PWM drivers on a video light???

[herculino]

Hi all,

After a lot of reading and searching trough the forum I cannot find information to solve my doubts.

I'm on the long way to design a couple or three lights for my underwater photo/video camera.

I've been looking around whats in the market and found that : http://www.uwimaging.com/sola1200d.html

They look really nice and all I've read about them is that they work great. The only point is the price :shakehead too much for me...

So I'll try to build something based on it's main characteristics, let's see if I'm able to cope with this.:duh2:

I'm compiling as much information as I can and I found that article about SOLA 1200 and Greenforce LED 1500:http://www.deepshots.co.uk/2011/02/...al-lm-sola-1200-vs-greenforce-squid-led-1500/

Now I'm full of doubts about the electronic, I was unable to find resources to build my own drivers so I was considering the ones built buy taskled or something similar.

My main concern is if there will be any trouble like the one that Greenforce seems to have with this drivers. Has any one try them with video lights? Did you find any trouble like this? Is there a "recomended/tested" driver for this application?

Thanks for the help, hope to be able to contribute as I learn a bit more about this world.

 

[Packhorse]

I am quite surprised that GF would release a light with such a problem. I guess it comes down to lack of in field testing.
As the author points out the solution seems to be to simply increase the PWM frequency. It would be interesting to know what a good frequency and a bad one is.

Its not something I have noticed in my lights but I have not used PWM in my video/ camera lights. Instead I have used buck drivers in the way of AMC 7135 "smart resistors".

I am in the process of building another video light and was considering a PWM driver. I may have to reconsider or at least do some testing.

 

[jspeybro]

the choice of driver depends a lot on which LED(s) you want to use. each LED has its own characteristics.
The trouble (@packhorse, it can be seen in the video in the second link of the OP) is that one of the frequencies for the of the readout circuitry of the sensor is similar to the frequency of the PWM of the driver. The one from that greenforce is probably too low. Unfortunately, the frequency of the PWM is usually not given, but it is possible to measure it with an oscilloscope. I'm familiar with this issue but only for linescan camera's so I'm affraid I can't give good guidelines for which frequencies are recommended for canon.
Typically, the higher the PWM frequency, the better, but the more expensive/complex the driver. I would think one would need something around 100 KHz or more to get a stable image, but I haven't tested it. Unfortunately I don't have the time right now to try and measure some things.
The second link mentions that the effect is worse around 1/1000 of a second shutter speed. This could be an indication that the PWM of the driver only uses a frequency in the KHz range.

do you already have an idea on which LED(s) you want to use?

Johan

 

[herculino]

Hi all!!!

Johan, I was quite sure about the election of the leds till I kept reading and learning about it:shakehead

I was planning not to use any kind of reflector or may be just something in the sola 1200 fashion; just a piece of aluminum with polished counterbores, more thinking on aesthetic than any other thing.

The leds were supposed to be cree XP-E that rates about 200 lumens per led at 1A so that 6 of them will give the expected 1200 lumens.

Then I read this post form wquiles: http://www.candlepowerforums.com/vb...or-Javier-in-Spain-with-3xMC-E-and-2xIMR26500 And realized that may be my approach is not good as I will have 6 different beams I need to align somehow

As what I'm looking for is a wide UNIFORM pattern of light, may be I need to think a little more about optics to "align" the beams in a way that I have not a hig lux spot but a uniform beam.

That's why I'm considering now cree XM-L and use just 3 of them driven at lower amps and may be using a diffuser so that I produce more than 1800 lumens before the diffuser than can end up in something near the 1200 objective.

Or may be it would be better to use a sst-50 or sst-90 for an easier approach, what I don't like on this leds is the low efficient in compare with cree leds. They need a bunch of battery power to give you the planned 90-100 min of running time.

WHAT A MESS!!! The more I leard/read the more confused I get. Please a bit of help would be really appreciated.

Best regards from Cerdeña.

 

[Packhorse]

Then I read this post form wquiles: http://www.candlepowerforums.com/vb/...and-2xIMR26500 And realized that may be my approach is not good as I will have 6 different beams I need to align somehow

I dont think its that much of an issue. I have 16 XP-G's in a light with no optics and the light blends quite nicely into each other.

A few simple rules help I guess. Like not having the LED next to something that will cast a shadow and keep the LED's as close to each other as practical.

 

[herculino]

You should know better Packhorse, as I know you have made/mod many candles...

Any way I will try to run a simulation on Light Tools (if I'm able to find out how) before going to the buy an try way.

I'm trying to build the cheaper I can as my wife don't let me spend more money in my "silly" hobbies...:shakehead So I need to do it well at first attempt.

