# PWM Noise led driver

I'm working on an constant current led driver, in particular the TPS92512 from T.I.

There's audible noise when PWM dimming.

I think that the problem is on the inductor, but I don't know how to solve it, I mean, do I have to pick an inductor with higher DCR, higher Isat, or what?

Here is the LED driver. I changed the inductor to SRR0805-150M.

• What's the switching frequency? How big is your PCB? Can you post a picture of your assembly? – Enric Blanco Mar 8 '17 at 14:39
• The PWM frequency is 500 Hz. The switching frequency is 696 kHz ( not measured but given by T.I. configurator ) – EnricoPallazzo Mar 8 '17 at 14:50
• Try shifting the PWM frequency to see if this increases or decreases the audible noise. Do you have to PWM at 500Hz? – M D Mar 8 '17 at 14:53
• "audible noise" kind of vague. High pitch? At what PWM duty cycle? 100%? 50%? In order to determine the problem the Output Voltage (or number of LEDs & part number) is important. What is the Max Current? What is the target Ripple Current? When you lower the PWM frequency what is the modulation frequency of the flicker? 10-20hz? What is the UVLO voltage? – Misunderstood Mar 9 '17 at 18:29

Piezoelectric effect in the ceramic capacitor due to the 500 Hz PWM should be your main suspect, although you can't rule out the role of the inductor.

Look at this excerpt from MPS Application Note AN021:

The familiar PWM dimming frequency range is less than 1kHz. Due to the piezoelectric effect of the ceramic capacitors and the oscillation of wire winding coil, the discrete low frequency (relative to the WLED driver’s switching frequency) dimming cycles can possibly cause audible noise in the system.

Debug your board by gently pressing all ceramic capacitors, one at a time, with something non-conductive and anti-static. The purpose of this test is to prevent the capacitors from vibrating. If noise ceases or greatly drops when pressing a capacitor, then you'll have found the culprit. Beware: you may found that not only one, but several capacitors contribute to the noise

Start with the output capacitor, $C_{out}$ P/N GRM21BR71H105KA12L. This part is a 0805 MLCC X7R. As you can see in this paper from Kemet, any multilayer ceramic capacitor (MLCC) with dielectric X7R (or any class 2/3 dielectric) can generate perceptible acoustic noise under certain circumstances:

Also note that the Murata capacitor catalog warns you that their GRMxx capacitors aren't "anti-noise" - a reference to acoustic noise generated due to piezoelectric effect:

So, which are the main options do you have to prevent this effect and its noise?

1. Replace the culprit with another capacitor, either a ceramic one that does not exhibit this effect, or a completely different one material-wise (electrolytic or film).

2. Stiffen your PCB by mechanical means in the vicinity of the culprit, in order to prevent the board from vibrating and acting as a speaker. If the inductor is the cause, this also applies.

3. Use a driver that can do PWM above the audible range.

There are other options, but you may want to research them for yourself.

• Hi Enric, thank you for your reply. OK I'll try to do this and I'll let you know. – EnricoPallazzo Mar 8 '17 at 16:19

PWM Noise in SMD wirewound core chokes with high pulse currents ( several Amps) is a common problem especially when the PWM rate is in the most sensitive midrange of the hearing range.

To alleviate this inherent problem the PWM rate can be moved towards 100Hz with the undesirable flicker tradeoffs

A better choice may be to change selection of choke to low acoustic noise types with some tradeoffs on uH size of composite printed inductors rated for several Amps.

e.g. EROCORE Ultra low buzz noise. LPC & PIHD style

http://www.core.com.tw/new_product.asp?id=xx&iPage=3

e.g.              （μH） (A) (Asat) (mΩ) (mΩ)Max.
LPC0312H-1R0M 1±20% 4   5     32    38


Wurth also offers these styles .

• Hi Tony, thank you for your reply. Yes if I change PWM to 100 Hz the noise is less annoying, but there's flickering, so I need a PWM frequency greater than 300 Hz where I don't see a flickering effect. I'll try to change the inductor – EnricoPallazzo Mar 8 '17 at 16:21
• I once had an SMD choke on PWM buck regulator sound like running water (chaos theory noise instability) with a low power boost PFM regulator load in a rather noisy lab with fans. I suppressed it with a lower ESR Cap in between. – Tony Stewart EE75 Mar 8 '17 at 18:14
• @EnricoPallazzo you should change the inductor if, and only if, you know you have made an error in its selection. To just randomly choose some other arbitrary value is just wrong. You need to justify why you are going to change by calculating its value based on the objective characteristics of the project (e.g. forward voltage, input voltage overhead, ripple, efficiency, switching frequency, current requirements, PWM modulation and duty cycles). Without knowing the number of LEDs and their part number, all anyone can do here is guess. Guesswork does not involved in product design. – Misunderstood Mar 9 '17 at 23:17
• It's not an error, just a nuisance that wirewound chokes make noise with audio PWM, try to stick to same parameters – Tony Stewart EE75 Mar 9 '17 at 23:22
• @TonyStewart.EEsince'75 good guess not knowing anything about the " audible noise", a description of the noise would have been a help. Using a different brand of inductor sounds like a kludge cover up rather than looking for a possible design error causing the noise. – Misunderstood Mar 9 '17 at 23:25

This is the schematic TI gave me.

