Test of Step-Up-Modules (sparkfun, Pololu & china-module) / any other?

[ahhk]

Hi,

i tested 3 step-up-boards in the same configuration /sensor:

https://www.pololu.com/product/2114/specs
Sleep-current: 103uA

https://www.sparkfun.com/products/10967
Sleep-current: 28uA

http://www.ebay.de/itm/Elevador-Tension-DC-D-Step-Up-Boost-0-8-3-3-V-a-3-3-V-Arduino-3-3-V-ESP-8266-/221886656856?hash=item33a979b158
Sleep-current:243uA

Isn't this the china-module, which normally should be "quite good" (linked at mysensors...)?

Are there any other step up modules as good as the sparkfun NCP1402-Module? Or better and cheaper china-modules?

Greetings

[ericvdb]

I bought these, both the 5V and 3.3V versions

http://www.aliexpress.com/item/8pcs-Very-small-DC-DC-0-9-5V-to-DC-5V-Step-UP-Boost-Voltage-Converter/32268508188.html

http://www.aliexpress.com/store/product/8-pcs-Ultra-small-DC-DC-0-8-3-3V-to-DC-3-3V-Step-UP/1525466_32268496143.html

[ahhk]

Same china-modules. They are crap compared to the sparkfun module......

[ericvdb]

Crap?? The ones I got consume +/- 60uA while the arduino is sleeping, wouldn't say thats crap...

[m26872]

I just measured the sleep currents for a few of my nodes (china step-up module linked from there).

First one really old I just found (I've no clue what sketch or lib is in there) equipped with one DS18B20: 440uA

Second node with two DS18B20 (one probe style) and a rather fresh code which I think I'd be able to trace lib and IDE versions etc for: 59uA (Not bad!)

Third one of the "famous two" node 106 (dht22 and bmp180): 98uA

One note is that the price of these step-ups is high to start with and seems to have doubled in one year, so I doubt I'll buy more of these. And probably and hopefully that goes for any other step-up as well. Better to design for low voltage directly, when you can.

[ahhk]

440uA and 98uA is far too much! 59uA is quite good but not very good.
Sleep current is in my config the biggest part, which sucks most of the juice out of the batteries.

Maybe this is a quality problem of the china-modules?

I tested my 3 modules with the same node - i just replaced the step-up and measured the current. So my results are based on the same sketch, hardware and lib....

[scalz]

@ahhk:
but your 59uA is not the booster only. if you want low power you will need to change things in software, including removing watchdog first. and cut off all you can. sorry I have no time for a tut for the moment. I am finishing my rev 1.1 for my board and need to make documentations for boards I started.

and you are right, there are design things which can affect power consumption of different modules. like inductors, capacitors quality, efficiency... that is why I decided to design entirely my board, so I don't depend on different random quality of external modules and it is reproductible..
Not easy task I think to depend on external module, mainly for ultra low power. The more variables you add in the system, the more luck you have to break your low power..

[m26872]

@ahhk
"Quite good" is far from "crap". 59uA looks coherent with @ericvdb 's 60uA. I never claimed I did a comparative test, I just measured what I had laying around. I doubt the differences I measured depend on the step-ups.

[scalz]

another thing too.. How do you measure power consumption? do you have uCurrent?
I say this because you get 28uA from sparkfun module. And in datasheet it says it is 30 uA if Vout is 1.9V and 45uA for Vout 3.3v NCP1402 33T1 model, so it is very near ericvdb results..

[ahhk]

My Multimeter is a Peaktech 3410. Manual says: "0,1 µA; +/- 1,5 %".

BOD disabled, 8mhz internal, only HTU21D and NRF are connected...

I still dont understand, why my china-module takes 243uA with same configuration...
I am just replacing the step-up module on the breadboard.

[scalz]

oki.
but for more low power you need to not use the watchdog. and it is used in gw.sleep.
others tips too are to set unused pins in output mode, and set them = 0 if I remember right. because it consumes too. look at gammon power savings tuto, you will learn interesting things..

for your china module, I can't see on pictures what is the reference of the booster ic...then looking at datasheet would tell more.

[ahhk]

what can i save without the watchdog?
gw.sleep doesnt work without watchdog, i think? what can be a workaround?

i found a blog, where someone tested the powersavings by setting the pins to outputmode. This is <1uA. not a big deal...

