3

I designed a pcb with the following dual led circuit based around the TPS92518: enter image description here

I would like to turn on and off the Powerleds in the <1ms range. At the moment our camera stills sees the leds a bit glowing. When I measure on the scope, my output shows a very slow off time curve. enter image description here

How do I get a sharper edge? I played around with changing Cboot values, but higher then 100nF doesn't have a positive effect.

Can it be the N-MOSFET? PSMN019-100YL Total gate charge is 72.4 nC

user3411864
  • 145
  • 6
  • 1
    Are these white LEDs? Keep in mind that the triplet state lifetime on a lot of LED phosphors is hundreds of microseconds or even more, so many LEDs will continue to glow even after the voltage reaches zero (although often only dimly). – user1850479 Feb 13 '20 at 18:51
  • Good point, yes they are white powerled arrays. Can it also been seen as a reverse voltage on the output? – user3411864 Feb 13 '20 at 18:57
  • I have to agree with @user1850479. If you are observing these LEDs with a photometer of some kind, then phosphors can have very long decays. (Usually, they don't make phosphors good enough to have just one decay tau, but a smear of similar ones -- it takes real talent in making phosphors to get the taus all close.) In any case, I'd add that triplet states take as long to charge up as they do to charge down, so you might narrow the ON-time and see if that has an impact. Another thing is that you may be saturating your photodetection system and it's taking time to come out. Lots to check. – jonk Feb 13 '20 at 18:57
  • However, phosphors have another important characteristic that will make this very easy to test. Their emission tau is HIGHLY dependent on their temperature. So my recommendation is to heat them up. You should see a drastic change (to the shorter) in the decay curve if this is the problem. Thermally isolate the LEDs and heat them (or cool them down.) Lattice phonons have a huge impact on the rate of triplet to singlet transitions. – jonk Feb 13 '20 at 18:59
  • 1
    BTW the same curve is also seen on the mosfet gate – user3411864 Feb 13 '20 at 19:01
  • @user3411864 Then heat or cool the LED. If you see a significant change in your observations, it is the phosphor. If you do NOT, then it's not the phosphor and perhaps you should focus on the MOSFET gate. Simple test to perform. I spent twenty years working with Dr. Wickersheim and company on phosphor thermometry. So I kind of know a little bit on that side. – jonk Feb 13 '20 at 19:02
  • @user3411864 You have a couple options. Change the temperature, get some monochromatic blue LEDs and see if the response is similar (no phosphore), or try driving your white LEDs with a function generator (should give a nice, fast off voltage). Once you know the problem you can pick the appropriate solution. – user1850479 Feb 13 '20 at 19:07
  • When I spray freeze spray on the leds to cool them down, I don't see any change on the scope – user3411864 Feb 13 '20 at 19:08
  • BTW the scope image is 6ms on 6ms off. But when running 1ms on 1ms off, I get the same slow curve. roughly 10.4V after 1ms off. – user3411864 Feb 13 '20 at 19:14
  • 2
    I suspect it's the current stored in your inductor. Try half the inductance...you should get more ripple and faster off time if I'm right. – Cristobol Polychronopolis Feb 13 '20 at 19:23
  • @user3411864 What is the off voltage for your LED array? – user1850479 Feb 13 '20 at 19:29
  • 1
    @user3411864 since the spray cold didn't change the decay rate, it isn't the phosphor. – jonk Feb 13 '20 at 20:35
  • You say you see the same curve on the MOSFET gate...maybe you can apply a pull-down there to turn it off faster. – evildemonic Feb 13 '20 at 20:46
  • Is this PWM controlled 81 Hz? – Tony Stewart EE75 Feb 13 '20 at 21:06

1 Answers1

2

If you are trying to turn off the LEDs using SPI control then the stored energy in the buck converter’s inductor will take time to decay to zero. If you need a solution that turns the LED off quickly then you might consider a more direct method of simultaneously turning off using SPI and dumping that stored energy into a low value resistor using another MOSFET.

Andy aka
  • 456,226
  • 28
  • 367
  • 807
  • This answer is supported by the scope image. Although OP will have an issue at some point with typical LEDs and phosphor. – MadHatter Feb 13 '20 at 19:36
  • @MadHatter I suspect he will. Any idea how long the phosphor will keep illuminating? – Andy aka Feb 13 '20 at 19:39
  • Would this decay energy show up at the mosfet gate? Because when I measure on the gate, I get the same curve – user3411864 Feb 13 '20 at 21:20
  • @user3411864 Yes, because the bootstrapping function in the chip tracks the source voltage via the boot capacitor so that it can adequately drive the gate above the incoming power rail when it needs to. False alarm in other words; normal operation for a bootstrap circuit. – Andy aka Feb 13 '20 at 21:34
  • Oh wauw, thanks a lot for pointing this out! Could it also be a combination of gate capacity and coil decay? Because an buck-down circuit in LTspice with the same component values (chip switching simulated as a square wave) shows a 350us decay time. – user3411864 Feb 13 '20 at 21:56
  • You should be able to view the gate voltage and source voltage simultaneously and work that out for yourself. – Andy aka Feb 13 '20 at 22:11