Thank for your help (in this and many other links)

Best regards from Zaragoza

 

[MikeAusC]

Mains powered room lights "flicker" at 100 or 120 Hz, but video cameras cope.

Maybe the PWM should be set at a multiple of 100 or 120 Hz - or use big capacitors on the output to ensure the LED gets DC.

 

[herculino]

You are right Mike, I'm reading Power Supplies for LED Drivers from Steve Winder but it's a bit too much for me. He talks about a "filter" made by a capacitor and an inductor... but it's a bit too much for me.

I was planning to build my own drivers but may be I must give up in my research as I cannot find any PCB any where.

I'm sure that George's drivers are great but the 40 bucks each its a bit expensive for my plans. May be I'm lucky and can use one of this KD or DealExtrem cheap ones, not sure about this point yet.

Keep studying.

 

[wquiles]

Mains powered room lights "flicker" at 100 or 120 Hz, but video cameras cope.

Maybe the PWM should be set at a multiple of 100 or 120 Hz - or use big capacitors on the output to ensure the LED gets DC.

You can't filter the output of a PWM for LED's to get DC - that is the whole point about PWM. It makes the output's current go from zero to its max value (say, 3 Amps), with the variable being the duty cycle from 0 (OFF) to 100% (always ON). If you filtered the output to get a steady DC current value, you would have no dimming - it would be a constant output.

To solve the problem, you either:
- use a constant current LED driver - no dimming (just ON or OFF)
- use a constant current output, but dim it with PWM using an ever higher PWM frequency than the camera (250-500Hz) but not high enough that would prevent the LED from turning ON/OFF per the duty cycle, or
- use a variable output LED current regulator design that does not depend on the use of PWM - the output is constant at each setting, and the variable output is achieved with some other method besides PWM (for example, a different user interface, button press, selector ring, etc..)

Will

 

[lucca brassi]

+1
-or swich simply between a few drivers with fixed PWM or current

 

[MikeAusC]

You can't filter the output of a PWM for LED's to get DC - that is the whole point about PWM. It makes the output's current go from zero to its max value (say, 3 Amps), with the variable being the duty cycle from 0 (OFF) to 100% (always ON). If you filtered the output to get a steady DC current value, you would have no dimming - it would be a constant output. . . . .

Then how does a Switchmode regulator produce a variable power dc output ?

The only difference is that PWM of the switchmode is at a lower frequency.

The capacitor on the output just averages the voltage from the PWM duty cycle.

Of course you have to make sure the capacitor is outside any feedback loop.

 

[herculino]

The more I read the less I know!!

It seems that I'll have to go to something already made and test for driving my lights. No a buck saved in this part!!

May anyone recommend me a driver for XM-L and XP-E leds? I'm still hesitating which ones to use:

3 XM-L will give me about 2700 lumens, if I use a diffuser may be I'll have a more uniform light beam with near the 1200 lumen I'm looking for. This leds are supposed to be the more efficient ones up to day.

6 XP-E will give me about 1300 lumens, and I'm not sure about the beam I'll get with them. This leds are less efficient than XM-L but I think it will be much easier to find a nice/cheap driver for them as are taking just 1A.

Please give your opinions about this two options and if possible candidates for driving them.

Thanks a lot for your help.

Best regards from Zaragoza

 

[Packhorse]

I use AMC 7135 drivers if you can closely match battery voltage to LED Vf they are great.

Not sure what advantage XP-E's have over XR-E's ( tighter focus with no optics 90deg vs 130deg meaning higher lux even if lower lumen) or XP-G's ( more efficient).

 

[wquiles]

Then how does a Switchmode regulator produce a variable power dc output ?

The only difference is that PWM of the switchmode is at a lower frequency.

The capacitor on the output just averages the voltage from the PWM duty cycle.

Of course you have to make sure the capacitor is outside any feedback loop.

I mean no disrespect, but No, that is comparing apples and oranges.

1) PWM on the output - the signal reaching the LED
When we talk about a PWM modulation on the output to the LED, the PWM is actually cutting the output off completely - it goes to OFF, and then based on the PWM duty cycle, the output goes to 100%. You are literally turning the LED fully OFF and then fully ON - the current goes from whatever the DC value is, to zero. Since the human eye can't react that quickly, PWM modulation allows dimming of an LED as the human eye only sees the "average" time it is ON. You can't NOT put a capacitor in this output to try to negate the switching taking place (so that it would not be seen by a recording camera), as it would then to create an average value where the LED will be basically ON all of the time - by putting a capacitor you loose the ability to use PWM for dimming. The frequencies that we are talking here for PWM on the output signal are in the 250 to 500 Hz range, which are meant to be high enough so that the human eye can't see the switching going on, but as the OP noted, that might not be fast enough for video cameras that have a higher sampling rate.