[NeverDie]

Running Gammon's "Sketch J" drops you down to 150nA, and it doesn't set unused pins to anything.

[scalz]

yes of course setting pins mode is the last optimizations. Like Neverdie said, you should try Sketch J.
but you can do all what you want, if your power supply is not good, you will never get <uA..

[ahhk]

To reach <50uA is the target. I dont need to get nA.
The difference between 100uA and 50uA is the change intervall of the batteries. every year or nearly every 2 years
I will take a look at "Sketch J"....sound interesting...

[ericvdb]

Just took the time to shoot a pic of my Step-Up consumption, including a Voltage regulator MCP1702-3.3

As you can see, it's consuming 54uA with nothing connected.

Step-Up module: link

The capacitors are really important, one on the input of the step-up module, the other on the output of the MCP1702-3.3, both are 22uF. Without them, consumption is 2.20mA

[icebob]

@ericvdb great observation. But this board is not an MCP1702, because MCP1702 is a voltage regulator IC

[m26872]

Reading this thread again makes me remember another observation I did when was trying to get my mini PIR sensor up and running.

I was measuring load current and switched between the old "big" china step-up and the now more common smaller sized one. Identical load. The sleep current was a lot more for the newer smaller one. Could have been those 240uA, not sure but I remember it was the double or something. I was really confused since the two boards look to be populated with the same componets. I didn't look into this further cause I could bearly make the PIR stable on boost supply in any way.

Edit. Power Led disabled, of course.

[icebob]

On my custom node PCB, I will use MCP16251 chip to boost supply to 3.3V.
http://ww1.microchip.com/downloads/en/DeviceDoc/25173A.pdf

By docs, If I use it from one AA battery (1.5V), the efficiency is ~80%, consumes ~14uA. I think it is not bad.

[ericvdb]

@icebob said:

@ericvdb great observation. But this board is not an MCP1702, because MCP1702 is a voltage regulator IC

Where did you read that it's an MCP1702??

This test setup has a Step-Up module to 5V AND a voltage regulator to 3.3V. I never mentioned the MCP is included in the Step-Up module.

[icebob]

@ericvdb Sorry, I read somethingh wrong

[rvendrame]

@ericvdb , do you have any data sheet for this step-up? Or do you know which is the core chip?

[ericvdb]

No time to look for the datasheet atm, but here are some pics:

[scalz]

Hi.

maybe it could help some people to "evaluate theoretical" quiescent current...

if I remember right (I am not in front of my stuff), we can calculate the theoretical power consumption at Input like this:

I_input = (Vout * I_circuit)/(Vbat * Eff)

where
I_input = power consumption at Input
Vout = booster voltage output
I_circuit = power consumption of the circuit
Vbat = battery voltage
Eff = booster efficiency for Vout and Vbat

Now, if we take MCP1651, and says we have:

• a circuit which consumes 50uA (including booster quiescent current, sensors, leakage and a well designed circuit...)
• a MCP16251 booster with 3.3V output on a single 1.5V cell. Efficiency won't be the same during the whole life. And it is not an ultra high efficiency booster or it it would be named like "ultra high" (when they can, they do advertisement ). This is why we can only see the efficiency at 1mA. But what is efficiency under very light loads (<100uA)? It should need to be tested. no matter, we assume it is like on datasheet.
  So here we have Eff=85% at VIN=1.5V and Eff=80% at 0.9V (near end of life)

Some maths gives us:
I_input = (Vout * I_circuit)/(Vbat * Eff)
I_input = (3.3V * 50uA)/(1.5V * 0.85)
I_input= 129uA
So it should use approximatively 129uA at Input/batt.

Another maths, if batt is at the end of life:
I_input = (3.3V * 50uA)/(0.9V * 0.80)
I_input= 229uA

So here we can see that quiescent current of booster is not always the biggest problem.

I hope it can help in your choice. And I hope to have not done a mistake lol!

[icebob]

@scalz: thank you for you calculation. It's interesting. I will measure it if I receive the MCP16251.

[Adi Vacaru]

This guy has actually tested the mcp16251 and posted his findings on YouTube https://m.youtube.com/watch?v=O5o6JUjz6Yc