Moteino Regulator Choice & Drain

[ichilton]

Hi,

I notice in the following post, it says: "I decided to use the MCP1702 3.3v regulator instead of the MCP1700 I started with. There's no significant difference and allows the Moteino to be powered from a 9V battery (up to ~13V actually)":
http://lowpowerlab.com/blog/2012/12/28/moteino-eagle-files/

Is that still the case?

In the BOM xls file on Github, it lists both the MCP1700 and MCP1702 (by the way, the 1702 link goes to the 3v version rather than the 3.3v - assume that's a mistake?).

Have you had any problems with excess drain when having an MCP1702 on board but connecting say 2x AA's to the 3.3v line (regulator's output pin)?

The MCP1700 apparently doesn't do that:

http://harizanov.com/2013/03/funky-v2s-ldo-to-blame-for-high-sleep-current-now-fixed/
http://forum.jeelabs.net/node/278.html
http://jeelabs.org/2011/06/21/mcp1702-current-draw/#comments

Or do you always connect a higher voltage and go through the regulator?

Thanks,

Ian

[Felix]

Yes the BOM is not totally up to date. I will find some time to update that.
No real reason to use the mcp1700. I don't recommend using 2xAA behind the regulator, so I don't care about drain in that situation, it's already a bit overclocked at 3.3V.
I care much more about being able to use a 9V battery instead of 2xAAs, that's why I chose the 1702.
Use anything above 3.3V in front of the regulator (VIN).

[ichilton]

Apart from the regulator thing, the BOM seems to match the schematic.

Is it a 9V PP3 that you normally use with wireless nodes for yourself?

Do you find that going through a regulator reduces battery life?

Thanks,

Ian

[Spexx]

I was also thinking about using two AA batteries, which are connected directly to the 3.3V line. So I have measured for some days ago the current consumption and I was wondering, that I need more current if I don't need the 3.3V regulator. But according your links, this is probably a problem with the MCP1702...

If I connect 5V to Vin and if the Atmega is powered down and the RFM69 is sleeping, I can measure a current consumption of ~6,2µA. According the datasheet, the MCP1702 needs a quiescient current of approx. 2µA. So maybe you could reduce the consumption from 6 to 4µA, if you don't use the regulator. But as Felix mentioned, the µC is already overclocked, so maybe the frequency should be reduced from 16Mhz to 8Mhz. But I think this isn't worth it.

I'm thinking about using 3-4x AAs or AAAs, because they would have a higher capacity compared to the 9v batteries, but need also more space. Maybe it would be an option, if you need a very long running time before changing the batteries.


Felix,
by the way, the MCP1702 supports an input voltage up to 13,2V. So it would be also possibly, to supply the Moteinos with a higher voltage, for example with 12V? If I remember correct, only one capacitor in your BOM does not support more than 9-10V, so probably this one has to be replaced.

[ichilton]

I've not actually measured the current, but on a non-moteino atmega328 board, i'm currently at over a year on a single pair of AA's connected to the 3.3v line.
(this reduces down to about 4 months with an MCP1702 on board).

The 1uF capacitor for C2 linked in the BOM is a 10v so if that's what's on your board, you'll have to replace that with a higher tolerance one to be able to use above 10v.

Ian

[Spexx]

At which frecency do you use the board?
I have running a TinyTX with an ATtiny84 with 2xAA batteries. After some month I have at the moment a battery voltage with 2,75V and it works fine without problems. But it runs only with 8MHz.
The advantage in case of connecting the batteries direcly is, that the voltage can be measured directly via the µC.


I don't need really an input voltage from 12V, but I have replaced the capacitor on my selfmade Moteinos. So maybe Felix could also use another capacitor in case of updating the BOM.

[ichilton]

My 328 ones are running at 16Mhz with an external crystal.

I have 14 TinyTx running away on the 8Mhz internal OSC.

All of them sleep most of the time and just wake up once a minute to read the temperature (ds18b20), transmit it and go back to sleep.