2) A switching current regulator has an internal loop where it can run from 10KHz to 100's of KHz, and some of the more newer ones in the MHz range. This type of regulator has a sensing element (very small resistor in series with the output) where the control IC (integrated circuit or "chip") would adjust/tweak the output current via a feedback loop which as I said above is running "very" fast, compared to the plain PWM. Due to this switching frequency, the output current has some ripples (at the base and at harmonics of the switching frequency), so you typically do have some filtering element at the output to smooth out the current to that it looks more like DC - you are basically trying to filter out the AC component - the ripple. This internal switching frequency is NOT the same as the PWM frequency as seeing by a video camera - the video camera can't sample fast enough to notice the very high frequency ripples from the internal switching frequency.

Here is a good link that talks more about switching power supplies in general:
http://en.wikipedia.org/wiki/Switched-mode_power_supply

Here is a very simple LED current regulator, using discrete components that works in the KHz range:
http://sound.westhost.com/appnotes/an003.htm

Here are some examples of switching LED driver IC's from TI (these run at a fixed 600KHz internally):
http://www.ti.com/product/tps61161



Now, what makes this problem with the video camera more interesting, is that many of these LED drivers can also support an analog or PWM input control signal to provide dimming. Here the LED driver is dimming the output by following the input control signal. For example, you can read here about the H6CC - which is a switching-type, LED current regulator, that has an external PWM input signal:
http://taskled.com/techh6cc.html


I hope this helps clarify the subject.

Will

 

[herculino]

Hi all!!

Thanks for the explanations Will. Now I understand better what I was reading in my book about drivers. The author refers to the ripple produced by the buck driver chip, the comparator if I can call it this way. I didn't got the idea since now. Thanks

I was reading the technical notes for all Taskled drivers and H6CC looks an interesting option as I can use a 10K analog potentiometer to dim the output current. Now I'm thinking how I can install this potentiometer in a dive light with no leaks.

About AMC7135 based drivers...it sound well till you think in dimming the leds to extend battery life. If I'm not wrong this drivers "spend" the battery at the same rate despite of dimming leds, I mean, the power that doesn't goes to the leds leaks like heat in the AMC's. SO if you dimm you get hotter but not more durable.

Sorry for my bad English, I don't know how to express better the idea.

I just got an idea for controlling the output of H6CC:
What about say 3 reeds attached to 3 different values to switch the output? May be in a ring with a magnet and so you rotate to the different values. Will the reeds handle the amps well or you need something more complex?

About the leds....I'm getting :duh2:. I think that 115º of the XP-E would be more interesting that 90º of the XR-E for video application. XP-G is like half a XM-L in lummens and amps so...what's the advantage? May be using 6 XP-G instead of 3 XM-L will through a more uniform beam? In that case I'll choose them.

What a mess... And I was thinking that I was close to the final design:sick2:

Thanks again for your comments, I appreciate them so much.

 

[georges80]

If you don't want video (or fast shutter camera) artefacts showing up you really want a constant current driver that performs dimming by varying the output current (versus PWM dimming). Something like the h6cc with 3 resistors and 3 switches (reeds or whatever) will provide 4 levels of output assuming each is individually turned on. Each level would be constant current and current regulated with no PWM flicker issues.

Now, we seem to also have some confusion about how the PWM is performed by the switcher IC or via the external controller. They are two different 'animals'.

The PWM by the switcher IC is a) analog and b) very high speed (often in the MHz range).

The switcher IC is usually configured to run the inductor in continuous mode, such that the current through the inductor never drops to zero. With the high speed switching, continuous current flow and output filtering the result is essentially DC with some ripple. Given the high operating frequency even 1/1000 sec shutter speeds won't "see" the ripple.

With external PWM, the base frequencies are typically much lower (a few hundred Hz) due to running in digital fashion and outside switcher IC control loop. With say 256 discrete PWM steps (to give a wide dynamic range) you end up operating the external PWM hardware at 256 x base frequency. There is an upper limit to this external PWM frequency since one must choose a minimum pulse width (1 / (256 x base frequency)) that the driver will still respond to (i.e. turn LED off and back on). Even at a relatively slow 200Hz the shortest pulse would be 1 / (256 x 200) = 20uSec. Often a driver will not be able to shutdown the output current in 20uSec and so either a slower frequency must be used or the dimmest levels (1/256 etc) can't be used.

