Solution Finder

LYTSwitch-6 constant current with PWM drive

Posted by: scaprile on Tue, 05/04/2021

Hi,

the idea is to power several LEDs in series at constant current. The LYTSwitch-6 would be configured for let's say 48V and current limited to 0.5A via the sense resistor connected to IS.

The LED current would be switched on/off by a series MOSFET in order to perform PWM light control at 100Hz, so the LYTSwitch-6 will effectively see no current for 5ms, then 500mA for 5ms, and so on. PWM control is a requirement to maintain the same color temperature over the whole light intensity span.

I wonder how this would affect control loop stability. Is this a suitable operation for these chips ? Any others ?

Regards

Comments

Submitted by PI-Zek on Wed, 05/05/2021

Thank you for your interest in using LYTSwitch-6 product. Can you please provide more details about this PWM light control at 100Hz (max voltage level of PWM, etc.)?

If the design is for an LED dimmable application, then LYTSwitch-6 is suitable for PWM dimming function with its variable duty cycle as the dimming input.

We have existing applications with external dimming control circuit compatible to PWM dimming input: 100 Hz to 3 kHz, 10V peak PWM, variable duty cycle. This dimming circuit converts the input PWM signal into an analog DC voltage (proportional to the duty cycle) which serves as variable reference of the LYTSwitch-6 to control the output current through the feedback pin of the controller. A maximum 100% duty cycle input at the dimming terminals will provide the maximum output current set at IS pin and the minimum duty cycle at 0% will be at minimum output current (usually light OFF).

Please see below reference design which is dimmable by a 10V PWM signal (100Hz to 3kHz; variable duty cycle). This is also designed for 48V LED load but at a higher output maximum current of 2.08 A.

https://www.power.com/design-support/design-examples/rdr-801-100-w-2-stage-boost-and-isolated-flyback-dimmable-led

I think I was clear enough but I see I wasn't.

The LED current would be switched on/off by a series MOSFET in order to perform PWM light control at 100Hz, so the LYTSwitch-6 will effectively see no current for 5ms, then 500mA for 5ms, and so on.

The PWM circuit will connect and disconnect the LEDs from the LYTSwitch power supply, with 50% duty cycle operation (for example) there will be 500ms load for 5ms, no load for 5ms. There is no modulation of the reference nor any analog tweaking of the power supply, it must work as a constant current source while it has load and constant voltage when in open circuit.

PWM control is a requirement to maintain the same color temperature over the whole light intensity span.

If I was to control light intensity by changing the current limit, a low current at low intensity levels will yield a different color temperature (at least with these white LEDs), a blueish light. The current must be kept around 500mA over the whole intensity span, which will be given by the LED duty cycle as in a muxed circuit. The LEDs are on for some time and off for some time. At mid point, 50% duty cycle, the LEDs are on half the time and at 100 Hz that is 5ms on, 5ms off. Extreme cases might be 500us on 9.5ms off and 9.5ms on 500us off, but given stability parameters I could adjust the duty cycle limits to keep the power supply stable. (If it doesn't work at 50% it won't work at the extremes...)

Of course if you just modulate the reference it will be stable and I've already seen most application notes and reference designs do that, that is the reason why I'm asking here; I'm concerned how such a load variation might affect control loop stability in the power supply.

If this is not possible I will resort to a two stage solution with some PI chip for off-the-line constant voltage power supply, enough capacitance for stability, and an Allegro chip for constant current PWM light intensity control.

 

I think I was clear enough but I see I wasn't.

The LED current would be switched on/off by a series MOSFET in order to perform PWM light control at 100Hz, so the LYTSwitch-6 will effectively see no current for 5ms, then 500mA for 5ms, and so on.

The PWM circuit will connect and disconnect the LEDs from the LYTSwitch power supply, with 50% duty cycle operation (for example) there will be 500ms load for 5ms, no load for 5ms. There is no modulation of the reference nor any analog tweaking of the power supply, it must work as a constant current source while it has load and constant voltage when in open circuit.

PWM control is a requirement to maintain the same color temperature over the whole light intensity span.

If I was to control light intensity by changing the current limit, a low current at low intensity levels will yield a different color temperature (at least with these white LEDs), a blueish light. The current must be kept around 500mA over the whole intensity span, which will be given by the LED duty cycle as in a muxed circuit. The LEDs are on for some time and off for some time. At mid point, 50% duty cycle, the LEDs are on half the time and at 100 Hz that is 5ms on, 5ms off. Extreme cases might be 500us on 9.5ms off and 9.5ms on 500us off, but given stability parameters I could adjust the duty cycle limits to keep the power supply stable. (If it doesn't work at 50% it won't work at the extremes...)

Of course if you just modulate the reference it will be stable and I've already seen most application notes and reference designs do that, that is the reason why I'm asking here; I'm concerned how such a load variation might affect control loop stability in the power supply.

If this is not possible I will resort to a two stage solution with some PI chip for off-the-line constant voltage power supply, enough capacitance for stability, and an Allegro chip for constant current PWM light intensity control.

Submitted by PI-Zek on Sat, 05/08/2021

Thank you for the details your provided. We have an application which also features the CCT (correlated color temperature) function effectively using PWM but using two LED load strings with different color temperatures. The on and off cycle of each string will vary to select the desired color temp. To provide the cool/neutral white, the full duty will be on the 5000K string (OFF at 3000K string) and to provide warm temp, the full duty will be on the 3000K LED (OFF 5000K). More details on the CCT function on page 21.

https://www.power.com/sites/default/files/product_document/design_example/der-847_50w_2stage_boost_isolated_flyback_led_ballast_cct_hiperpfs-4_…

As for the loop stability, I will get more information about the device's capability of the load variation response then come back to you. Thank you.

 

Well, my application is close to this one if you remove Q8, R59 and R60. There is only one LED branch that is turned on and off. Also, as there is no power selection, I wouldn't use the 3-in-1 circuit, the resistor divider connected to the FB pin would be set to obtain an output voltage higher than the LED branch max voltage, and the current feefback resistor connected to the IS pin would be set to obtain a 500mA output current (35.9mV/500mA=72mohm).

My expectation is that the power supply will go into constant voltage regulation when the LEDs are off (48V). When the LEDs are on (providing the total series LED threshold voltage is lower), it will enter constant current regulation, so output voltage will be determined by the LEDs themselves; the former controlled by Q7 at 100Hz (or so), with continuously variable duty cycle to be determined within safe stability and regulation limits.

I wait for you to provide further data, thanks.

Submitted by PI-Crumb on Thu, 05/27/2021

hi,

I apologize for the late reply. I will be responding to your further inquiry since the other engineer got transferred to another branch. Your expectation is correct in that when the LEDs are ON, the unit will go into constant current regulation, and when the LEDs are OFF, the unit will operate in constant voltage.

However, there will be an output current overshoot everytime you turn ON the LED, because the power supply voltage is sitting at a higher voltage than the LED voltage, and the output capacitor will discharge to the LED. The overshoot current could be several amps higher than the operating current. 

DER-847 works well because there are two LED branches and they switching is complementary - if one LED is ON, the other is OFF, and vice versa. However, since you only have one LED branch, then overshoot will occur.

Thanks, yes, the current overshoot is something I've been worried about, depending on the number of LEDs and cap size it could be really dangerous.

We are re-thinking the application and we'll start without dimming.