Circuit module power up drawing too much current

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
I've just finished a small circuit that uses a PIC16LF1823 to switch on and control a WiFi module using a high-side NX2301 pFet. The circuit is powered by an XC6215B332MR-G, which is a‎ linear 3.3V regulator that has a nominal 200 mA current capacity. The entire circuit, when powered up, consumes around 130 mA. That's well within the regulator's capabilities, and in fact, it works beautifully.

1611103108376.png

I'm quite content with the entire setup, because it's battery powered (4xAAA) and it draws only about 14 µA (without C1) from the pack when the WiFi module is switched off. The WiFi module is intended to be used only once a month, and that for only less than a minute. So the battery pack should be able to last quite a few years before it needs to be replaced.

The problem is that, if I weren't to place a 470 µF electrolytic capacitor (C1) as shown above, the MCU would reset itself due to the sudden power draw demanded by the WiFi module whenever I switch M1 on. With C1 everything works flawlessly... and without it all hell breaks loose...

Here's the thing: I HATE THAT CAPACITOR ...

Why my intensely negative animosity towards said innocuous and innocent component?... a fair question... There are mainly two reasons:
  • It's big, bulky and ugly.
  • It draws an unacceptable amount of waisted current due to its inherent internal resistance.

Although the latter reason is the most significant, the former is a PITA just as well.

It's as simple as this, the leakage current of said capacitor subtracts several years of blissful operation from my precious battery pack.

I have tried applying PWM to the pFet's gate so as to gradually ramp up the power to the circuit to no avail. Even if I were to switch the MCU_pin low (remember, it's a pFet, so MCU_pin is high when the transistor is off) for one microsecond every 0.00017 seconds (that's a 0.2% duty cycle) the regulator would be overloaded and the MCU would reset itself.


Here's my question:
In the event that keeping the odious capacitor becomes unavoidable, is there an easy technique that would allow to gradually charge the capacitor, and then switch the entire circuit on, but without using another of the MCU's I/O pins? ... and then afterward to disconnect both the WiFi module and the cap so as to go back to the ultra-low power draw operating mode?
 
Last edited:

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
On the other hand, in your schematic C1 would be permanently active too... but with a leakage much lower than the original 470 µF since it would be a smaller cap.

Question: Why not connect C1 to V instead of to ground? That way C1 wouldn't leak when the MCU pin is high, which is its normal idle state.
 

Danko

Joined Nov 22, 2017
1,829
Yes, it may be small ceramic cup (almost zero leakage).
C1 connected to V will turn M1 ON when powered (not good), but it will only one first time in some years.
You easily can try it.
Edit:
Thanks, @crutschow !
So, capacitor must be connected to V!
1611103520231.png
 
Last edited:

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
Yes, it may be small ceramic cup (almost zero leakage).
C1 connected to V will turn M1 ON when powered (not good), but it will only one first time in some years.
You easily can try it.
That would be an acceptable glitch... thanks for confirming my assumption.
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
Danko lined out his second line, so I believe he realized that there is no power-on glitch due to the capacitor.
The capacitor in uncharged to start, thus the Vgs will stay at 0V when the power is applied, keeping the MOSFET off.
Interesting... it's a neat scenario that I think I'll sim in LTSpice and see how it plays.

Also, I'm installing an inverse parallel schottky diode with the resistor, so as to make the discharging charging of the gate faster when the pFet is turned off.

1611112252551.png
 
Last edited:

Ian0

Joined Aug 7, 2020
9,667
Wouldn't it be better to give the WiFi module its own regulator and switch that on and off to avoid disturbing your MCU supply at all? LP2951 is a LDO regulator with a shutdown pin (there are plenty of others). Then you don't need the MOSFET, the big capacitor or any of the bits that generate the slow rise time. (Just wondering if 3.3V is enough to make sure that the MOSFET is on - there aren't too many logic level P-Channel FETs out there)
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
Wouldn't it be better to give the WiFi module its own regulator and switch that on and off to avoid disturbing your MCU supply at all? LP2951 is a LDO regulator with a shutdown pin (there are plenty of others). Then you don't need the MOSFET, the big capacitor or any of the bits that generate the slow rise time. (Just wondering if 3.3V is enough to make sure that the MOSFET is on - there aren't too many logic level P-Channel FETs out there)
I too thought about using a separate regulator as you've suggested. But my board's pinout makes that idea more difficult to implement. Also, it's easier (and cheaper) just to add the resistor and small cap to the pFet's gate.

The pFet I'm using is designed to work at such low voltages, and it's working fine. Check its datasheet.
 
Top