Startup problem TPS54332

The worst case should be to start up at 4.95 V under those conditions.  Can you post your start up waveform?  Be sure to include EN, SS Vout and the switching node.

Hello John,

Here is one start up waveform. Is there anything specific you want me to measure or give details about levels and such?

Best regards,

Sebastian

There are possibly more than one issue with your circuit.  That is a lot of output capacitance.  It will take a bout 4 A current to charge them up during the slow start time.  That may be dragging your battery voltage low.  The zero and pole frequencies of your compensation network are very high as well.  They may be completely ineffective.  What is the ESR of your output capacitors?  How did you derive the circuit component values?  From the EN waveform, I can deduce that your input voltage is fluctuating a lot during start up.

The ESR of the output capacitor is 80mohm. I used SwitcherPro to start off with for the circuit component values.

Schematic from switcherpro

Gain and phase simulation.

Here is a startup waveform of 7.5V input voltage, which works.

In spite of what it seems, switcherpro does not accurately support mixed output capacitors.  That feature was never fully implemented before development was discontinued.  Can you input supply handle the 4 A charging current for the output caps?

Datasheet for our batteries list 3C as high discharge , so 2550mA. We will do some testing and see what we can deliver.

Removing 2 of 3 680uF capacitors and adjusting the compensation network still results in poor startup.

We removed all cap, so we only had 22u on input and output and setup the schematic according to webbench solutions, still same startup problem.

Any suggestions?

What kind of load do you have a t start up?  If the load is very light you may run in to low headroom drop out.  See this app note.  It is applicable to TPS54231 as well.

http://www.ti.com/lit/an/slva547a/slva547a.pdf

Hello i am Sebastians colleague.

We have read the app note and tried different solutions. Tying a diode from in or out to boot is not enough. 

The dummy load has to be a low value to work. 1k is to high. This is the start up waveform with 410ohm load, and only two of the 680uF caps mounted

Is that not a very light load? Do you have any other suggestions?

//Claes Jonsson

I would start by removing all the 680 uF caps.  Re-calculate the compensation  using 3 x 22uF ceramic only.  switcherpro does not support using two types of output capacitors and will not correctly calculate the compensation.  I might be able to compensate your circuit.  Can you tell me the type of 680 uF capacitor?  I need to know the ESR.  It will probably be next week before I can do it though.

What happens if you increase the load to 4 ohm?

I have re-calculated the comp for 3x33uF and this is what it looks like

and with 4 ohm load the supply gives approx 775mA and the waveform looks like this.

The 680uF caps we use: http://www.farnell.com/datasheets/1790597.pdf

Can you post your current schematic?

As Sebastian wrote earlier, we have changed R6 = 270k and R7 to 101k. This should according to calculations give a startup voltage of 4.31V.

And C5 has max value of 27nF.

//Claes Jonsson

I'm not sure about your second to last waveform but the last one looks almost normal.  It looks like when you start to draw current, the VIN voltage is drawing down and your EN is disabling.  You may be able to defeat this by adding some additional bulk capacitance to the input and / or adding more hysteresis to your EN threshold.