Monitoring a few analog values and a few binary bits over the internet?

[Nigel Goodwin]

While I think about it, I mentioned earlier I was going to test a non-rechargeable battery EVE ER26500 once the 18650's ran out - but I'd got enough bits to build another unit (and a suitable SIM card), so I've build that up and got it running now, feeding a separate MYSQL table for simplicity. So we'll see if it makes 60 days!!.

Just an update, I gave the wrong battery number - the ER26500 is only 8.8aH, it's actually an ER34615 that's 19aH, and it's one of those we're sacrificing in this test. Ooh - just checked the price of them, £21.13+vat from Farnell

 

[Pommie]

I remember being shocked at the price of Lithium Thionyl Chloride batteries. Good capacity though, 70Wh per cell.

Mike.

 

[k7elp60]

Here is an idea that I helped with about 30 years ago. I had a friend with the same problem as MikeMI. The friend had a telephone line at the remote location. I was in the process of assembling telephone answering devices. He used that and would call the number at the remote location , and if he got a certain message on the answering machine he knew there was a problem.

 

[Nigel Goodwin]

I've also got a wager coming up - many of the envisioned uses for the project will use non-rechargeable LiIon, as existing GSM types do - and these commonly transmit once a week only, with an expected five year battery life. The batteries are 19Ah, and usually used with at least two in parallel, with a 'best' capacitor across them - the batteries have high internal resistance, so can't supply the high current pulses the modems require. So the next test will be to power it from one of these batteries (adding a best cap - I've got a space for it on the PCB) and see how long that lasts. My boss reckons that it will last more than 60 days, I reckon it won't make 60 days - so we've got a bet on it, with the loser buying lunch!

As the 18650's haven't run out yet, it's a bit difficult to predict - the 18650's are 2.2Ah each (so 8.8Ah), but are second hand batteries - so that makes it even more difficult. I'm guessing they won't make three weeks - if that proves the case, and they are truly 8.8Ah, then the 19Ah shouldn't make 60 days!.

Anyway, enough rambling on, hope this gives you some ideas.

Well the 18650's ran out, they lasted 18 days, 4 hours, and 12 minutes - while transmitting every five minutes - a total of 5166 transmissions.

By my calculations the EVE 19Ah should make around 38.86 days - which would easily win me the bet

The EVE version was started on 17/8/2018 at 11:08, and so far has sent 3287 transmissions - so 60 days would be October - annoyingly the battery voltage of these cells is incredibly flat, until they suddenly die - so the battery voltage readings aren't helping

 

[Nigel Goodwin]

OK, the EVE test board stopped on 6th September, it only managed 20 days, 6 hours and six minutes, and sent 5454 transmissions - a rather disappointing total I thought? - still it easily won me the bet

I've stuck a voltmeter on the battery, and it reads about 3.2V, but resetting the PIC and starting it running again produces dips below 3V on the battery, as the GSM modem draws current from it.

Considering I was taking no low current precautions, not even putting the PIC to Sleep, running at 64MHz, and transmitting every 5 minutes I don't suppose it's too bad. The intended application will only transmit probably once a week, then go back to sleep. Ignoring the standby current in sleep, 5454 transmissions is over 100 years at one a week

Even once a day (which is likely to be an option) is well over ten years (almost 15 years) - and the aim is to achieve 5 years plus battery life - which is why the EVE batteries are used, as they have a very long shelf life.

As far as consumption in Sleep goes, I'm no further forward at reducing it than I was before - but apparently under 100uA is the standard aim of other commercial units - which I'm easily exceeding.

Oh, as far as the bet went, I had a free Breakfast at Weatherspoons this morning.

 

[Pommie]

Oh, as far as the bet went, I had a free Breakfast at Weatherspoons this morning.

I'll bet it was the best breakfast ever.

Mike.

 

[Nigel Goodwin]

I'll bet it was the best breakfast ever.

Mike.

Weatherspoons is a lot cost pub chain, but they do decent food at a decent price - and their breakfasts are pretty good - and there's one not too far from where we work

 

[Nigel Goodwin]

Just been having a quick measure, it's taking about 13mA quiescent (just running a loop waiting), so that alone used 6.3Ah of battery life during the test - adding a SLEEP command dropped it to around 50uA, which is only 0.025Ah for the same period.

I couldn't use sleep for the test (which in any case would have took too long), as I've got no provision on that test board for a 32KHz xtal, and I'm actually using those pins as I/O. The Li-Ion test used the exact same settings, so it gave a valid comparison between the two.

My new test board, which is designed to fit an existing commercial enclosure, has a 32KHz xtal onboard, and uses a 16F18446 instead of an 18F.

While I've been typing this the sleeping 18F has actually drifted down to about 30uA.