https://www.ti.com/microcontrollers/msp430-ultra-low-power-mcus/overview.html#usb

Dirk wrote:
> So I was thinking of ATtiny85 or Arduino Nano kind
> of setup with some extra storage connected.

Aduino Nano is not a good idea because of it's USB-interface. In this 
case, the A* ProMini is a good choice, sleep with timer around 5µA: 
http://www.home-automation-community.com/arduino-low-power-how-to-run-atmega328p-for-a-year-on-coin-cell-battery/

If you use the version with 8 MHz, it can run directly from one 
LiIon-Cell.

However, you need to think about powering of your sensors, can work 
between 4.2 and 3.0 Volt? How to powering down while the AT328 is in 
sleep mode?

How you can handle powering down your sd-card and initialize again after 
restart? The usual China modules have an xx1117-3V3 with too mouch 
quiescent current, replace by MCP1702 / 1703 but with different pinout.

Hi Dirk,

do you really want to fire and forget?
I could imagine it is quite unpleasant if you leave the hardware at the 
lake and the firmare/hardware does weird things - for whatever reasons.

And of course do you want to collect data in shorter periods than one 
year? What if a sensor runs away or gets broken?

I'd suggest to use some rf magic like Lora (if monitoring is a thing), 
also you should have a not too bad time reference - you could of course 
use the ambient light which is accurate for longer periods but why so 
complicated ;)

A external RTC is quite accurate and not to expensive.

For the MCU selection - if you want to stick to Atmel/Microchip go check 
the tiny816 or the amtega 0 series (eg. Atmega3208) those are quite 
decent IMO.

If you want to have low power the EFM32 claim to be very low power.

If you're concerned about power consumption and don't mind larger 
footprint, use a solar cell both as light sensor and energy source (or a 
solar cell as energy source and an off the shelf ALS if you don't want 
to bother calibration)

Just some thoughts :)

you somehow lost me on bightness and clarity already ...
sounds like one and the same;
or you want to measure the tint and opacity?

Anyhow, that sounds like a more sophisticated optical sensing than a 
good ol' photoresistor.
More like a full fledged albeit low res camera module.
I'm only asking because that will (well - might) be the main cause for a 
drained battery;
and as that it's almost of no importance if you throw an ATTiny10 or an 
ATMega16 at it.
In terms of reliability I guess any modern day µC that has a wide enough 
temperature range for your lake in question will do just nicely.
So the only thing I can think of is low power consumption.

So in order to answer the question in a reliable way, it'd be important 
if you could let us know what three sensors you have in mind.
(make and model and if possible linked datasheets)
To properly guess the required amount of battery juice for a single 
reading and then expraopolate from there.

Too bad btw you said lake not river...
I saw a hobbyist project once that used a tethered submarine-style 
controller with a tiny turbine..
the flow charged an array of caps which then powered the gear.
that was nice.. (IIRC it was only temperature and time to interpolate 
flowrate or such.. can't remember exactly.. hence I cannot find it atm)

Another possibility might be to equip the gear with a small solar cell, 
salvaged from an ikea garden light or such..
(I think there are also swimming pool lights that work alike)
a small amount of rechargeable double-A batteries (mostly just one, 
sometimes up to three) power those lights,
and usually should hold enough of a charge to power your gear for quite 
a while.. if it's enough to cover all of a long winter IDK,
depends on the requirements again.


'sid

Dirk B. wrote:
> 1xOpacity:

That part alone needs 11mA during Operation.  Did you calculated the 
total power requirements of all sensors during the your campaign? 
Methink processor power is neglectable.

Btw. 10 meter depht means 1 bar pressure (1kg on each cm²). On a first 
view i do not see how you can tighten this sensor against it (i made 
measurements much more deeper and offshore).

Putting sensors in natural waters invites all the tiny little creatures 
in fouling and covering your sensors. Are you aware of this? Then you 
have leakage out from moisture. It is hard to tighten a case over longer 
periods without special hardware. Putting ebay buying parts together in 
a case means  a hard way to prevent this.

So nine sensors, not threee all of a sudden then..
and lacking the info on what sensors will be involved but one
(an ebay link is not a datasheet..)

If that's your type of research, I'm afraid I don't bother too much...
The footwork at least should be your part not ours!


good luck with the project.

'sid

Opacity : forget about that. Whatever surface will usually be covered 
with algae sooner or later.

Check this blog out. This person has a lot of experience in low-cost 
underwater data loggers utilizing Arduinos:

https://thecavepearlproject.org/

See also here:

https://thecavepearlproject.org/2020/04/08/diy-data-logger-housing-from-pvc-parts-2020-update/

Here are a few more useful links:

http://www.gammon.com.au/forum/bbshowpost.php?bbtopic_id=123
https://northernwidget.com/
https://esajournals.onlinelibrary.wiley.com/doi/abs/10.1890/0012-9623-95.2.68

okay for time_based logging;
@Dirk you didn't understand..
the exact sensor is important to get it's demand and respnse time 
properly.
"some photoresistor" is just not going to cut that.