My 328 without a 1702 on board looks similar in battery voltage over time to the TinyTx, which are at about 14 months on the same pair of AA's.

All of my other 328's had a regulator on board but not used and I noticed I was changing the batteries in them every 4 months, which is why I went investigating the other week!

Ian

[Spexx]

Hmm so with my understanding there would be two possibilities:
1) Moteino with regulator, Vin > 3.3V
2) Moteino without (or not used MCP1700) regulator and a pair of AA's (connected to 3.3V)

The first possibility needs probably a little bit more currenct because of the regulator, but you will always have a stable voltage at 3.3V. Depending on the connected sensors, this variant would have a benefit.

In case of extending the battery life, I would prefer the second possibility. But the Atmega will be run out of spec with 16MHz. Because of this, I would reduce the frequency to 8MHz, to be on the "safe operation area"...

Maybe someone can tell us about his experience with the battery life in case of using the regulator.

[ichilton]

Yeah, i've heard about the safe operating thing, but it seems to work fine anyway.

The voltage seems to drop below what the rfm12b can transmit at first.

I actually had an OpenEnergyMonitor emonTx get as far down as about 2.55v and it was still working ok and transmitting, but started giving some dodgy temperature readings from the ds18b20.

You do need to be careful running without a BOM set though as i've heard funny things can happen when the voltage gets really low and can start executing random instructions like overwriting itself. You'd normally see the low voltage and change batteries before that happens though.

My plan is to build up some spare units in different configurations - regulator, no regulator, batteries into regulator, batteries after regulator etc and running them on an increased sample rate and see what life I get out of them. i.e if you run at a 2 second sample rate, that simulates a month of use at a 60s sample rate in a day, so in a few weeks you can simulate a few years of use.

Ian

[ichilton]

I just pulled out some of my graphs -
Sorry, the whole timeline doesn't fit on the 1 image...

This is a 328 in my loft, 16mhz with no regulator - 2xAA straight onto 3.3v:



This is a TinyTx (ATTiny84A) in a bedroom - 2xAA straight onto 3.3v:



This is a 328, 16mhz, 2xAA straight onto 3.3v, but there is an MCP1702 on board:



Interesting huh!

Ian

[Felix]

Hey .. nice graphs and documentation of what is happening, thanks Ian.
My 2 cents is that a 9V has been proven to run a Moteino for about 10 months (with sleeping code), and is also more compact than 2xAA, and also keeps the atmega very close to spec. I don't see a real big reason for using just 2XAAs. I would see a good reason to use 3XAA/AAA where they could provide more juice and also keep close to spec.
Still trying to get to assemble some of those power shields ... they will yield 3.3V/5V from a single cell of any kind (not coin cells tho) for ultra compactness.

[ichilton]

Not too bad, but that's lower than i'm seeing with AA's...

My main reason for using AA's is cost - you can get an 8 or 10 pack of cheap AA's for £1 (GBP), whereas you only get 1x PP3 for that.

Now i'm up to nearly 30 nodes, that's quite a bit!

What are the power shields? - using a boost convertor? - those are on my list to try out too.

I'm going to have to do some tests with all of these different scenarios/options!

Thanks,

Ian

[Felix]

Good point, 9V batteries are a little more pricey.
Power shield is a little shield board for Moteino which gives you a bunch of options on how to power it, and includes a lipo charger and a boost regulator, a little prototype area.

[ichilton]

Ah, interesting!!

Ian

[smni]

Looking at the spec sheet for the MCP1702, it requires an input voltage of Vr +Vdropout. For 3.3V, the dropout voltage is about 600mV (for a current draw of 250mA, per the spec sheet). So that would suggest that you need about 3.9V into the regulator.

Is this correct, in practice? What happens if you only supply 3.3 or 3.4V to the input of the regulator instead of the recommended 3.9+? How does the dropout voltage change for a smaller current load?