Hello All,

I have a circuit which I have designed for a customer. The circuit contains a lot of white LEDs which are all driven by a single PWM signal. This PWM signal is generated by a microcontroller and passed to a very strong FET and from there to all the LEDS anodes. I can also control each led on/off state using a separate FET on each led cathode to GND.

What powers the FET is the output of a DC/DC converter that outputs 3.3V and can carry up to 4A load (even 5.5A if taken to the limit).

The problem I am facing is that this DC/DC converter will go into hiccup mode when I try to operate many LEDS together even in a low duty cycle mode.

Now before you all jump and say "That's easy - you have overloaded your DC/DC output with too much current" - well, that is not the case.

The overall current taken out of the DC/DC is about 1A (1/4 of its capability). If I reduce the number of LEDS that I want active and increase the duty cycle things work well and I can measure as much as 4.5A on the output of the DC/DC and so it seems that the number of active leds is the cause of the problem and not the current they all draw.

I am attaching a single schematics which I have copied and pasted into it the important parts from my design. In it you can
see the DC/DC circuit, the PWM circuit and a sample of two leds circuit (all other leds circuits are exact duplicates).

BTW, the lower FET of each led (2n7002n) has a resistance of 2ohms.

Any help with solving this would be greatly appreciated.

Mod: re-imaged,

 

What computer are you using? What is the supply voltage for the IO pins? 0V and 3.3 or 0/5V???
Do you know if Q20 is turned on well. What is the voltage at Q1-pin3 when hickups?
Same question for Q1, Q2 Drain? Are the transistors turned on well. The data sheet said they are not turned on well at 3.3V of signal and they are not going to work at high current. Are they running hot?
How are you measuring current?
What LEDs are you using? LEDs are not designed to be run at 3V. They are designed to be run at some current. How are insuring the right current for each LED?

 

Hello All,

I have a circuit which I have designed for a customer. The circuit contains a lot of white LEDs which are all driven by a single PWM signal. This PWM signal is generated by a microcontroller and passed to a very strong FET and from there to all the LEDS anodes. I can also control each led on/off state using a separate FET on each led cathode to GND.

What powers the FET is the output of a DC/DC converter that outputs 3.3V and can carry up to 4A load (even 5.5A if taken to the limit).

The problem I am facing is that this DC/DC converter will go into hiccup mode when I try to operate many LEDS together even in a low duty cycle mode.

Now before you all jump and say "That's easy - you have overloaded your DC/DC output with too much current" - well, that is not the case.

The overall current taken out of the DC/DC is about 1A (1/4 of its capability). If I reduce the number of LEDS that I want active and increase the duty cycle things work well and I can measure as much as 4.5A on the output of the DC/DC and so it seems that the number of active leds is the cause of the problem and not the current they all draw.

I am attaching a single schematics which I have copied and pasted into it the important parts from my design. In it you can
see the DC/DC circuit, the PWM circuit and a sample of two leds circuit (all other leds circuits are exact duplicates).

BTW, the lower FET of each led (2n7002n) has a resistance of 2ohms.

Any help with solving this would be greatly appreciated.

Mod: re-imaged,
View attachment 255609

Assuming the FETs are on / off, you seem to have no current limiting on the LEDs.
QED a short circuit,


It is probably to late , but is does seem a very very complicated way to achieve this
why did you not use on of the may led driver chips that are designed for just this application

https://www.ti.com/power-management/led-drivers/overview.html

 

What is the current rating of each LED?
How many is 'a lot'?

 

How many is 'a lot'?

It appears he is putting 3.3V on 3V LEDs and not really knowing what the current it. The only current limit I can see is the on_resistance of the MOSFETs.
The first power supply makes 3.3V.

 

Hello Guys,
Thank you for your feedback. Some answers to the many questions posted:

1. The Leds I use have 250mA max forward current with a forward voltage of 2.9-3V. (Osram)
2. I am trying to lit up more than a 100 LEDS at once.
3. The I/O voltage is 3.3V
4. Yes, Q20 is turned on well. I have also try replacing it with a short circuit to GND but that did not solve it.
5. The 2n7002n is rated for more current than what I am working with (3.3-3)/2 = 150mA.
6. It is true that the on resistance of the 2n7002n is the only limiting current but do not forget I am controlling the PWM
of the signal and so can control the total energy driven to the LEDS (which BTW works very very well for over 2 years now)
7.I can not use a led driver since in total I have 576 leds on the board and need to control the on/off state of each of them separately and have a common PWM signal to all of them.
8. Non of the transistors are getting hot, they all stay in normal temperature.
9. Current at full 100% PWM duty cycle (DC 3.3V) was measured between the LED_PWR signal and the drain of Q195.

What do you guys think caused the DC/DC to go into hiccup mode? And... How can I solve it?

Some ideas I had:

1. Current spikes?
2. Large capacity generated somewhere?

 

I suspect it may not be the DC-DC converter having issues. What is the source of the 12 volt supply? Hiccuping is exactly what happens when you overload a switching power supply – it may be rated for the full LED current, but shutdown on the startup surge. Switchmode power supplies have a strong survival instinct.

 

A very strong battery (12V/100W) is feeding the DC/DC input so I do not think the problem is there

 

Hello Guys,
Thank you for your feedback. Some answers to the many questions posted:

1. The Leds I use have 250mA max forward current with a forward voltage of 2.9-3V. (Osram)
2. I am trying to lit up more than a 100 LEDS at once.
What do you guys think caused the DC/DC to go into hiccup mode? And... How can I solve it?

(some text removed from quote for clarity)

Some ideas I had:

1. Current spikes?
2. Large capacity generated somewhere?

It looks like @ronsimpson 's diagnosis (LEDs are driven too hard) seems to be consistent with what you describe. You are driving the LEDs in a way that can only be done under carefully control conditions with LEDs that have narrow current distributions at a particular voltage. Because there is variation in the LEDs, the on resistance of your MOSFETs, and your (trimmed?) power supply, the best solution is to add a small series resistance to each LED.

What is the manufacturer and part number for the LEDs?
Can you post a datasheet to help us understand exactly what is going on?
What is the magintude of PWM_PWR? 3.3V?

Toward your other ideas:
1. Current spikes? - Not likely.

2. Large capacity generated somewhere? - Similarly not likely.

Drive your LEDs in a way in which the current is predictable and you will probably see the problem vanish.

 

Yes, PWM_PWR is 3.3V. BTW, when trimming down the 3.3V down to 3V I can get more LEDS run together at high duty cycle.

Attached is the data sheet for the LED I am using.

Question remains - what is actually causing the hiccup on the DC/DC output? The same design once run directly
from a single cell li-ion battery (ranging from 4.2 down to 3.V) instead of the DC/DC and everything was working fine.

 

Yes, PWM_PWR is 3.3V. BTW, when trimming down the 3.3V down to 3V I can get more LEDS run together at high duty cycle.

Attached is the data sheet for the LED I am using.

Question remains - what is actually causing the hiccup on the DC/DC output? The same design once run directly
from a single cell li-ion battery (ranging from 4.2 down to 3.V) instead of the DC/DC and everything was working fine.

Depending what you mean by "Hiccup"

The DCDC is either
over loaded
in sufficient input
under loaded.

Some graphs would be of use here,

One thought,
the DCDC is to a first degree "a feedback loop amplifier",
when the leds are turned on, they appear momentarily as a dead short across the output,
and the DCDC has a relatively sophisticated over current circuit,

A LiIon cell, is a much lower output impedance than the DCDC,
and has only a very slow over current circuit,
So the battery rolls over the initial short circuit
and its internal limit circuit is probably providing the current limit you want,

If you scope the leds you should see the difference between the cell and the dcdc system,

In the circuits I referenced above,
you will see thats its normal to run multiple leds in series,
and use a step up led driver,

You could try a bodge, put w resistance on th eoutput of the DCDC,
or idealy in series with the parallel leds,

 

Yes, PWM_PWR is 3.3V. BTW, when trimming down the 3.3V down to 3V I can get more LEDS run together at high duty cycle.

Attached is the data sheet for the LED I am using.

Question remains - what is actually causing the hiccup on the DC/DC output? The same design once run directly
from a single cell li-ion battery (ranging from 4.2 down to 3.V) instead of the DC/DC and everything was working fine.

Some power supplies shut down when the current limit is exceeded then go into "hiccup mode" during which the output is periodically turned back on to see if the overload is still there.

Everything in your response (above) indicates that your LEDs are drawing too much current as @ronsimpson suggested. Series resistors are still a good way to make the current draw more predictable and to balance the current among the paralleled LEDs (if you have some in parallel).

 

1. The Leds I use have 250mA max forward current with a forward voltage of 2.9-3V. (Osram)
2. I am trying to lit up more than a 100 LEDS at once.

So the maximum current draw could be 100 x 0.25A = 25A., but your power supply is rated for a peak of only 5.5A. Do you see the problem there?

 

Hello,

How are you controlling the current in the leds?
Leds are current controlled devices and NOT voltage controlled.

Bertus

 

Interesting that your project has worked for a year. Have you tried changing the power supply? Maybe it broke. What changed? Did you add or change anything?
-----added-----
When you build one hobby project you can get away using LEDs in voltage mode. 3V on a "3V" LED. When we build for production the LEDs need to be in current mode. The FETs need a good strong Gate signal. When we get millions of LEDs a month there are variations in the parts. (FET also) All parts come slightly different. The reason several people kick against your design is that if I copied it with LEDs I got from a different batch, likely the current will be very different. You are lucky it works.
-----added-----

I am thinking about a project that is almost the same. 100s of LEDs but I want color. Here is a picture of a little LED/IC that does PWM and controls the current. Controlled by a computer. Food for thought.