• Vin 24-36v
• Vout 18.6v
• Iout .89A
• Ripple 10%

If you provide me a link to the tool you use to configure this driver, I would like to take a look at it.

You have Iadj (pin 6) pulled high. So the driver is going to try and output the maximum current of 2.5 Amp. The PWM from your micro has to do all the dimming.

If for some reason your PWM fails, the current goes to 2.5 Amp and you will likely lose all your LEDs.

Iadj should throttle the current back to 0.89A and if you want it to stay at 0.89A you do not have to do any PWM.

Always set the max output current to your maximum value. Then use the PWM only to make and adjustments to dim below the max current.

The inductor you had that you replaced. It was a much better inductor than the Bourns you are now using. The Bourns has about 10x more DC resistance.

I do not know why you are only using a 15µH inductor it seems too low. If the TI app suggested a 15µH inductor you fed the app the wrong parameters or there is a bug in the app.

You want the driver to operate in the continuous conductive mode. In the discontinuous mode your inductor can produce ringing, your LEDs may flicker. If you can see flicker there is a serious problem in the design. The human eye and mind cannot perceive 100-200hz flicker, it is measured not seen.

If the amplitude of the ringing in the inductor is so high you can hear the ringing, there is a problem in your design, not the manufacturer of the inductor.

To operate in DCM it is better to have a low value inductor. The ripple current is not good for the life of the LED.

The inductor is what keeps the current flowing when the switching voltage swings.

The TI documentation is very good and very detailed. They go into detail on each and every component. You need to understand why they do that and do it.

This is not simple stuff. There are resources.

You need to take a look at your UVLO too. The following link will explain that.

The following is from TI's app note Understanding Buck Power Stages in Switchmode Power Supplies

Section 5, Component Selection is an easy read. Well, compared to the prior sections.

In switching power supply power stages, the function of inductors is to store energy. The energy is stored in their magnetic field due to the current flowing. Thus, qualitatively, the function of an inductor is usually to attempt to maintain a constant current or sometimes to limit the rate of change of current flow.

The value of output inductance of a buck power stage is generally selected to limit the peak-to-peak ripple current flowing in it. In doing so, the power stage’s mode of operation, continuous or discontinuous, is determined. The inductor ripple current is directly proportional to the applied voltage and the time that the voltage is applied, and it is inversely proportional to its inductance.

In addition to the inductance, other important factors to be considered when selecting the inductor are its maximum dc or peak current and maximum operating frequency. Using the inductor within its dc current rating is important to insure that it does not overheat or saturate. Operating the inductor at less than its maximum frequency rating insures that the maximum core loss is not exceeded, resulting in overheating or saturation.

There are many types of inductors available; the most popular core materials are ferrites and powdered iron. Bobbin or rod-core inductors are readily available and inexpensive, but care must be exercised in using them because they are more likely to cause noise problems than are other shapes. Custom designs are also feasible, provided the volumes are sufficiently high.

Current flowing through an inductor causes power dissipation due to the inductor’s dc resistance; the power dissipation is easily calculated. Power is also dissipated in the inductor’s core due to the flux swing caused by the ac voltage applied across it but this information is rarely directly given in manufacturer’s data sheets. Occasionally, the inductor’s maximum operating frequency and/or applied volt-seconds ratings give the designer some guidance regarding core loss. The power dissipation causes a temperature increase in the inductor. Excessive temperature can cause degradation in the insulation of the winding and also cause increased core loss. Care should be exercised to insure all the inductor’s maximum ratings are not exceeded

• thank you. Here you can find the tool : [link] (webench.ti.com/webench5/power/…). There are 6 leds in series XPGWHT-L1-0000-00H51, Iled = 1A Vmin = 24V Vmax = 26 V tAmb = 40°C . There's the RSENSE that "hardware limit" the current on the Leds. When PWM @ 100% I measure 0.9A on the LED – EnricoPallazzo Mar 14 '17 at 12:34
• @ Misunderstood the different value for the inductor comes from the ripple % . I have a 30% ripple and this gave me a 15uH inductor. Do you think it's better to have a lower ripple % and so increase the value of the inductor? Thanks – EnricoPallazzo Mar 15 '17 at 11:25
• Yes, 10% ripple is much better for the LEDs, the inductor ringing, and keeping it in Continuous Conduction Mode. That is why the inductor was making noise. – Misunderstood Mar 16 '17 at 5:19
• @ Misunderstood, I'll change the inductor and I'll let you know!thank you! – EnricoPallazzo Mar 17 '17 at 7:14
• There is more than just the inductor. When you change the inductor you need to change the switching freq. Check: RT/CLK, UVLO, RSENSE, Iadj – Misunderstood Mar 17 '17 at 13:50