Author Topic: AB1815 Power Management Abilities  (Read 6015 times)

ChemE

  • Sr. Member
  • ****
  • Posts: 419
  • Country: us
AB1815 Power Management Abilities
« on: September 10, 2016, 01:05:24 PM »
I've been through the datasheet twice and still can't work out how much current the AB1815 is capable of switching on/off safely without using an external MOSFET.  It looks like Vss switched mode in section 3.1 is the easiest to wire up and would use less power than would other modes.  Are there any major drawbacks to this versus Vcc switched mode?

TomWS

  • Hero Member
  • *****
  • Posts: 1930
Re: AB1815 Power Management Abilities
« Reply #1 on: September 10, 2016, 01:33:23 PM »
I've been through the datasheet twice and still can't work out how much current the AB1815 is capable of switching on/off safely without using an external MOSFET.  It looks like Vss switched mode in section 3.1 is the easiest to wire up and would use less power than would other modes.  Are there any major drawbacks to this versus Vcc switched mode?
This is from "AB18X5 Family System Power Management", Abracon Drawing # 453728 (emphasis on 1Ohm was added by me):

Quote
Figure 1 shows the recommended implementation, in which the internal power switch of the AB18X5 is used
to completely turn off the MCU and/or other system elements. In this case the PSW/nIRQ2 output is
configured to generate the SLEEP function, and the PWR2 bit is asserted. Under normal circumstances, the
PSW/nIRQ2 pin is pulled to VSS with ~1 ohm of resistance, so that the MCU receives full power. The MCU
initiates a SLP operation, and when the AB18X5 enters the SLEEP state the PSW/nIRQ2 pin is opened and
power is completely removed from the MCU. This results in significant additional power savings relative to
the other alternatives because even very low power MCUs require more current in their lowest power state
than the AB18X5.
The AB18X5 normally powers up selecting the OUTB register bit to drive the PSW/nIRQ2 pin, and the
default value of the OUTB bit is zero. This ensures that the power switch is enabled at power up. If the power
switch function is used, software should only change the PSW/nIRQ2 selection between OUTB (0b111) and
SLEEP (0b110) to ensure no glitches occur in the power switching function.

The only thing that bothers me about VSS switched modes is the potential of having an input pin biased to VSS (which would be leaky) and how cumbersome switching the ground plane would be, especially in an RF system.  An SOT23 FET is a small price to pay for eliminating these concerns.

I certainly would NOT switch an RFM69 radio ground with a 1 ohm switch.  It might work with RF coupling caps across the switch but I wouldn't bother.

Finally, if you need a Load Switch(VCC > 3.3V) then it's easy to use the N Channel FET in the Load Switch with its gate to 3.3V and the Source to the PSW pin.  You can't use VSS mode in this case (other than driving the N Channel FET).

Tom


ChemE

  • Sr. Member
  • ****
  • Posts: 419
  • Country: us
Re: AB1815 Power Management Abilities
« Reply #2 on: September 10, 2016, 02:43:08 PM »
Thank you Tom and I was afraid of that; it seems my lack of understanding in the datasheet and my inability to understand a good deal of your explanation means that this may be more than a non-EE can handle.  So an emphasis on a 1ohm resistance means I should use Ohm's law to work out how much current can be passed at a given voltage?  If so 3.3VDC/1ohm = 3.3amp which seems much higher than I would have expected this teeny chip to be able to switch on/off.  Regarding ground planes and RF, I'm 100% out of my depth there.  I hear you saying that Vss mode is ill advised if my circuit includes a radio which it will.  In looking at FET datasheets it isn't at all clear to me how much current they leak when they are off.  I trust that the FET that Felix used on the TPL5110 breakout board is uber low leakage but it seems most FETs leak more current than a sleeping mote so one has to be careful in their selection.

ChemE

  • Sr. Member
  • ****
  • Posts: 419
  • Country: us
Re: AB1815 Power Management Abilities
« Reply #3 on: September 10, 2016, 02:48:19 PM »
I suppose the main thrust of my questioning is whether the AB1815 + a low leakage FET will use less current than the TPL5110 breakout board.  14nA WDT current on the AB1815 + the leakage on the FET seems like it could well be lower than the WDT current of 35nA on the TPL5110.

TomWS

  • Hero Member
  • *****
  • Posts: 1930
Re: AB1815 Power Management Abilities
« Reply #4 on: September 10, 2016, 03:43:56 PM »
Thank you Tom and I was afraid of that; it seems my lack of understanding in the datasheet and my inability to understand a good deal of your explanation means that this may be more than a non-EE can handle.  So an emphasis on a 1ohm resistance means I should use Ohm's law to work out how much current can be passed at a given voltage?  If so 3.3VDC/1ohm = 3.3amp which seems much higher than I would have expected this teeny chip to be able to switch on/off.
You would have to be careful about power (IxV) first.  But, the way to think about 1ohm is that if you're using an RFM69 transmitting with a nominal current of 65mA, you would have 65mV of ground bounce on the VSS side of your entire design when the transmission takes place, and then a much lower drop when just the 328P is operating.  All of this voltage change would appear as signal shifts at the control pins between the AB1815 and the processor.  It might be ok, but, with high frequency spikes exceeding this voltage, it probably won't be.
Quote
  Regarding ground planes and RF, I'm 100% out of my depth there.  I hear you saying that Vss mode is ill advised if my circuit includes a radio which it will.  In looking at FET datasheets it isn't at all clear to me how much current they leak when they are off.  I trust that the FET that Felix used on the TPL5110 breakout board is uber low leakage but it seems most FETs leak more current than a sleeping mote so one has to be careful in their selection.
'Most' FETs are power FETs with large area which will leak a lot more than simple signal FETs used in the processors and such.  The leakage is generally correlated with voltage across the FET, negative bias on the gates, and temperature.  Complex relationship to be sure.

I suppose the main thrust of my questioning is whether the AB1815 + a low leakage FET will use less current than the TPL5110 breakout board.  14nA WDT current on the AB1815 + the leakage on the FET seems like it could well be lower than the WDT current of 35nA on the TPL5110.
I suspect that other than the 'academic' side of your concern, I think you will find that the power difference between an AB1815 and TPL5110 implementation is so far into the noise that neither one materially affect the battery life of your project.  It will be the 'awake' operating current and duty cycle, however small, that will overwhelmingly dominate the net power requirements.

The AB1815 is useful when you want to set specific times for wake up, especially if these vary.  TPL5110 is small, easy, and very accurately repetitive, but only occurs at ONE fixed rate - that is determined by the programming resistor.  This will get you pretty close to a specific interval, but the exact time will be determined by when the device was last powered on. 

By the way, the AB1815 is a very precise timer, but you have to calibrate it to compensate for any error in its oscillator frequency.  Abracon purposefully shifts the frequency to slightly higher than 32.768KHz so that you can program 'dither' calibration coefficients to get the average time to exactly 32.768KHz +/- 2PPM.

Tom