why LED current fliping in negative direction in LT spice simulation?

Thread Starter

meena udayasankar

Joined Mar 29, 2017
14
Hi all,

I tried to build a LED driver circuit to drive my LED.
I am giving PWM pulse to the driver chip and checking the output current across the LED.
The currents across the led flipping in negative direction when simulating it.
Can anyone give me a reason for it ? why the LED current flipping in negative direction.
 

Attachments

wayneh

Joined Sep 9, 2010
16,390
The brief transient showing a tiny negative current isn't really your problem and could even be an artifact of the simulation. The problem is a lack of positive current to light the LED.

Could it be as simple as the LED placed backwards?
 

Thread Starter

meena udayasankar

Joined Mar 29, 2017
14
The brief transient showing a tiny negative current isn't really your problem and could even be an artifact of the simulation. The problem is a lack of positive current to light the LED.

Could it be as simple as the LED placed backwards?
Ya. It could be the transient. But actually I am trying the below circuit. I gave the pulse with (2000:1) dimming ratio as per the below graph.
I should get the graph like the below one right. But my current is flipping down. LED is the load. How can it produce negative current

upload_2017-3-31_9-15-28.png
 

Alec_t

Joined Sep 17, 2013
11,268
In your sim, you could reduce the values of C2 and C3 to 4.7nF to make the sim faster (with the present values of 0.1uF it takes ~12mS to get going). However, the LM3517 operates with an internal switching frequency of ~1MHz, so I don't think you are going to be successful when using a PWM input having a pulse width of only 3.5nS.
 
The graph from Linear Tech shows a PWM signal high time of 4 us (0.000004 s) not 4 ns (0.000000004 s). If you look at the current through the LED at the 4 ns time, it is 0 A. The IC has not had time to do anything.

The datasheet also suggests that the minimum HIGH time of the PWM signal should be 4 cycles of the internal oscillator, which leads to the 4 us HIGH in the example.

Voltage on the opposite side of the LED, vo2, will show that the IC is not doing what you expect. I would also agree that the reverse current on order of nanoAmps is not the main issue here. Once you get the circuit working, i.e. drawing forward current through the LED, the small reversal in current will not longer be a bother.
 

crutschow

Joined Mar 14, 2008
24,914
Here's the simulation with the values of C2 and C3 changed. and the PWM duty cycle changed to 50% (yours had 3.5ns not 3.5μs).
(Why do you want such a high PWM frequency?)
The LED current doesn't pick up until after 1.1ms.
The spikes are due to capacitive feedthrough from the fast switching edges.
They would not appear in the LED light output.

upload_2017-3-31_10-20-8.png
 

Thread Starter

meena udayasankar

Joined Mar 29, 2017
14
Hi
Here's the simulation with the values of C2 and C3 changed. and the PWM duty cycle changed to 50% (yours had 3.5ns not 3.5μs).
(Why do you want such a high PWM frequency?)
The LED current doesn't pick up until after 1.1ms.
The spikes are due to capacitive feedthrough from the fast switching edges.
They would not appear in the LED light output.

View attachment 123690
Hi,

Ya correct.. Got the point. But in the above simulation , the current of LED spikes goes to 2A. Will it make the LED to glow? (because for dimming purpose only i am giving PWM pulse)
 

Thread Starter

meena udayasankar

Joined Mar 29, 2017
14
The graph from Linear Tech shows a PWM signal high time of 4 us (0.000004 s) not 4 ns (0.000000004 s). If you look at the current through the LED at the 4 ns time, it is 0 A. The IC has not had time to do anything.

The datasheet also suggests that the minimum HIGH time of the PWM signal should be 4 cycles of the internal oscillator, which leads to the 4 us HIGH in the example.

Voltage on the opposite side of the LED, vo2, will show that the IC is not doing what you expect. I would also agree that the reverse current on order of nanoAmps is not the main issue here. Once you get the circuit working, i.e. drawing forward current through the LED, the small reversal in current will not longer be a bother.
Hi,
Thanks a lot. In the datasheet they mentioned a PWM dimming ratio formula. (i.e) PWM(Ratio)=Tmax/Tmin.
From the graph, I took the period and calculated Tmin with 2000:1 dimming ratio. That's y I gave ns pulse.

PWM= tmax/tmin

Tmax/Tmin= 2000/1=7us/Tmin

Tmin=7us/2000=3.5ns

Tmin=3.5ns

Tmax/Tmin=7us/3.5ns=2000:1
 

Thread Starter

meena udayasankar

Joined Mar 29, 2017
14
In your sim, you could reduce the values of C2 and C3 to 4.7nF to make the sim faster (with the present values of 0.1uF it takes ~12mS to get going). However, the LM3517 operates with an internal switching frequency of ~1MHz, so I don't think you are going to be successful when using a PWM input having a pulse width of only 3.5nS.
Hi,

Thanks.. I changed the pulse from ns to us. I got the LED current spike of 2A. In that case, led will glow know..
 

crutschow

Joined Mar 14, 2008
24,914
the current of LED spikes goes to 2A. Will it make the LED to glow
As I stated in post #13 the spikes are due to capacitive feedthrough and will not cause the LED to glow. The spikes are coupled by the junction capacitance and do not generate carriers that emit light.
Thanks a lot. In the datasheet they mentioned a PWM dimming ratio formula. (i.e) PWM(Ratio)=Tmax/Tmin.
From the graph, I took the period and calculated Tmin with 2000:1 dimming ratio. That's y I gave ns pulse.
You have it backwards.
The 7μs is Tmin, not Tmax.
So for a 2000:1 dimming ratio the pulse period should be 7μs * 2000 = 14ms.

Incidentally, the word "y" is spelled why.
 
Top