Testing power consumption

It sounds like your Mayfly is not actually going to sleep.  An idle Mayfly will use around 6.5mA, and a sleeping Mayfly will use between 0.25mA and 0.5mA, depending on whether you’ve got a microSD card in the slot and the type of card.  Here’s the power consumption results from Mayfly current tests:

Mayfly current when board is powered and idle: 6.5 mA
Mayfly current when MCU put to sleep and no microSD card in socket: 0.27 mA (or 270 uA)
Mayfly current when asleep with microSD card in the socket: 0.43 mA (or 430 uA)

Gosh @shicks that is great to hear.  I wonder would you share how you made the sleep measurements with the microSD present. Any sketches for putting the mega1284 to sleep. Getting to 0.5mA would be fantastic!

Matt Interesting real data with a coulomb counter. Great stuff.

My current measurements on 0.28.01 for sleep are 2.5mA . I haven’t got a test setup for other modes at this point, but probably will have in the next couple of weeks.

For my current measurements I run it only from the battery and put a Adafruit INA219 in series, with a simple driver polling once a second, running on a separate Mayfly and then visually monitor current.

I’ve put together a spreadsheet for how  power is used. I tend to calculate in mAseconds and then of course 3600mAseconds = 1mAhr.

Since I’m also wanting to tweak production code, I need debugging. I modified the FTDI cable so that only the processor Tx and the Gnd are connected, and debug output will advertise what state the processor is in. For an FTDI connector –  male to male pins and just snip all pins except critical ones.

So from the spread sheet, for sampling period 900sec, using Readings 6.5mA@10secs, LTE 35mA@50secs,

Then for 24 hrs, 96*15minutes

sleep at 6mAs the calculation is 182mAHrs  close to yours

sleep at 2.5mA then   104mAhrs

sleep at 0.5mA  59.6mAhrs   ~ (I want it!)

If the “Switched 3v3” or “Switched 5v” is turned off, then nothing connected to those pins will be drawing power.  But if your Mayfly board is not actually going into deep sleep, then it’s sitting in idle mode instead of actually sleeping, and that’s why it’s drawing 6ma.  Wasn’t there some recent threads about how the I2C device was keeping the board from sleeping properly?

What’s the reason for having a separate ADC board?  Are you just worried about reading a voltage greater than 3.3v with your Mayfly onboard ADC?  Because if that’s the only reason, then it would be easier to just put a resistor divider on the sensor output and connect the center of it to your Mayfly ‘sADS1115.  If you pick 2 perfectly matched resistors (like 10k) then you simply double the ADS reading to give you the actual sensor voltage.

@mbarney an observation, the power is being turned off on the off board ADS1115, but not for the Turner. The ADS1115 Analog inputs are only specified for Vdd+0.3V. If either of the Turners outputs are above this, then it will feed through to the ADS1115 Vdd  Sw5V.  Probably won’t account for what your are seeing.

I found a I needed to know when in sleep v current consumption, so used the Ttty to know when its in sleep mode, and  to measure the current at that time. (I have Adafruits INA219)

@mbarney the trace looks good for the waking portion. What I see on  my system is that it wakes every minute and then sleeps inbetween.  So if you have it on a 15minute setting it will do above, then sleep/wake 14times. The current can be read when it goes to sleep.

For the differential case I would think you would have two sets of dividers.  BTW total error is additive. So you can probably match the R, but whatever the error is eg 0.1% then the final error that gets introduced is 0.4% . It may not matter if you are looking for dynamic changes rather than absolute numbers.?

@mbarney  its a common problem with low power, sleep circuits.

Just to be clear/restate what the issue is. When in “sleep” mode and powered off, all instrument pins (and actually all other processor pins) need to be set to output low, or input with preferably pull down.

On power-up, the pins are activated output and set high within 50mS so as to not trigger the wipe.

On power-down, its a race condition which to do first, and the global control of the power, but with a min 50mS pulse I would think you are safe to make the wiper pin low and hope that the power is turned off within 50mS.

KellerParent.cpp has ::powerUp() and ::powerDown() which is where I change the pins.

You are using ::setup() for setting the port direction, which is I think is run only once, so I would think you move the wiper setup to ::powerUp

Sounds like its coming along, and I’m fascinated that you are getting 0.5mA.

BTW with my FTDI cable setup, I just found the hardway that the Ftdi Rx pin (tx pin from processor) hadn’t been clipped off and on my stability test station; the 4.4Ah battery mysteriously suddenly drained down in about 3days and then recharged from he solar. It turns out my monitoring PC Win10 did some form of automatic update and then powered down. The FTDI cable  also powered down, including this unclipped Rx pin, that then drained off the battery current.  I only noticed after I had restarted the PC, and the monitoring, and then two days later was checking the results and saw the battery had mysteriously drained down and come back. So the purpose of stability monitoring is check for anomalous behaviour in a secure setting, so I needed to find out what had happened.

I would do as Neil suggested and make sure the trigger pin for the Turner is set low before sleep.

We’ve had the same problem with RS485 adapters that pull power through the Tx/Rx lines.  Because the stop bit is high those lines will stay high if you don’t force them low at the end.  It’s kind-of a PITA with the Tx/Rx bits; you have to actually write it setting those low into your loop because ModularSensors doesn’t know anything about the Tx and Rx pins, only the serial stream.

It could also be that the wiper is causing a whole lot of noise on the line which is keeping the Mayfly awake (acting as a pin-change interrupt).  But I think if that were the case you’d see the going to sleep/waking up lines continuously, not just once a minute.  The once a minute thing is because the alarm on the DS3231 doesn’t allow for every x minutes, just at one exact time or every second, every minute, every hour, etc.  To get it to wake only in 5 minutes (or 15 or whatever your logging interval is) you’d get the current time and then calculate and reset the alarm for 5 minutes in the future each time.  Since the board should only wake for a few ms every minute, it seemed safer to use the every minute alarm than to reset it to a new time each time.

I’d put it in the power up, since it’s supposed to happen right away and setting the pin is nearly instant and won’t slow down other things powering up.