DC/DC Boost Converter Startup Problem with Very Slowly Rising Input Voltage

[bendras]

I am trying to use a DC/DC boost converter to increase voltage coming from a super capacitor from 2.5V to 3.3V (see the schematic below). The problem is that since the power source can provide only a few mA of current the supper capacitors takes multiple seconds to charge up  and in turn the slow voltage rise makes the boost converter oscillate and then output around 1.2V instead of the expected  3.3V (see the graph below). Anybody has any ideas on how could this be solved?

Note: The DC/DC converter enable threshold voltage is  0.8V

Schematic:


Graph:

[mikerj]

You will likely need to add some kind of supply supervisor to hold the switcher in it's disabled state until the caps have charged to sufficient voltage.  You can get reset controllers in three pin packages (TO92, SOT23 etc) with various voltage thresholds which may be suitable.

[ogden]

R1 and R2 supposedly shall be KOhms, not Ohms. Also try using capacitor values which charges within 1sec or so. Now your simulation is kinda useless, at least for others - because it covers just input 0.8V to 0.9V range and that's it. I want to see whole 0V to 2.5V

[edit] Also C6 shall be some real electrolytic capacitor, with some internal ESR.

You will likely need to add some kind of supply supervisor to hold the switcher in it's disabled state until the caps have charged to sufficient voltage.

There is built-in supervisor in form of EN pin. It triggers at 0.9V 0.8V. So what you do is - add voltage divider. If you add 200K/200K resistors - it supposedly will start at 1.6V

[mikerj]

There is built-in supervisor in form of EN pin. It triggers at 0.9V 0.8V. So what you do is - add voltage divider. If you add 200K/200K resistors - it supposedly will start at 1.6V

No hysteresis though, so still likely to get instability at switch on threshold.

[ogden]

There is built-in supervisor in form of EN pin. It triggers at 0.9V 0.8V. So what you do is - add voltage divider. If you add 200K/200K resistors - it supposedly will start at 1.6V

No hysteresis though, so still likely to get instability at switch on threshold.

Which exactly datasheet did you read, if any? I see 0.25/0.8 V hysteresis:

[mikerj]

That's the exact same data sheet I read.  I see maximum Vil and minimum Vih values, exactly as you would see for any standard logic input, but no mention of of a Schmitt trigger input.  The block diagram doesn't show any kind of Schmitt trigger input either.  Did you find something else in the datasheet that mentions hysteresis?

[ogden]

I see maximum Vil and minimum Vih values, exactly as you would see for any standard logic input, but no mention of of a Schmitt trigger input.  The block diagram doesn't show any kind of Schmitt trigger input either.

You are right. EN input may not be Schmitt trigger - there is no clear mention of such. My prior knowledge let me believe that this IC is no different to some I know (LT switchers).

[SiliconWizard]

Indeed. But anyway, as ogden said, the main issue is with the values of R1 and R2. Way too low, so they load the boost regulator way too much. I simulated the op's circuit with 172k and 107k instead and it appears to start correctly and get stable in approximately 1.5s with some minor ripple. So the feedback resistors were the main issue.

Still, the output current it can deliver will be pretty limited (as the low-value feedback resistors showed).

[bendras]

I have done the following changes to the model:
  1. As suggested I have set the R1 and R2 resistors to appropriate values
  2. Added realistic input and output capacitor models (Taiyo Yuden JMK107ABJ106MA-T 10uF 6.3V  ceramic caps.)
  3. Added RT, RB and RHYS resistors to shift the turn-on voltage to 2V while keeping the turnoff voltage at 0.8V.

Now the startup seems to be reasonable.

New schematic:


Zoomed out graph:


Zoomed in graph: