Difficulty driving LEDs with higher current?

Discussion in 'General Electronics Chat' started by kcb4365640, Jul 29, 2010.

  1. kcb4365640

    Thread Starter New Member

    Apr 9, 2010
    17
    0
    Hello everyone. I created an old thread that describes some of what I am doing http://forum.allaboutcircuits.com/showthread.php?t=36791. Now my problem is understanding why my LEDs seem to randomly not turn on when I driving them with a larger current? I know this is a very long post, but I wanted to include everything that may be relevant to why this is happening. Also I am at a complete loss on what is causing this problem, so I would be very grateful for any suggestions that might be causing this!

    First let me remind everyone what I am trying to do. I have ~40 NIR LEDs that I need to rapidly turn on/off individually (i.e. only one LED is on at a time), and I take an image only when the individual LEDs are on. I use a digital output from a USB data acquisition board to "switch" a reed relay that completes the LED circuit allowing a specific LED to turn on/off. This digital output signal is also used to send a trigger signal to the CCD camera, so it only exposes an image when the signal is "high". The 40 LED image sequence needs to be completed in the shortest amount of time possible. Therefore, I use very short pulses to turn the LEDs on/off (e.g. ~5ms). The NIR LEDs I use can be pulsed up to 1A for very short times (see the attached L850-40series.pdf from last thread).

    During the previous thread I took some good advice from Bill Marsden and others and I decided to just use a single resistor and a 5V power supply (see http://www.powerconversion.com/assets/ssl20c_1191534424.pdf for a description of the 5V power supply). During that time I ran tests to verify that the LEDs performance was not changing over time. During the test I turned on/off an individual LED many times (e.g. 500) and I recorded the average pixel intensity for each of the 500 images. The idea was that if resistor was too small (i.e. current too high) it would damage the LED and it would degrade in performance over the 500 images, and it might not come back to the original intensity. From my tests I decided that a resistor of 5.6 ohms was best. With this resistor I should have a current of 5V-2.5V=I*5.6ohms => I~450mA. At that time with this resistor I showed that that the LED brightness degraded over the 500 images (not significant though see image in last thread), and the when I repeated the test the pixel intensity came back to the original value so it wasn't permanently damaging the LEDs. Also from the manufactures data sheet (see attached in last thread) it was clear that it is safe to do this as long as duty cycle was taken into account.

    This brings me to my current dilemma on why higher currents seem to cause the LEDs to not turn on sometimes... Even before when I posted the old thread I noticed that if I used a smaller resistor (3.3 ohms) sometimes the individual LED would not turn on. At the time I didn't give it much thought because I planned on using the larger 5.6 ohm resistor, because I didn't want the current that high anyways. Since then I built a prototype device with a customized circuit board that takes the sequence of ~40 individual LEDs. Now I noticed that this problem seems to occur more frequently, and it occurs more often with some of the LEDs? Although when I use a very large resistor (e.g. 100 ohm) the LEDs seem to always turn on when they are supposed to (see attached image). I am attaching 4 images that illustrate what I mean. 2 of them used a 100 ohm resistor and the other two used a 5.6 ohm resistor, and I tried it with 2 different LEDs. The only difference is that I replaced the resistor! Why does the LED sometimes not turn on when I use the smaller resistor (larger current), but everything seems to work just fine with the larger resistor (smaller current). You can see when the LED is not on by looking at the figures when the average pixel intensity is ~ 37 counts.

    Some of the ideas that I have that might be causing the problems include: 1) I need a different power supply, but the current can output up to 4A and it is 5V (see http://www.powerconversion.com/assets/ssl20c_1191534424.pdf for a description of the 5V power supply). I also use the 5V power supply to power some IC chips (transistor arrays, OR logic gates, etc., but I don't think they take too much current especially up to 4A? 2) For some reason when I created the customized circuit/ prototype there are some type of cross talk because the higher currents lead to higher electrometric fields? 3) The reed relays that I use might have a problem (part number HE3621A0510, and spec sheet is located at http://www.hamlin.com/specsheets/HE3600 revised.pdf). 4) For some reason the LEDs themselves might be the problem, but from my understanding they should be fine with a larger current (please see attached PDF spec sheet), and if they are overheating because of the larger current why is it so random?

