TPS564247: TPS563231DRLR + TPS564247DRLR selection of capacitors and total output ripple

Hi Hossam,

My colleague Shipeng will give you response soon. Thanks!

BRs

Lucia

Hi Hossam,

We don't need polymer cap or electrolytic cap at output side.

For TPS564247, the ripple test result can be found in datasheet. I think you should use at least two 22uF to achieve 15mV ripple. And I advise to add one more position as backup.

For TPS563231, pls note that this is PSM mode device so the ripple will be larger at light load. You may need to choose a FCCM mode device to get a low ripple.

For transient response, it really depends on the load current slew rate so it's better to test on real device.

Thanks.

Regards,

Shipeng

hello,

I have enough space so I can use multiple output caps in parallel. I currently have 3x22u 1210 X5R caps in parallel with 5x10u 0805 X5R caps... + a small 1u or 100nf in parallel too. I also added a 10u elec cap with high esr (3 ohms) for damping. please check this and report if it is gonna be ok.

my application won't have huge transient anyway... rated at 2 amps full load but could increase if additions were installed but generally won't be more than 3 amps full load. load is expected to be constant so now big transients. the transients (big or small) are expected to happen at 50 or 60 hz frequency since the application is a retro gaming console which is ntsc or pal.

I wanted to get DCAP3, so TPS564247 is my preferred part for 5v rail @ 2-3 amps but i need 3.3v at 0.6 amps rated (bump it up to 1.5 amps max) which is why I got a cheaper part (TPS563231) but i didn't know it has that eco mode for light load efficiency. 

if i want to use TPS563231 but achieve very good low ripple results, how many milli-amps of load is required? i mean to get out of "light load" situation.

I know newer parts are available but I really want something available at lcsc.com specifically... please advise.

Hi Hossam,

I see you use ~100uF ceramic caps and a 10uF 3Ohm elec cap. I run a simulation by using a 1.5uH inductor and I get 37kHz, 39° bode plot. So I think it's OK.

If you want to use TPS563231 but achieve very good low ripple results, it needs more than 0.6A load to enter CCM mode. So I advise to choose a FCCM mode device.

Thanks.

Regards,

Shipeng

hello,

can you please check with 2.2 up to 4.7u inductor with tps564247? or will 1.5uH inductor be enough to have < 10 mv p-p ripple output and excellent transient? i source all my passives from lcsc, i could get very low DC resistance inductor for cheap... but i need to understand the relationship between inductor value and total output ripple figure... will 1.5u be enough or is it good to pick say 2.2, 3.3, 4.7u? with the same capacitance i choose.

as for the 3.3v i think TPS562208 is good and available cheaply. it is dcap 2 not 3... it will only need up to 0.6 amps, maybe 1 amps in very rare cases. it is very cheap at lcsc which is a plus.

i think for small current i can pick small inductor and capacitors... 2x22u 1210 + 3x 10u 0805 + 1u 0402 + 10u 3 ohm esr elec is gonna be fine. please check.

last thing, when picking capacitance should i go with nominal value or actual capacitance after dc bias?

Note: if you have a software where you calculate the bode response and stuff like this please give it to me.

Hi Hossam,

My colleague Shipeng will give you response soon. Thanks!

BRs

Lucia

Hi Hossam,

Large inductance will cause better ripple but worse transient response, so you need to trade off. You can choose inductor by the recommendation table in datasheet.

TPS562208 is suitable for you but the ripple will not change a lot with load current, so you should not decrease capacitors.

You should picking capacitance by actual capacitance after dc bias.

You can use Pspice model on ti.com to run a simulation.

Thanks.

Regards,

Shipeng

Hello Cheng,

What do you recommend I do for getting best ripple (10-15 mv p-p) while having good transient? I can put more capacitance if there is a need for smaller inductor. getting multiples in parallel will make esr very low which enhances transient.

for tps564247, i see evaluation module having 8 mv p-p ripple with just 0.82u inductor and very small caps, this had good transient too. so I can use like 1u or 1.5u as you suggested but with larger caps and more caps in parallel. this theoretically will give us better ripple and better transient response on expense of more cost of more caps... i have no problem with this!

