TLV3012: Instability on power on

Julien

we are off today on holiday break but I want to make sure that I understand the connections before analyzing.  I am unclear about the connection at the output of the comparator because the circuit got blurred a bit.  Is there a connection point at the junction of the 1M, 1.87M, comparator output, and gate of the p-ch MOSFET?

If so, I presume the 1M is to weak pullup to keep the MOSFET off when the comparator is powering up.  I know the TLV3012 output goes low momentarily for about 6us when it is starting up but other than that I am not seeing any problem with the circuit implementation. You mentioned something about delaying the start of U5 which is on the other side of the switch.  There is turn on delay with the TLV3012.  The delay is less than 1ms, so that may not be a problem.  One thing that I have seen help the TLV3012 start in a more controlled manner is if the power supply voltage ramps slowly (ramp rate of greater than 200us/V.

What is the range of the supply that is being monitored?  I am a little concerned since that is powering the TLV3012 which has an operating range up to 5.5V and an abs max value of 7V.  I presume that your system remains within that range at all times.

Please clarify the connection and I will analyze this further tomorrow.

Chuck

Hi,

Yes, the gate of P-Mos, 1M87, 1M and the comparator ouput are connected.

The delay of the TPS63000 (U5) is about 100ms.

The range of the power supply is 4 to 5.5V and could sometime increase up to 5.6V. I know this voltage is out of the standard range but I presume that could not destroy the component. The voltage range is guaranteed by the linear regulator LP38693 (U3).

The large hysteresis is need because the hydrid Supercapacitor voltage increase up to about 4.5V when the P-Mos is open and the power consumption of the circuit is release.

regards,

Julien

Julien

Thanks for verifying the supply range and connection points.  I do not see any issue with how you configured the TLV3012.  One thing that may help your circuit is to put a series resistor between the output of the comparator and the junction of the two resistors and the gate.  I am thinking somewhere in the 1k to 10k range just to slow down the response a bit.  I am wondering if the switch is getting turned on too fast and this is causing the supply voltage to dip below the 4V range and then turn off immediately.  If this were to occur, an oscillation could occur where the TLV3012 keeps trying to turn the FET on and off.  Putting a series resistor of say 10k will limit the sink/source capability of the TLV3012 to less than 1mA which prevent any damage from occurring.  Please give this a try and let me know if this helps.  If you can use a 4-ch oscilloscope and probe the output, supply voltage, and the two inputs, then we will have a full picture of what is happening.  Comparators are relatively simple devices and respond to what they see at their inputs.  I am guessing that this picture should show something that may help us analyze the problem.

Chuck

Chuck,

The oscillation could not be cause by a fast change of the voltage of the power supply because of the large capacity of the hybrid supercapacitor.

I agree with you, the use of a 10k resistor will limit the output current and slow down the maximum switching frequency.

We already observed the behavior in case of oscillation of the circuit. The capacitor voltage just slowly decreases from 5.2V to 4V, but the TLV3012 oscillates and the output changes fastly (over 10kHz) and the TLV3012 overconsumes. We're not able to verify if the current path is Vdd to Gnd (internal consumption of the TLV3012) or Vdd to Output (consumption of the gate of the P-MOS).

It's sure that the start of the TPS63000 (U5) have an influence on the oscillation of the TLV3012.

Julien

Julien

Unfortunately I am still a little confused by what you are describing.  It sounds like you are saying that when U5 receives power through the switch that is turned on by the TLV3012, the output of the TLV3012 starts to oscillate.  I am not sure how or why that would happen but my suggestion of putting a series resistor between the comparator output and the gate of the MOSFET may help.  To further stabilize that node, you could even add a capacitor to ground to make sure there the comparator output is not seeing some sort of transient from the turning on of U5.  You could also add a cap to ground at the middle of the resistor divider that the TLV3012 is monitoring.  If you can stabilize that node, maybe that could help.  This is definitely a hard problem to troubleshoot in this fashion.  Another possibility is the TLV3012 being a micro power device, has very limited bandwidth and thus a limited PSRR frequency range.  Please make sure maybe a series diode to the power pin with a bypass capacitor on the V+ pin of the TLV3012 would prevent the comparator from seeing any supply dip when U5 turns on.  Sorry for the rambling response but I am trying to give you any idea I have to help mediate the problem.

If we cannot solve this via E2E, we may need to take this offline.  Do you have a TI representative that calls on your account?

Chuck