High powered LED control using 2 independent MOSFET's in series

Thread Starter

Luukmaster

Joined Sep 6, 2019
9
Hi, i'm trying to build a 14 segment number display using these 10W high powered LED bars. Every LED is controlled by a IRF530N and de final MOSFET at the bottom is an IRLZ44N used to dim the whole thing using one PWM pin on the arduino mega controlling it. Every LED bar has current limiting built-in but in the schematic it is displayed as 13.3Ohm resistors. This circuit works fine until we start turning off LED's, whenever there are 4 or less LED's active the PWM seems to stop working causing the LED's to ramp up to full brightness. In this configuration the LED's consume 480-600 mA on full power depending on the amount of enabled LED's (less LED's = Higher Amperage).
i hope you can help me with a few questions:

Q1: What would be the proper way to wire up 2 MOSFETS in series while being able to control them indepentently? this is the only way i found to get them to work but it feels like there is a much better way to do this.

Q2: Would it be better to use a logic level MOSFET as PWM switch? we're planning to have a PWM frequency of 1-2KHz

Q3: Why does the LED's power consumption decrease when more LED's are enabled? the supply is not limited.

Also, i know it would be better to control every led using PWM individually but we dont have enough PWM outputs available.
upload_2019-9-6_9-4-56.png
 

Sensacell

Joined Jun 19, 2012
3,432
Eliminate the master FET at the bottom, ground the FET Source terminals.

Replace the 5V supply to the switches with a single buffered 5V PWM signal.
Use a logic level gate N-CH FET to be sure that the channel is fully on. Choose a FET with low gate charge, which should be easy since the individual load current is rather small. ( > 1 A )

The combined gate capacitance will be the sum of all the FET's - so use a gate driver IC to buffer the PWM.

In your current circuit, the bottom FET is not getting enough gate voltage to fully turn on, so it's operating in the constant current part of the curve, limiting the total power available. (look at the datasheet)

I imagine you designed it this way to avoid heavy currents in those switches?
 

TeeKay6

Joined Apr 20, 2019
573
Hi, i'm trying to build a 14 segment number display using these 10W high powered LED bars. Every LED is controlled by a IRF530N and de final MOSFET at the bottom is an IRLZ44N used to dim the whole thing using one PWM pin on the arduino mega controlling it. Every LED bar has current limiting built-in but in the schematic it is displayed as 13.3Ohm resistors. This circuit works fine until we start turning off LED's, whenever there are 4 or less LED's active the PWM seems to stop working causing the LED's to ramp up to full brightness. In this configuration the LED's consume 480-600 mA on full power depending on the amount of enabled LED's (less LED's = Higher Amperage).
i hope you can help me with a few questions:

Q1: What would be the proper way to wire up 2 MOSFETS in series while being able to control them indepentently? this is the only way i found to get them to work but it feels like there is a much better way to do this.

Q2: Would it be better to use a logic level MOSFET as PWM switch? we're planning to have a PWM frequency of 1-2KHz

Q3: Why does the LED's power consumption decrease when more LED's are enabled? the supply is not limited.

Also, i know it would be better to control every led using PWM individually but we dont have enough PWM outputs available.
View attachment 185583
@Luukmaster
With your present circuit, when the PWM drive is used, you are forcing the MOSFET of every "on" LED to switch on and off, even though I believe it is your desire that the IRLZ44N provide the PWM control. Move the negative connection of V1 from ground to the common net tying all the MOSFET sources together. Then any "on" device remains "on" even while the IRLZ44N cuts current to zero.

@Sensacell
I do not read the IRLZ44N datasheet as you do. The device should (per datasheet) be operating in the ohmic, not constant current region. Ron is shown in the datasheet as quite low with Vgs=5V. A quick voltage measurement of Vds of the IRLZ44N will reveal whether the device is remaining in saturation.
 

danadak

Joined Mar 10, 2018
4,057
Sim seems to work -

upload_2019-9-6_6-32-30.png


Of course thats with ideal gate drivers. And IRF530 not a logic level MOSFET.