However, that was for 1v output but for 5v output datasheet says minimum is 1.5u and typical is 1.8 with maximum being 4.7. i can use power inductors with values of 1.5u and 2.2u... also 3.3 and 4.7 but as you mentioned those could give bad transients.

so kindly do simulation by using 1.5u and 2.2u inductors (dcr=5mohm), combined with the following caps:

22u 1210 X5R x3

10u 0805 X5R x5 (can increase them to any required number)

1u 0402 x1

10u electrolytic with 3 ohms esr (for damping only).

and please report back to me the final result of ripple figure and transient, not necessarily fully detailed though. you could compensate for dc bias in an approximate way.

you could add more caps of similar types or even add big 470u polymer cap (8moHm esr) or multiples of it... basically anything to make it work as intended... while keeping inductor small enough.

if big 22u 1210 not good enough i can change them to 1206 or lower but i find them good because of lower ESL and lower DC Bias,.

as for tps562208, i told you the caps i intend to use and i can increase them. please tell me the optimal choice for low ripple... since there will be no big transient events or so.

i am not really good in spice, i thought there was a tool which calculates stuff automatically.

Hi Hossam,

Shipeng will check and get back to you soon. Thanks!

BRs

Lucia

Hi Hossam,

I run the simulation by 1.5uH and 3x 20.4uF,3.1mΩ, 5x 4.8uF, 1.8mΩ. And feedback is 220k, 30k, Cff=100pF. And get 143kHz, 59° bode plot. I run the simulation by 2.2uH and 3x 20.4uF,3.1mΩ, 5x 4.8uF, 1.8mΩ. And feedback is 220k, 30k, Cff=100pF. And get 101.7kHz, 65° bode plot.

For TPS562208,  you can use our recommendation table in datasheet if no big transient events.

Thanks.

Regards,

Shipeng

hello,

thanks for your recommendations! how much ripple and noise did you get in your 2 simulations?

plus, are they safe to use? like what is the safe bode plot phase and frequency limits? so I can know what to choose.

looks like you used actual capacitance after dc bias, nice.

Hi Hossam,

I see the ripple at 4A load, it's less than 10mV. My simulation cannot see any noise. And the bode plot is safe. Thanks.

Regards,

Shipeng

hello,

thanks for your effort.

as for bode plots, what is the limits of safety? both phase and frequency?

if you can, try adding 2x 470u polymer caps 8mohm esr to the capacitor bank mentioned above and see if it is safe.

also, please state the transient response of both results you got... to see if 2.2u vs 1.5u is good or not.

thanks!

Hi Hossam,

Thanks for your feedback. Shipeng will get back to you during these days. Thanks!

BRs

Lucia

Hi Hossam,

If adding such polymer caps, the crossover frequency will be 13.8kHz with 2.2uH inductor, 17.5kHz with 1.5uH inductor. The overshoot/undershoot voltage is 20mV both 1.5uH and 2.2uH(Transient: 0A-4A-0A, 2A/us). As there's large capacitance here, it's better than without polymer cap.

Thanks.

Regards,

Shipeng

hello,

for such polymer caps and their cross over frequencies, are they safe to use? or better stick with pure ceramics?
what is the total ripple you see when using them vs just ceramics?

thanks

hello,

by safe i mean it will not oscillate or go out of control and lose regulation.

hello Shipeng,

i read that dcap3 has maximum recommended limit of about 100uF for this regulator. so i assumed above this value it will become unstable and give problems.

having significantly larger capacitors like these well above the maximum limit does not seem to do any harm, so why the warning in datasheet and all that about stability and bode response...etc?

eventually for as low ripple and noise as possible, do you recommend I stick with just the ceramic caps or it is good to add those 2 extra big polymers? any downside of both methods? 

thanks 

Hello sir,

this was my concern to begin with. since i cannot do bode plots and stuff like this, i asked your help. 

so Cff in case of just ceramics is going to be 100 pF, but with larger polymer caps in addition to ceramics... how much can it be? I can use 2x 470u 7~8 mohm esr polymer caps for the 5v output how can i pick a proper value without getting into simulation and bode plots?

I am totally fine with pure ceramics but i fear that it could not be enough to get ripple down to < 10mv p-p and filter all noise if exist. what did the addition of those polymers do to total output ripple and transient in your simulation? 

in my application, huge transients are not really expected but low ripple and noise is a must. thus, do you recommend sticking with those ceramics in parallel + 100pF Cff or add the polymer caps and increase the Cff to something like 220pF?