The above is just a simplistic overview of some of the tradeoffs that go into designing a driver and dealing with external PWM.

My personal preference is to dim via current regulation (lower the output current) versus PWM dimming since it is more efficient (if long runtimes are important) and also reduces all flicker artefacts. That said, I do have some PWM based lights (torches/flashlights) and for those uses the simpler implementation of the PWM hardware makes a lot of sense.

In summary, we have choices of what driver and driver topology to choose, we need to make the choice based on meeting our own specific needs for each light.

cheers,
george.

 

[hanachan]

interesting issue.

this issue (flickering) may depend on not only the PWM frequency but also the digital video camera's frame rate & its shutter speed.

i have experimented. with my DIY two lights (SST-50 & 3xXML) and 4 digital camera movie (G10-canon, TZ7-panasonic, TX5-sony and GF1-panasonic).

my DIY SST-50 light has 5kHz PWM and a 3XML light 1kHz PWM frequency.
frame rate is, G10:24fps, TZ7:30, TX5:about 30, GF1:60 but output is 30.
and GF1 can record by 1)video mode and 2) pushing video button at P, A, S mode.

result:

flickering strength:
light--------- SST50(5KHzPWM) -----3XML(1KHzPWM) duty ratio:about 20%
camera
G10---------------(-)------------------(-)
TZ7---------------(-)------------------(+)
TX5---------------(-)------------------(++)
GF1(video)---------(-)------------------(++)
GF1(Pmode)--------(-)------------------(++++)

the 5KHzPWM produced no flickering.
the 1KHzPWM produced significant flickering except for G10.
flicker was observed at higher duty ratio.

IMO the low frame rate & slow shutter speed decreases the possibility of flickering.

herculino,
if you can DIY, i recommend
PWM AMC7135 about 5Khz (a bit lower frequency may be OK)
or vary AMC7135 VDD voltage (AMC7135 can limit current not only PWM
but also VDD voltage change; my experiment).

anyway you may have to test the driver (light) with your own video camera.

later i would like to load up the video.

 

[jspeybro]

I just got an idea for controlling the output of H6CC:
What about say 3 reeds attached to 3 different values to switch the output? May be in a ring with a magnet and so you rotate to the different values. Will the reeds handle the amps well or you need something more complex?

I'm exploring this idea as well. I made the circuit but wasn't able to test it yet because my battery pack is not yet assembled and 2 18650's cannot provide the current I want to use.
I'm using 2 FETs and 2 read contacts. Depending on which read contact is closed, there will be a certain resistor connected to the potentionmeter input of the H6CC giving me 2 modes, which is enough for my needs. The FETs also switch the power to the driver. I'm using an SST-50 for my light, but that's because the XML was not yet available when I was buying my components. I might swicht to the XML later if I'm not happy with the beam, but my light is not a video light.

Johan

 

[tplas]

------ Is there a "recomended/tested" driver for this application?------[ post 1]

I know the problem.
These are good :http://www.shiningbeam.com/servlet/the-132/3-dsh-Mode-Regulated-Circuit-Board/Detail
and these :http://kaidomain.com/ProductDetails.aspx?ProductId=10995
I use them both, no waves or flickering in the images on low , mid and high.

 

[georges80]

------ Is there a "recomended/tested" driver for this application?------[ post 1]

I know the problem.
These are good :http://www.shiningbeam.com/servlet/the-132/3-dsh-Mode-Regulated-Circuit-Board/Detail
and these :http://kaidomain.com/ProductDetails.aspx?ProductId=10995
I use them both, no waves or flickering in the images on low , mid and high.

Those are linear regulators. Work fine **IF** the battery input voltage is in the 'efficient' range of your output load. They also provide no dimming unless you PWM them (or they have a uC onboard to do the PWM control). Basically they are 'smart' resistors, no more, no less.

If your battery pack is far above or below your total LED Vf you need a buck or a boost driver to do actual power conversion. e.g. if you want good runtime you may have a 4 x li-ion battery pack (series) running say one or 2 XML's you would want to use a buck driver to convert the input voltage/power down to the total Vf at 90+ % efficiency versus burning the rest off as heat.

PWM is basically turning the load on and off at some frequency and at some duty cycle ON_TIME/(ON_TIME + OFF_TIME). Depending on duty cycle and frequency you may or may not have a problem with your camera or video camera based on shutter speed. The guaranteed way to NOT have problems at any shutter speed is to use a fully current regulated driver (dims by adjust current) versus using a PWM driver.

This really isn't difficult...

cheers,
george.