WEBENCH® Tools/TLV61220: Output voltage "sagging" with 30-40 mA load

Hi Ethan:

May you check the schematic firstly? It seems nothing connect to the Vbat. It maybe a typo as the layout has something there.

You could try to connect a small size (0.1uF) cap between the Vout and GND pin, to make the loop as short as possible.

Besides, I'm in business travel these day. Please forgive me if any relay in delay.

Hi Minqiu!

Thanks for your prompt response.

Yes, my apologies, that schematic was wrong, as you note. This is the correct one:

Thanks for your suggestion re a small capacitor between VOUT and GND. Is that capacitor necessary because my C15 output capacitor is a little too far from the TLV61220? I can easily re-do my board layout if that might help. But I like the idea of adding more capacitance, perhaps as a short-term fix to confirm that's the issue.

Thanks again

Ethan

Hi:

Your C15 output capacitor is a little too far from the TLV61220, that's why I recommend to add a small capacitance to make the road shorter.

You can solder it to see if it has some helps.

That worked, thanks!

Also spinning a new board design with thicker traces and improved layout.

Hi Minqiu!

Well I got the new board in, and it has the same issue. Below please see my new layout. The wide traces are mostly 30 mil, with one 22 mil. The narrower traces are all 8 mil. The schematic and BOM are the same as last time.

How about I just add a 0.1uF output capacitor to my board layout?

Thanks as always,

Ethan

Hi Ethan:

I remember the circuit worked when you add a small capacitance, right?  Generally speaking, the small package capacitance has less  parasitic parameter. So I'll recommend you to add a small cap close to the converter. 

Hi Minqiu!

Are you saying I might want to switch to a smaller package size for my input and output capacitors? Is 0603 small enough? It seems that's the package size in use on the evaluation module.

Hi Ethan:

Nope, It's better to add another small size capacitance to make the shortest line. And the line for feedback resistance could be wider.

And additionally, I want to double check with you one more time. Did the circuit work well when you add a smaller size close to the converter before?

Besides, there is a application note introduce how to do the better layout: http://www.ti.com/lit/an/slva773/slva773.pdf.

Hi again!

Yes, you're correct that we have seen this circuit work properly, but only under the circumstance shown below, where my colleague soldered in a .1 uF capacitor directly on top of the TLV61220, using little leads made from the leads of a thru-hole resistor. This is difficult to photograph, but below are my best efforts:

Obviously this isn't a viable design solution for production :)

I tried to mimic the above circuit by soldering an 0805 0.1 uF cap on top of my C15 output cap--stacked right on top of it and soldered to it--but this didn't resolve the issue. This makes me dubious that adding a second output cap is going to resolve the issue. But I'm willing to try another board spin to find out!

I also fattened up my feedback pin traces. I have some reason to think this will help, as you say, because another thing I've tried is changing the feedback pin resistor divider resistors from 1M/100k to 100k/10k and even 10k/1k, which did improve things slightly--the Vout now sags to 4.9 instead of 4.8V under the same 30-40 mA load.

The resulting revised layout is below. Please do share your comments on it! if you can! Thanks as always!

Oh I forgot to add, C28 would be a 0.1 uF as you suggested.

Hi Ethan:

So sorry that I'm in oversea travel these days. Please allow me reply next Monday. Thanks for your patient waiting.

Hi Ethan:

Thanks for your kindly waiting. 

May I know the status now?

It's better to exchange the place between C28 and C15. And it's recommended to use a small size cap for 0.1uF.

Besides, the line for FB road could be wider if there is no other conflict.

Awesome, I'll make up a revised board layout.

When you say a "small size" capacitor, do you mean a smaller package size like 0603? Or a small capacitance value?

Is it better to switch to 0603 for all the caps? I would rather do that than stay with 0805, if there's some chance of it resolving the issue.

Here's a version with 0805 capacitors still:

Hi Ethan:

Small size means package size like 0603,0402, with less parasitic parameter.

You could use wider line to connect the R14 to GND, and the PIN of EN, as there is enough place, and no harm to choose wider line. Same suggestions to other line with enough place. 

OK I'm changing all my caps to 0603. Reducing parasitic inductance seems like a good thing.

I fattened up a couple traces--the SWVBAT line on the bottom side and the short trace between R14 and its GND via.

How's this look?

Hi Ethan:

So thanks for you taking my suggestion into considered. The last thing I consider is if the Cout has enough effective value. As the MLCC has DC bias character. You could check with the datasheet of the MLCC. If it's not enough, you may need to enlarger it. Otherwise, I don't see other obvious risk.

OK I'm going to order the PCB for this layout. Delivery takes 4-7 days. If there's still an issue, I can always try changing capacitors without changing the board layout.

I've assembled a couple of tables showing most of the 0603 capacitor options suggested by WEBENCH for this circuit. COuld you offer some guidance on selecting the best options? I tried to include all the options with higher DC-bias derated capacitance than the old 0805 capacitors, and lower ESR. It seems all 0603 capacitors have higher ESR than 0805 capacitors, but not necessarily much higher.

These are the options for Cin:

And for Cout:

All of the data above are taken from the WEBENCH component selection tool for the TLV61220DBVR, with my requirements entered in.

谢谢你！ 很高兴我收到你的帮助