Anyhow, you'll need to cover that battery reqquirement on your own once 
you get the infos yourself then.

Timely based logging:
No problem, just don't :D
Honestly, since you log in consecutive order (write storage in 
consecutive bytes) and you know the length of each dataset;
you can calculate the time of each entry with your reader later on and 
save a significant amount of storage
(say each set has 20 bytes) so the 400-420 block is the 20th reading and
with one reading every 4hrs it came in 80 hours after it's initial 
start.
(simple math easily to be done even in a simple spreadsheet excel like 
reading application)

In order to get the most accurate times, a RTC would come in handy..
most require a significant amount of juice unfortunately or get quite 
expensive easily.

So what I've seen to work well is a cheap LCD Quartz watch with an alarm 
function acting as a trigger.
it itself is powered by it's own tiny button cell (as usual)
with an alarm set every four hours.
the buzzer is replaced with a signal line to trigger the powerboard.
And the "alarm off" button is connected to a datapin on the arduino.

When the alarm goes off the powerboard then sends the 5V (or what have 
you) to the arduino,
which in turn holds the powerboard active and reports the "I'm awake 
stop beeping" buttonpress to the watch (to stop the alarm).
Once the arduino has done it's work and logged all data properly
it cuts it's own powerline to the powerboard which in turn then falls 
back to sleep automatically and neither the arduino nor it's attached 
sensors get any power.
So all you need is a low standby current powerboard, which should be 
fairly easy be done in <7µA or such..
IDK if there are any prefabbed chinese ones out there (I guess so) but 
plans on how to make one are widely spread over the internet.
(the infamous amazon order buttons used a very similar power-approach 
for example powering a wifi reporter with a single triple A over several 
thousand button presses)
And best of all the arduino's watchdog doesn't need to wake up every 
couple of seconds to count up to the four hours, but it just hangs there 
waiting for the alarm to tickle it'S power switch.
So all idle power is part of the cheap wristwatch movement.
even the lousiest LCD watches can last well over a year from those tiny 
button cells..
and if you worry about that,
you can just replace the buttoncell with a triple A to have it last for 
decades if you want.


'sid

An RTC perhaps a DS1302 or so takes far less current than anything self 
built. in the order of 400nA or less. There are RTCs which can provide a 
timer interrupt eg for wakeung up a controller.

Dirk B. wrote:
> We will be cleaning
> the sensor from algae if necessary as well.

And reading the data during inspection is not an option?

Dirk B. wrote:
> almost weekly but do not want to remove and readout
> all the time.

Why? you can replace an external battery and SD card (and hold a backup 
in the probe).

Dirk B. wrote:

> 2. to not log the date/time information. That reduces the storage down
> to ~34kByte for 1h logging and ~8kB for 4-hourly logging: Then it could
> already match into the Flash, but readout might be less comfortable than
> with SD cards. I wanted to mark them and can other people also to
> transport them. But with first approach it doesn´t matter too much
> anymore.

This sounds to me that you plan to use the program flash of an ATMega328 
as data storage. Do you have some experience in doing that? It's 
possible but not a simple task.

If you do not want to use SD cards, why not just taking a separate SPI 
EEPROM to store the data. If you get a PDIP package and put it in a 
socket, you can just take it out and read the data externally.

Dirk B. wrote:
> Well I do not want to open the box sealing regularly. The sealing will
> suffer too much and finally the electronics gets watered. :-(
>
> The sensors I will put into Epoxy resin to isolate from internal
> electronics. The outside is the one which can be easily cleaned.
>
> Dirk
>


Maybe a 2 part design with watertight connections makes sense. One with 
sensors and one with battery and memory. Than you can use enough power 
and gets rid of 1 year storage demand (and risks like data loss by 
device loss).

Doing this and long term storage and sensor operation demands are an 
option and nice to have.


Me think there are conflicting design goals. A very low power very 
special device made out of standard components. Designing that way runs 
often in unforeseen conditions that are hard to handle, lacks 
flexibility and ends in extensive component search.


> p.s. any idea on the clock timer?

Sorry no, i will made the whole project with an XLP PIC. Simply because 
i can. Last time i had sleep current below 100nA. Due to periodic wake 
up the clock timer is always included (or complimentary).

Hi,
I am in implementation phase. I have got a problem not yet solved with 
the alarm output from a DS3231 ZS-042 module: The alarm does not switch 
the SQW port. Now I am reading adverse information on removing 
resistances (there is a 4xCascade pullups or not.
PIN2 of Arduino mini can be switched an is set to HIGH as default.
Output of DS3231 stays at LOW. According to data sheet it says that 
output requires an external pullup resistor which I would say is the 
cascade. But stays on LOW. IT evens pulls down the Interrupt pin2 of the 
Arduino board. What´s wrong? THe clock generally works and gives out 
TIME, ...

I have used several implementation sketches.
So first question what to do with these resistance cascade?

Thks