TPS62740: How to eliminate Ripples from DC-DC converter using PI filter

Hi,

This looks like the device is operating in the PFM mode which is for lower load currents and that's the reason for that type of ripple.
If you are already using 10uF for Cin/Cout and 2,2uH for the Inductance then the best way to improve the ripple further would be to increase the load current or replace the TPS62740 with a device which has the 'forced PWM mode' function.
If you apply another filter that could lead to instability in the feedback loop.

Could you send me the schematic? and let me know what the load current was during that measurement?

Regards,
Alex

Hi Alex,

Thank you for your response.

This is the DC-DC schematic we are implemented on our device. 

V(in)=3.6V (Battery Operated)

V(out)=2.5V

I(out)Min=5mA

I(out)Max=100mA.

usually our device consumes less than 5mA,  only during wireless communication it consumes up-to 100mA for short period. Can you assist us on this issue.

Regards,

Vinayak P

Hi Vinayak,

Everything looks fine.

In order to improve the output ripple you could use for the Cout = 22uF and for the L 3,3uH.

As per datasheet:

Regards,
Alex

Okay I'll check and update you. will it be okay if component size is other than 0805 ?

That should be no problem you just have to make sure that the effective capacitance is enough.

Okay Got it.

We will check and Update it here.

Thank You

Regards,

Vinayak

Hi Alex,

I am checking the DC-DC converter for our Circuit, if the load is low ( i.e. 5mA) we have to select DC-DC converter with forced PWM rather going with PFM. because I observed most of the DC-DC Converters are switching to PFM for power save mode. How do I select the new DC-DC for lower loads.

Thanks, 

Vinayak

Hi Vinayak,

Unfortunately I could not find an alternative which would be pin to pin compatible.

Otherwise you could check this two devices which would have similar specifications as the TPS62740 and the 'forced PWM mode' function:

LM3676 data sheet, product information and support | TI.com

TPS62240 data sheet, product information and support | TI.com

Lm3676 would still be available in case that would be important and the TPS62240 would fit better but is currently out of stock and both devices are not pin to pin compatible.

Regards,
Alex

Hi Alex,

Thank you for the updates,

we actually using it for ultrasonic water meter, we are operating our device with Battery of 3.6V , we need Buck converter with ultra low Iq (nA), After the searching in TI's Website, we were interested in TPS62840 which has 60nA and used in smart-meter applications. 

which one will be more suitable for the metering devices ? LM3676 , TPS62240 or TPS62840 ?

And I also observed in TPS62840's Simulation if we select Forced PWM the efficiency get reduced to 2.616 % from 91.703 %. why so much of difference ?

and by selecting Forced -PWM mode will it affect my circuit, Battery or device's performance ? 

Regards,

Vinayak

Hi Vinayak,

Considering you your battery powered system the TPS62840 could be the better choice.

Forced PWM mode devices usually have a low efficiency for lower loads.

But you now have to consider what is more important to you: 

Very low ripple with lower efficiency (using a forced PWM mode device) or

Keeping like it is with low ripple and good efficiency and in addition if you use for Cout = 22uF and for the L 3,3uH. that could improve the 2,8mV ripple.

Regards,
Alex

Hi Alex,

As we are working on TPS62740 will it be good to continue with our existing device ?, we will check the Cout=22uF and L3.3uH with our device. can you suggest Pi filter for our design ? previous day I was looking in TPS62840's  DC-DC Technical Documents and I found that TI is using Pi circuit with DC-DC converter for lower loads, so can you suggest us values for them ? 

As in  the TI's documents  suggested to use as below. L1 value as 600E (Ferrite bead) will this work my circuit ( Vout =2.5V, Iout(min)=200uA  ,Iout(max)=100mA ) 

Regards,

Vinayak

Hi Vinayak,

Yes it will be probably the best to keep the TPS62740.

Yes that second filter that you have there will improve your ripple.

But here is a guide on how to design a 2nd stage filter with all calculations needed to get the values of the capacitances and inductances.

Regards,
Alex

Alex Da Silva said:

But here is a guide on how to design a 2nd stage filter with all calculations needed to get the values of the capacitances and inductances.

You mean this or any other values?

Thanks,

Vinayak

Hi Vinayak,

Other Values the guide which I sent you has a step by step guide in which you can pick the values according to your switching frequency and its not an PI filter it is an 2nd stage LC filter with a damper resistor but it would have a better gain margin.

Now you could try both filtering methods and see which one fit your design better.

Regards,
Alex

Hi Alex,

These are the results of using Cout= 22uF, L=3.3uH.

the waveforms are different from old values. can we say the output is improved as compared to old values ?

Need your suggestion  on this.

Regards,

Vinayak

Hi Vinayal,

Yes the new diagram looks better.

Is that enough for you or do you want to use a filter aswell?
If you decide to use a filter you have to change the L and Cout again because if you just add a filter now you would exceed the max inductance and max capacitance allowed. 

I would recommend you this guide in order to design a filter

Design a second-stage filter for sensitive applications - Power management - Technical articles - TI E2E support forums

Regards,
Alex

Okay we will check and use either new L C values or Pi filter.

Thanks, 

Vinayak

Hey Alex,

What is the significance of R damp circuit after Pi Filter ? 

How these will help the circuit ?

Regards,

Vinayak

Hi Vinayak,

The R damp is used to damp the Q value of the 2nd filter. 
A high Q will lead to a low gain margin and consequently unstable operation, such as an undamped response to a step input.

Here you will find how to select the right damper:
Design a second-stage filter for sensitive applications - Power management - Technical articles - TI E2E support forums

Regards,
Alex