TPS62933: audible noise

Hi Fred,

Duo to National Day holiday, my team will give you response early 10/8.

Hi Harry,

got it , please respond as soon as you're available, thanks

Hi Fred,

Thanks for reaching out! Reply will be given once my colleague come back to office!

BRs

Zixu

Hi Fred

Would you please do me a favor to capture the output voltage and inductor current so that I can check whether there is any low frequency noise coupled. Could you figure out which device is noisy, inductor or capacitor? And what is the loading? Why do we have such wide range frequency in the loading?

Thanks.

BR

Ruby

HI Ruby,

its' for LED backlight PWM signal , and  I already show the inductor current in the previous graph which is same as output current.

for the loading like I said , you can assume it's a normal step load signal with 50% duty cycle and frequency ranging from 60Hz~20KHz,

The switching frequency of SW seems  higher than 20KHz whether at low side or high side of load, but do you think this noise is coming from the frequency variation ( the load) , which is changing according to the load frequency that may below 20KHz?

and no, EVM is too small I can't tell which component is the source.

I'll update output voltage .

Hi Fred

I would like to check whether this noise is caused by the coupled loading frequency or our spread spectrum feature. Changing the inductor and the capacitor material can help. Adding output cap can also help. If the spread spectrum feature caused the noise, TPS62933F which is B2B with TPS62933 and does not have the spread spectrum feature can help here.

Hope this helps.

BR

Ruby

Hi Ruby,

I don't think it's spread spectrum ,I try constant load (eliminate load couple factor)  last week ,  but can't hear anything.

so if it's load coupled , does  F version also helps?

since the fsw is fixed.

Regards,

Fred

Hi Fred

I am afraid not. Regarding this case, I would suspect that the output cap makes the noise since it provides the ripple current to the loading. As a result, you may enlarge the output cap (or choose a cap with higher voltage tolerance to get a higher effective value), reducing the ripple current. Or change the material of the output cap may help here.

Hope this helps.

BR

Ruby

Hi  Ruby,

I'll try to capture the Vout and see if there's a low frequency ripple there tomorrow since we're in holiday today.

so you don't think this frequency variation will create noise, right?

Hi Fred 

As we talked offline, we may have some actions here:

1. Use TPS62933F to check whether there is any noise here.

2. Capture the output voltage during load transient

3. Check the output cap voltage tolerance.

You may also try to add one Cff here to check whether it works.

If there is any more question, please let me know.

Thanks

BR

Ruby

Hi  Ruby,

thanks,

update output ripple for 20KHz and 100Hz for reference, yellow is output voltage and green is step load current, ignore the blue.

the ripple is about 1.5V, do you think this is large enough to create audible noise?

BTW, both measurement are without dummy load, but I think this doesn't matter much

Hi Fred

Yes, the peak to peak value is high enough for the capacitor to get the noise.

Hope this helps.

BR

Ruby

Hi Ruby,

per checked, the F version still has audible noise.

the slew rate of the step load is in ns scale shown below 

1. do you recommend add E-cap or MLCC

2. for this fast slew rate , I add 820pF Cff but no helpful at all. any other value recommended?

Hi Fred,

My colleague Ruby will give you response by tomorrow!

BRs

Zixu

Hi Ruby,

Cff 820pF is not helping.

 just tried it. CFF only help on low frequency . I think that's why