Hello sir,

I really like these DCAP3 regulators mentioned here, and they fit my budget. I used TPS62913 before but could not achieve the extremely low noise figure I wanted. plus they are expensive for my project.

I have Meanwell power module as input to the buck regulators, it has common-mode choke and I follows it with bulk polymer and ceramic caps to filter everything... so with good layout it is safe to say that very low noise can be achieved.

the challenge is ripple and transient. since you confirmed ripple below 10mv at 4 amps and very nice transient then i guess it can be ok. my end use has power sips (or transient events) at 50 or 60 hz since it ntsc or pal device... but no huge transient or jumps are expected since the full load is 5v at just 2 amps anyway.

what i want is the transient to be good enough so that it + ripple + noise to be < 10 mv p-p total figure. 

For this prototype i guess i will go with 7x 10u 0805 + 3x 22u 1210 + 1x 100n 0402 + 10u 3 ohms esr elec cap (for damping),,, all with 150pF 0402 Cff. I will test to see if all this works fine with my device. having very low total esr is always a plus for low ripple and good transient. i guess this will show if we really need bigger caps at output or not. DCAP3 supposed to allow good ripple and transient without big output caps right?

as for bode plot, can you guide me to which hardware tools are used to measure it on real circuit? prefer if budget solution exists. 

also, what online simulation tool do you have and use with this regulator?

best regards

hello,

sorry to re-open this, but i have yet another design which requires 3.3v at 3 amps and 5v at 0.5-1 amp. I want to use TPS564247 but require very low noise and ripple PLUS stable operation, no humming sound, and good transient.

 my input is 12v via meanwell module which should be around 70mv p-p worst case.

here are my main questions:

1- please tell me if above output caps (full ceramic) are good enough or i must add polymer caps? or maybe reduce them? my goal is to achieve about 10mv p-p give or take.

2- what type of input caps are necessary? i can easily add bigger polymers combined with ceramics.

3 - is there any damping or compensation needed? like having electrolytic caps at input and output? if added a small one with big esr, or bigger one with big esr, how will that affect the circuit output behavior? i ask this since some circuit may not be stable without damping especially starting to produce humming or buzzing sound due to ceramics in parallel.

4- where to route feedback from? the output connector itself or just the output capacitor

5- can 4 layer board with layers 2, 3, and 4 being fully ground help with noise and stuff?

6- how do you recommend i put vias? i mean the output caps vias... or just let them be without vias to ensure low inductance route.

best regards

Hi Hossam,

You can read the datasheet TPS564247 data sheet, product information and support | TI.com and EVM User's Guide https://www.ti.com/lit/ug/sluucj5/sluucj5.pdf first. Most of your questions can be answered. You can buy or apply for a EVM board to test at your side to try your idea. That's why we always provide EVM board for each product.

Thanks.

hello,

i have read those numerous times but i am still not very sure. i know it could be a hassle but please answer my questions so i can take your answer as a reference for future.

getting evm is not an option for me right now, i am designing already and sure that i need this part. 

If you don't want to evaluate on EVM, you can add capacitors as much as you can (including damping) when layout. Then when you get PCB, you can try to remove capacitors to see if problem would occur.

For layout, you can copy our EVM board.

4 layer board with layers 2, 3, and 4 being fully ground  can help with noise.

hello,

i will try adding caps and remove them to see what happens. however, in your tests did you need any of these besides couple of ceramics? i mean, is dcap3 good enough to be very stable with very low esr output caps? 

i tried doing online simulation in your site powerdesigner but it tells me that simulation is not available for this design.

however, if you can give me actual oscilloscope captures of the device if you have them, that would be fantastic. real data always better. since your evaluation board only has data for 1v output which is not what i need.

i really want low noise and good enough transient, my usual method involves using bulk parallel polymers with ceramics but i read in datasheet that high capacitance values may make the device unstable, what is your opinion on this? i can add Cff of 220pf value for example but what negative effect could it do if it didn't help?

thanks 

hello,

please check my latest reply.

also, i want to send you my layout for checking. how can i do it in a way suitable to you?

regards,