Regards, Dana.
 

Thread Starter

Luukmaster

Joined Sep 6, 2019
9
@Luukmaster
Thanks for posting the updated schematic. As others have noted, 5V is marginal as gate drive for the IRF530. Why not use the 12V as gate drive rather than the 5V? (i.e. bypass V1)
Are you now happy with operation?
Because they are controlled by an arduino which as far as i know can only supply 5V unless i add more components. As for if im happy with the operation? i thinks so, i'll have to test it in the real world soon to say for sure.
Thanks everyone for the help!
i'll try to update if its succesfull
 

Thread Starter

Luukmaster

Joined Sep 6, 2019
9
After messing around in the simulation i did notice that using the current circuit with a PWM signal the LED's that are suppost to be off get a very brief 40 ns pulse of 40-80mA everytime the PWM signal switches on/off. im not sure if that will be visible or if the LED's will even turn on (i hope not) is this due to the internal capacitance of the MOSFET's?
 

Thread Starter

Luukmaster

Joined Sep 6, 2019
9
Alright, it seems the current schematic works for the most part, only one problem. whenever we turn up the brightness (PWM dutycylce) all the inputs on the arduino (displayed as 5V source in the schematic) get really unstable. we think this is because our ground is kind of floating. The 12V source in the schematic is a DC-DC converter and the arduino is fed by a another 5V DC-DC converter, and the whole thing is powered by a salvaged laptop charger (19V) plugged into mains. could we solve our problem by connecting the negative outputs from the DC-DC converters to the laptop charger? and possibly even directly to mains?
 

TeeKay6

Joined Apr 20, 2019
573
Alright, it seems the current schematic works for the most part, only one problem. whenever we turn up the brightness (PWM dutycylce) all the inputs on the arduino (displayed as 5V source in the schematic) get really unstable. we think this is because our ground is kind of floating. The 12V source in the schematic is a DC-DC converter and the arduino is fed by a another 5V DC-DC converter, and the whole thing is powered by a salvaged laptop charger (19V) plugged into mains. could we solve our problem by connecting the negative outputs from the DC-DC converters to the laptop charger? and possibly even directly to mains?
@Luukmaster
In your schematic of post#8, you label Q16 as IRFZ44N; I believe that should be IRLZ44N. A low Ron with only 5V Vgs is certainly desirable.

Are the Arduino drive signals able to drive the gates to both ground or +5V? Or are the drive signals only able to drive to +5V? If the drives can do both ground & +5V, then the 6.4K resistors between each gate and ground should be removed (i.e. replaced with open circuits).
 
Last edited:

Thread Starter

Luukmaster

Joined Sep 6, 2019
9
In your schematic of post#8, you label Q16 as IRFZ44N; I believe that should be IRLZ44N
I've checked, and this is indeed wrong, we have an IRLZ44N. However, all the others are IRF530N's, which are not perfect but do the job.
Are the Arduino drive signals able to drive the gates to both ground or +5V? Or are the drive signals only able to drive to +5V? If the drives can do both ground & +5V, then the 6.4K resistors between each gate and ground should be removed (i.e. replaced with open circuits)
I don't quite understand what you mean, the switches don't seem to heave a problem switching fully on, or atleast enough for our needs.
Try PWMing each LED string independently using one of these..
https://www.adafruit.com/product/815
This would probably work, but we would still need to switch every LED with a MOSFET because the LED's consume up to 10W each. we will probably try.

As for the current problem, we double checked if our ground wasn't floating and the problems persisted. Every connected input fails to read properly without any measurable cause. i'll add the inputs to the schematic soon.
 

dendad

Joined Feb 20, 2016
4,452
This would probably work, but we would still need to switch every LED with a MOSFET because the LED's consume up to 10W each. we will probably try.
Yes, you still need the FETs. But, as mentioned, you can PWM 16 different strings independently. And bu linking the solder address pads, extra boards can be used to increase the number of strings in x16 lots.
 
Top