    Thanks again for any suggestions/comments on what might be causing the LEDs to not turn on sometimes when I use more current to drive them.

    Kevin
     
  2. Bernard

    AAC Fanatic!

    Aug 7, 2008
    4,172
    397
    Have you tried a transistor as a switch, replacing the reed relay to see if there is any change. Getting down to 10 ms & shorter may be a struggle for reeds.
     
  3. mcgyvr

    AAC Fanatic!

    Oct 15, 2009
    4,770
    970
    Typical reeds have operating times in the .5 to 1 ms range. (.5ms bounce and .2ms release) that shouldn't be the problem
     
  4. kcb4365640

    Thread Starter New Member

    Apr 9, 2010
    17
    0
    Thanks for both of your responses. According to the manufacture's spec sheets http://www.hamlin.com/specsheets/HE3600 revised.pdf, it says the max operate and release time is 0.5ms. All the images that I took had 5ms, and the problem didn't occur with the 100 ohm resistor.

    Currently I am trying different resistors and it looks like a 15 ohm resistor is the smallest resistor that never gives me a problem (I have tried it on 12 of the 40 different LEDs). Thanks.
     
  5. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    Specs are tricky things, a reed switch is still a mechanical device, and .5ms is also 500µs. I have trouble with anything mechanical operating in this time frame reliably, especially a cheap part such as reed switch. This is just an opinion. Is there any chance you could replace it with a solid state part?

    I don't remember the thread (so don't have a hope of pulling it back up) but I remember someone talking about an LED that was being over driven flashing in a mode that strongly resembled a relaxation oscillator. Basically (we thought) the LED over heated, shut down, cooled, came back up, repeat indefinitely. This was an accident, probably not very repeatable, but I think there is a slight chance it might apply here.

    If you have access to a high speed camera it might be interesting to watch the sequencing.

    Is there any way you can divide the current between two LEDs? Something similar to what was suggested in the original thread.
     
  6. kcb4365640

    Thread Starter New Member

    Apr 9, 2010
    17
    0
    Thanks for the response.

    Reed Relay: I also have a few solid state relays that can try, but that would require that I redo my entire circuit... Here is the spec sheet of the solid state relay I have a few of: http://www.crydom.com/en/Products/Catalog/d_o.pdf
    I just don't understand why the larger resistor current would still work with the same timing of 5ms? The only thing I can see is that according to the reed spec sheets it says the "Current, Switching" is 0.5 A max. From what I am measured with the multimeter it looks like it should only be 400mA?

    LED acting as an relaxation oscillator: It sounds like if this is the problem there is nothing I could do. Although from the graphs that show the Forward Current - Pulse Duration I should be fine (see figures in PDF attachment)?

    I don't have a high speed camera, but I am thinking of locating an oscilloscope and measuring the voltage across the power supply, resistor, and the LED to see if at some point the voltage drops to 0.

    Unfortunately, I need to have only one LED on at a time and it needs to be as bright as possible so I can have the "on" time short as possible.

    Thanks again for your help.

    Kevin
     
  7. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    I think the higher current pulses causes the supply voltage to drop which causes your sequencer circuit to miss a few outputs.
    Maybe the switching power supply is slow to regulate following a high current pulse.
    Maybe adding a pretty big supply filter capacitor will hold the supply voltage high enough during high current pulses.
     
  8. kcb4365640

    Thread Starter New Member

    Apr 9, 2010
    17
    0
    Thanks Audioguru. I was also wondering if my problems could be because I am using a switching power supply. I ordered a linear regulated power supply from Digi-key and should have it early this week. It is rated up to 1.5A. Do you think just by using that it would fix the problem if it is what you mentioned?

