Power supply for MCU

[404notfound]

Hi all,

I am having some issues with a TPS61023 from Texas Instrument.

I am using WEBENCH Power Designer from TI to design my power supply. (Link https://webench.ti.com/appinfo/webench/scripts/SDP.cgi?ID=596F065B4F17B3A1 to "my" design).

In theory, I can work from ~1.5V to 3.6V to boost voltage up to 3.3V.

However, below 2.5V, the IC "shuts off"...

I checked the feedback pin, it ranged from 0.3V bellow 2.5V input to 0.4;0.6V with input voltage above 3V.

This "power supply" powered an ESP 12F. You can check my PCB design here: https://pro.easyeda.com/editor#id=a962e7a75ae043559788beba2e07cb2e

Is there an obvious problem?
Thanks !

 

[Tony Stewart]

eda design not avail.
Did you substitute any reactive parts with higher ESR?

Cin = 238 ns
Cout = 175 ns

 

[404notfound]

Cin 100Uf ; Cout 47uF.
I'm trying to dumb evrything in a shunt resitor to see where it fail

 

[ronsimpson]

eda design not avail.

Did you login? You have to have an account.

 

[Tony Stewart]

I am logged in but didn't 2x click enough

It seems EDA substituted lower quality ceramic parts from non Japanese caps with high DF = 10%

 

[404notfound]

I am logged in but no editing rights(?)

right !

I got some value for debug ;
By using the refenrece in the drawing on easy EDA (see attached) with a led as load (~20mA) :
- Vbat @3.1V - VCC = 3.07V - Vfb = 0.554 V ; i =26mA
- Vbat @3.0V - VCC = 2.97V - Vfb = 0.541 V ; i = 16mA
- Vbat @2.8V - VCC = 2.81V - Vfb = 0.510 V; i = 4mA
- Vbat @2.5V - VCC = 2.57V - Vfb = 0.464 V; i =? (led is on-ish..)
- Vbat @2.3V - VCC = 2.57V - Vfb = 0.465 V; i =? (led is on-ish..)
- Vbat @2.0V - VCC = 2.57V - Vfb = 0.467 V; i =? (led is on-ish..)

AND, without a load, evrything fine..
- Vbat @3.0V - VCC = 3.320V - Vfb = 0.805 V;
- Vbat @2.5V - VCC = 3.321V - Vfb = ;
- Vbat @1.7V - VCC = 3.321V - Vfb = ;

Magic Smoke ; I kill it while probing Vfeedback, i'm soldering a new one.

Is the compound resistor the issue ? Like RFBT + Rload = Some wierd value ?

 

[404notfound]

right !

I got some value for debug ;
By using the refenrece in the drawing on easy EDA (see attached) with a led as load (~20mA) :
- Vbat @3.1V - VCC = 3.07V - Vfb = 0.554 V ; i =26mA
- Vbat @3.0V - VCC = 2.97V - Vfb = 0.541 V ; i = 16mA
- Vbat @2.8V - VCC = 2.81V - Vfb = 0.510 V; i = 4mA
- Vbat @2.5V - VCC = 2.57V - Vfb = 0.464 V; i =? (led is on-ish..)
- Vbat @2.3V - VCC = 2.57V - Vfb = 0.465 V; i =? (led is on-ish..)
- Vbat @2.0V - VCC = 2.57V - Vfb = 0.467 V; i =? (led is on-ish..)

AND, without a load, evrything fine..
- Vbat @3.0V - VCC = 3.320V - Vfb = 0.805 V;
- Vbat @2.5V - VCC = 3.321V - Vfb = ;
- Vbat @1.7V - VCC = 3.321V - Vfb = ;

Magic Smoke ; I kill it while probing Vfeedback, i'm soldering a new one.

Is the compound resistor the issue ? Like RFBT + Rload = Some wierd value ?

In the datasheet, i dont find the relationship btw Vfb and Vcc ;
Can i make a linear approximation with two points saying something like :
- For Vcc = 3.3v , Vfb must equal 0.8v
- For Vcc = 2.8v, Vfb must equal 0.5v

And so if this is "acceptable", i just need to tweek my voltage div ?

 

[Tony Stewart]

No. Vfb is always constant and the FB R ratio determines the output voltage. 3.3V here.

It seems efficiency is poor from choice of reactive parts.

 

[404notfound]

ok so in therory, the comput'Omatic from WEBENCH Power Designer said 85% efficiency.
Even 50%, i dont care, this project is online for 1520ms evry 35min.

Is there a way to correct my board ?

 

[Tony Stewart]

I don’t know how pro eda can do simulations without valid parts or specs. Or if this is a prototype, what p/n we’re used for all LCs