDs in parallel, that will give you uneven lighting at best, destroy the LEDs at worst.

If the light output from this circuit should not be crisp enough for your purposes, there are ways to accelerate the edges of the current signal using additional circuitry.


Thanks so much for this carefully crafted reply. It is just at my level. I will order a few components and report back once I have some results. Interacting with this forum has been good for me. I was overly intimidated by the "over-engineered" model I was trying to follow. After all, who I am to second guess an EE graduate student at MIT? I should have been able to figure out how to do this myself. Next time I will make a bigger effort to do so.
 

Bluejets

Oct 5, 2014
6,901
Joined
Oct 5, 2014
Messages
6,901
As a side note for high current resistance sources, I have in the past used strips of stainless steel in varying lengths, thickness and width.
Typical application (model glider winch) was to control the power output of a C39 auto starter motor in 3 stages via a foot switch with the various resistance legs selected via microswitch on the foot pedal to starter solenoids, each bridging out the former until full honk was needed.

Might need a bit of trial and error to cut strips to the correct size but using a test circuit and a tong meter shouldn't be difficult.

Yes, some of those Lipo's poke out quite some grunt.
Around 1:00 in the video below will give you a bit of an idea from the telemetry feedback.
 

mvo

May 13, 2020
5
Joined
May 13, 2020
Messages
5
Success!

I took all the advice from Harald Kapp and Bluejets and got my strobe working. Kapp was absolutely right--no need for an over-engineered solution and Bluejets introduced me to LiPo batteries, which were a good solution for this project and will also be useful in other high-current settings. It was the high currents that initially intimidated me about improvising my way to an LED-strobe circuit. The only thing I ever put 50 A through before was the starter motor on my Tundra pickup.

Here is a schematic of the circuit I ended up using:

2020-07-11 09_18_38-Strobe1 _ Scheme-it.png
I collected data with V = 7.8 V from a two-cell LiPo battery and R1 = 0.0167-0.025 Ω (two or three 0.05Ω/2W resistors in parallel). R2 just pulls down the MOSFET drain when the the MOSFET (Nexperia PSMN4R2-60PL N-channel 60 V, 3.9 mΩ logic level MOSFET) is off: by allowing a few mA to flow through the LED (a blue OSRAM B P3W 01 high luminance LED for projection applications rated for 30 A continuous, 48 A pulsed application) when the MOSFET was off I could monitor the voltage of the drain with a cheap oscilloscope. R4 just pulls down the control pin when it goes low. Both R2 and R4 can be removed without affecting the circuit's function.

Here is a screenshot of a digital oscilloscope recording: the pulses are clean square waves.
2020-07-01 15_52_22-Window.png

The image shows a voltage drop of 1 V across R1 when its value was 0.025 Ω. Due to the internal resistance of the LiPo battery, the applied voltage was 6.5 V when the MOSFET was on (down from 7.8 V when it was off). Hence, the current was ~40 A. I was operating the strobe at 10 Hz with 1 ms pulses. I tested the circuit at 60 Hz with 200 μs pulses and currents of 50 A with comparable results. As Kapp indicated, a cheap switching power supply that can provide 60 A at 12 V also works fine. The oscilloscope traces are just not as pretty since the power supply's output is a bit squirrely. I could not clean it up with a capacitor across the outup. There is obviously a large built-in capacitor since the PS keeps putting out ampere-level current for several seconds after it is unplugged.

Finally, the system works as a DIY strobe. Here is a photo of a ball drop:

DSG_0398-1-4.jpg


So, thanks again. This has been the perfect coronavirus project.
 
Top