    I am not sure what you mean by adding a pretty big filter capacitor? Can you briefly tell me what a "big" one would be and I can add it to my circuit to try it out. Thanks.

    Kevin
     
  9. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    If the switching regulator can tolerate it, add a 0.1uf ceramic disc capacitor and a 1000uf "big" filter capacitor across the power supply voltage.
     
  10. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    You can turn a SMPS into linear with the addition of a linear regulator. All SMPS have some planned AC riding on them, it is inherent in the design. The LEDs won't care, but it might through your logic off a little on timing (unless you are using a precision time base).

    Just curious, have you slowed this down and verified the sequence?

    I'm also wondering about the sync. You mentioned earlier that the CCD scan rate was synced with the LED flash. If they got out of sync for whatever reason that would look very similar (I had mentioned aliasing earlier).
     
  11. kcb4365640

    Thread Starter New Member

    Apr 9, 2010
    17
    0
    Thanks. I too am wondering if the syncing is causing the problem now. A couple of times when I was running the 500 LED pulses I could visually "see" the LED turning on/off, and it looked o.k. (although it is pretty hard to tell since it is so fast). I can't really slow it down, because then the LEDs would be on too long and the current is too much and would damage the LEDs, and it doesn't have the problem with the smaller current. Then when the 500 test was done and I looked at the results of image pixel value vs. image number it only showed that the LED was "on" a few times (less than 10). This only happened a few times, but still... Since I wait ~100ms between each image and the LED pulse is 5ms there could be some syncing problems. As I mentioned it seems to only happen when I have a smaller resistor (larger current through LEDs). It is very possible that for some reason when I have the larger current my LED pulse is not synchronized with my camera exposure anymore...

    I looked at the power supply spec sheets again and there are several parameters tolerance time frames on the order of a few ms. http://www.powerconversion.com/assets/ssl20c_1191534424.pdf
    For example, it says the "overshoot/undershoot" is less than 10ms, the "hold-up time" is 10ms, etc. I don't really know what these parameters are, but if they would somehow give me a time offset to my LEDs, but not the camera exposure trigger signal then it might cause this problem.

    Thanks again for your help, and if someone could look at the power supply spec sheet to see if those or any other parameters might be causing my problems.
     
  12. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    The problem might be caused by the "undershoot".
    It is when a high load current causes the supply voltage to drop (they say it drops 10%) and it takes nearly 10ms for the voltage to return to normal.

    Cmos logic has good supply voltage fluctuation immunity but old TTL logic has less.
     
  13. kcb4365640

    Thread Starter New Member

    Apr 9, 2010
    17
    0
    Thanks Audioguru and Bill_Marsden.

    I am not sure if it will help, but I ordered a 5V linear regulated power supply from Digi-key (http://www.power-one.com/resources/products/datasheet/lin.pdf), and it should arrive by Tuesday. I don't know much about the linear regulated power supplies, but from the description it seems that it might work better... At least it doesn't hurt to try it.
     
  14. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    I don't know which linear power supply you ordered but they seem much better than the switching one.
     
  15. kcb4365640

    Thread Starter New Member

    Apr 9, 2010
    17
    0
    Well it looks like the problem is "sort of" fixed... When I ran the 500 image test with the new linear regulated power supply it worked for 36 LEDs, it didn't work many times for 1 LED, and it did not work for 2 LEDs only one time during the 500 images? I am attaching three images that illustrate what I mean (one figure that shows how the LED works all the time, one that only shows how the LED only failed once, and one that shows how the it fails many times).

    I am still not sure why it happens for some LEDs, but not others? Also why does it seem to only occur when I use a smaller resistor (larger current)? I will say this that when using the linear regulated power supply the problems are less frequent (number of LEDs that it occurs is fewer and the number of times it happens is less often).

    Thanks again for any suggestion on what might be causing this problem.

    Kevin
     
  16. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    Try putting 0.1µF caps as close to the source for the switching components (if you haven't already done this). You may need a surge source for current.
     
Loading...