This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM3409HV-Q1: Ripple Voltage and conductive EMI

Part Number: LM3409HV-Q1
Other Parts Discussed in Thread: LM3409, PMP15004, TPS92515

Hello Group, 

I have been struggling with this for a bit and wanted to get this question out here. I see a lot of information in these forums, so I am eager to see what is said.

I have a circuit with 3 LM3409 drivers, but this issue is present with even one driver. There is a ripple voltage from 360mV to 420mV present on my Main power rail. This ripple is at the same frequency as my PWM switch controlling the current to the LED. I am trying to reduce the ripple with input capacitance but I am up to 440uF and now inrush current is becoming a concern. I have researched other causes and ringing is a possibility. I have added a snubber to the Mosfet (D to S). This did eliminate the ringing and improved ripple from my first prototype. 

1) Since this is a P Channel switch, should I still use the Snubber on the Mosfet, OR should it be placed to the ground?

2) I saw this "slew rate control for EMI" article:

https://e2e.ti.com/blogs_/b/powerhouse/posts/how-to-use-slew-rate-for-emi-control

I am wondering if I can use CSP and CSN with a series resistor to control slew rate and eliminate any other type of ring on the switch? I wanted to ask here before investing my time in a redesign. 

3) Are there other ways to combat this like isolating and filtering the three drivers from the other circuit. The rest of the PCB is just linear control and MCU that is controlling the PWM switch.

Circuit Characteristics:

1A Output 

Mosfet used:  SQS401EN-T1_GE3CT-ND

Inductor used: SRN8040TA-220M

Switching Frequency of IC: 500KHz

Switching frequency of PWM: 20KHz

47uF Output Capacitance 

  • Hello Ben,

    Using a worst case 1A to 0A current waveform at 20 KHz, 50% duty cycle I calculate a voltage excursion of 28 mV for 440 uF (this does not include any ESR, trace resistance or parasitic inductance.  What does the voltage ripple look like?

    1)  It can be placed across the MOSFET drain to Vin or even MOSFET drain to source as well as it could be placed across the freewheel diode (probably easier to do and it's connecting to GND.

    2)  This shows snubbers for a synchronous buck, there are similarities so it's similar.  There is no timing concern for efficiency or shoot through though.  To reduce the slew rate you would place a resistor between pin 6 and the MOSFET gate (this is what is shown in the EMI article you are looking at).  Adding resistance to the gate will affect efficiency.  CSN and CSP are for current sense, I wouldn't add any series resistance to these pins.

    3)  You may want to add an EMI filter to the input if you don't have one.  An example would be PMP15004.  This uses a TPS92515 for the buck converter.  It shows an input filter consisting of capacitance, differential mode inductance and capacitance.  This board was designed to be very small and meets CISPR 25 Class 3 conducted.

    The inductor you are using is not a fully shielded inductor so it will emit B-field that can couple into wires and other circuitry.  Also make sure the noise is real and not being picked up by the loop of the scope probe and ground.  You can check this by touching the scope probe to the ground connection, if there is noise it's being picked up in the loop of the scope probe and ground.

    Is your noise at the PWM frequency or switching frequency?  Are they noise spikes or?  If you can attach a scope picture that may help.

    Best Regards,

  • Irwin, thank you for the response.

    1) Thank you for the clarification, I have it working and will attach the scope of the switch

    2) Yes, of course. I did not see the resistor on the datasheet but should have provided the spot for one.  

    3) I will bury myself in that information now. This might give me the guidance I needed. 

    The inductor is an issue but I have cleared all traces from the inductor area to prevent this. I will move to a fully shielded inductor for more testing. 

    Below is the switching node vs ripple voltage. The ripple is pretty clear and not stray noise. The Ripple Voltage is @ the 20KHz frequency set by the MCU. 

    I want to add, this ripple is on the Main voltage rail or "Input" voltage. This is also seen in other circuits on the PCB as a 12V linear regulator and the 5V regulator for the MCU

  • Hi Ben,

    Yes, that's just voltage ripple from the PWM.  If this is going to a linear regulator most of that should be gone on the output.  What is the power source to this?

    Note and EMI filter will not get rid of this.  This is caused by source impedance and the amount of capacitance on the input.  The voltage ripple shouldn't be an issue unless you have sensitive circuitry also running from this.  Note that this will look quite different when all three channels are running.  If they are set to the same switching frequency they could synchronize or they could run at slightly different frequencies which will cause an appearance of a low frequency ripple as well as the 20 KHz ripple.

    Best Regards,

  • Irwin,

    First, thank you for the information in the previous reply. That has been most helpful. 

    My current source is a 12VDC variable power supply but the end application would be a vehicle battery. 

    I have had issues of conductive EMI from designs and figured the ripple voltage would be the issue because some designs have more interference and this directly relates to the amount of ripple voltage. I.E. Less ripple voltage = less radio interference in the vehicle.

    Luckily the LM3409 is a terrific driver and I am able to get all three to switch at the same time.

  • Hi Ben,

    So the 20 KHz ripple shouldn't affect the EMI unless there are harmonics above the 7th from that.  The switching portion on the oscilloscope plot does show some ripple on the down slope of the input voltage waveform so that might be an issue since it's above 150 KHz (500 KHz and harmonics).

    Best Regards,