BQ25570 VOUT Problem

Where are the pulsed loads applied, VSTOR or VBAT? How much current are in the pulsed loads?

Pulsed loads are applied at VOUT. Peak current is ~26mA, applied for around 600ms, applied every few minutes.

The large resistor values and high impedance feedback pins are highly sensitive to parasitic resistors and capacitors created by residual solder flux. For example, residual flux can leave parasitic resistors in the ~1Mohm range.  Such a resistor in parallel with your large resistors will change the regulation value.  I suspect that is the problem and have confirmed it on other boards. 

Although it sounds ridiculous, the quickest test is to exhale directly onto the board.  The humidity in the exhaled breath almost always causes the parasitic capacitors and resistors to react. 

Thank you for the inputs.

I've read before in other posts, that the large resistor values and residual flux can be a problem.
So i've reduced the values by a factor of roughly 10 (see values my first post), except the MPPT resistor string. The resistor deviders are now all in a Sub MOhm region. I've also reduced the MPPT setting resistor values.

I've tried to exhale directly onto the board - no observable effects.

As mentioned before, the problem seems to only occur when increasing the VBAT voltage above 5V.

Could it be a problem that a quite large cap is connected to VBAT (4.5mF), wheras there is only a small one (original configuration of eval board) connected to VSTOR?

What is the general recommendation where to connect a storage element like tantalum caps? VSTOR/VBAT? I connected to VBAT because in shipping mode the input current to BQ25570 is specified less than von VSTOR so that the cap more or less only experiences his own self discharge.

Thansk,

Reinhard

I modified an EVM with your resistor values and saw the flux problem worsen when breathing on the board.  I gave the board back to my technician to attempt to reclean but I am out of office until next week. 

VBAT is typically the pin for a large capacitor.  When VSTOR is greater than VBAT_UV (~2.0V) the PFET between VBAT and VSTOR will close and tie the two pins together.  If a load transient is making VSTOR=VBAT dip below 1.8V, then you will enter the less efficient cold start mode.  

Another possibility might be the 64ms sampling of the output voltage.  If your input source can provide close to the maximum PIN that the IC can handle, then the boost converter can overshoot until the next 64ms VRDIV sampling period.

Thank you Jeff,

I will see if the "flux problem" might be our problem - at first sight no, but lets see.
Are there any changes/problems noticable at your board if you set the VBAT voltage to the maximal specified value of 5.5V?
(As mentioned, our problems start when VBAT is set to levels above approx.5V)

Let me allow me another question:
Do you have design tips or recommendation for coin cells as energy source?
The usable capacity of these cells is negatively influenced by high current discharge pulses, so the idea was to use the MPPT of the BQ25570 in combination with tantalum caps as energy buffer to "smoothen" the current drawn from the battery.

After having my technician clean the EVM that I modified to match your design, I saw the same infrequent overshoot so flux is not the problem.  I also reduced the top VBAT_OV resistor so that VSTOR=5.0V and did not see the overshoot.  I was able to capture the attached waveform, which shows the IC momentarily going into cold start.  Can you repeat the attached waveform?  After discussing with the IC designer, our theory is that high amplitude switching noise is false tripping the cold start circuit.  I plan to add additional decoupling capacitors (~1nF) to see if that prevents the cold start false trip.

Regarding using low impedance coin cells, keep in mind the following:

1.  The peak input current will be ~200mA unless you use MPPT set very high so that the battery's internal impedance trips the input voltage regulation (i.e. MPPT) circuit.

2.  The MPPT sample period is for 256ms every 16s.  Is your load current going to collapse VSTOR during this time?