Go Down

Topic: 3.7 Volt Battery for 3.3 Arduino ATmega 328 (Read 596 times) previous topic - next topic

wvmarle

Bad idea. That boost converter may be quite efficient (losing just 10-15%) but your linear regulator loses 34% no matter what. That's nearly half the energy of your battery being converted to heat, instead of doing something useful for you.

There exist boost/buck converters that can go both ways. That may be a solution. I was actually quite surprised to see the SIM7600 not being able to run directly off a 3.7V battery, many GSM adapters are designed to do just that.

Looking at the datasheet of the MIC29302 it looks like the drop-out is low enough to have it run directly from your battery. No need for the boost converter.

The thing is, most of your components will not draw much power normally, just spikes. Check the actual power consumption. You may be able to get it done with the MCP1700 and a big-ass capacitor (maybe even a small supercap) to support the spikes on the 3.3V side. I'm thinking of a capacitance in the order of 1-100 mF (that's m, not µ).
Quality of answers is related to the quality of questions. Good questions will get good answers. Useless answers are a sign of a poor question.

gpssignal

Do you have anything that needs 5V or can you run everything at 3.3V?
Is it possible to use 2 batteries, so you can step down from 7V?
Mmm...
ATMega2560 use 5V, SIM7600 use 3.8V, SDReader 3.3v, Sensor 5V... Before I was thinking in ATmega328 to run 3.3V but always the 5V sensor was a problem. Maybe I could use a external circuit to convert the 0.5-4.5V output sensor TO 0-3v3 volt, but It was more simple use 5V for main power supply.

gpssignal

#17
May 10, 2019, 08:07 pm Last Edit: May 10, 2019, 08:12 pm by gpssignal
Bad idea. That boost converter may be quite efficient (losing just 10-15%) but your linear regulator loses 34% no matter what. That's nearly half the energy of your battery being converted to heat, instead of doing something useful for you.

There exist boost/buck converters that can go both ways. That may be a solution. I was actually quite surprised to see the SIM7600 not being able to run directly off a 3.7V battery, many GSM adapters are designed to do just that.

Looking at the datasheet of the MIC29302 it looks like the drop-out is low enough to have it run directly from your battery. No need for the boost converter.

The thing is, most of your components will not draw much power normally, just spikes. Check the actual power consumption. You may be able to get it done with the MCP1700 and a big-ass capacitor (maybe even a small supercap) to support the spikes on the 3.3V side. I'm thinking of a capacitance in the order of 1-100 mF (that's m, not µ).
Yes I know (but I am not expert) that Linear Regulator are not good for a battery powered project. But the 5Volt sensor actually is a big problem, with output of 0.5 to 4.5 volts.

To avoid lose a lot of power I will control the modules power on/off using a load switch from MCU.
I will search for your IC and the SIM7600 need 3.8V recommended. I could connect directly to battery I will thing your idea a bit more.
The MCP1700 Rated 250mA Output Current only. SIM7600 need peak of 2A

wvmarle

Yes I know (but I am not expert) that Linear Regulator are not good for a battery powered project. But the 5Volt sensor actually is a big problem, with output of 0.5 to 4.5 volts.
Which exact sensor is that? No 3.3V version or replacement of it?

Quote
To avoid lose a lot of power I will control the modules power on/off using a load switch from MCU.
Use the Enable pin of modules when available.

Quote
the SIM7600 need 3.8V recommended
The SIM7600 datasheet that I found says it needs 3.0-3.6V. Apparently it's designed for 3.3V operation. The capacitor would have to deal with the current spikes.

SD cards can also be pretty power hungry.

ATMega2560 use 5V,
Why not a 328p? I don't see any reason for the 2560 based on your component list. The ATmega328p runs fine on 3.3V, 8 MHz. At that speed it runs at anything from 2.4V up to 5.5V.
Quality of answers is related to the quality of questions. Good questions will get good answers. Useless answers are a sign of a poor question.

gpssignal

Which exact sensor is that? No 3.3V version or replacement of it?

Use the Enable pin of modules when available.

The SIM7600 datasheet that I found says it needs 3.0-3.6V. Apparently it's designed for 3.3V operation. The capacitor would have to deal with the current spikes.

SD cards can also be pretty power hungry.

Why not a 328p? I don't see any reason for the 2560 based on your component list. The ATmega328p runs fine on 3.3V, 8 MHz. At that speed it runs at anything from 2.4V up to 5.5V.
2560 instead of 328p because 2560 has 256kB flash memory. I am not sure if 328p has enough memory also IF I use 3v3 volts, how I said before, The sensor 5V anyway will need 5V step up. So I will use the same 5V line for ATMega2560 and avoid rescale the 0.5-4.5 analog output of sensor

I made first version using 328p and SIM800C, now I need a LTE network and I think more memory flash simply. I I would like to use more effective/addecuate sensor but I only adapt my design to the hardware available, sorry for that

My first Idea was use 328p 3V3 @ 8MHZ directly of 3.7 li-on battery and add a load switch for each modules when the 328p is running in sleep_mode... Removing all leds and regulator possible to save power. This idea changed when I knew the sensor We have is a basic sensor 5V power and when someone told me I will need more memory flash to use libraries and code for our project (I'm not embedded programmer)

I would like to test before but I dont have arduino, neither modules :D . I only have a reference circuits of IC manufacturers and my laptop :/

I dont have  two chances to pay a PCBA services.


gpssignal

Which exact sensor is that? No 3.3V version or replacement of it?

https://www.iot-store.com.au/products/dfrobot-gravity-analog-water-pressure-sensor

DFROBOT ANALOG PRESSURE SENSOR

srnet

I dont have  two chances to pay a PCBA services.
What does that mean please ?
http://www.50dollarsat.info/
http://www.loratracker.uk/

srnet

I would like to test before but I dont have arduino, neither modules :D . I only have a reference circuits of IC manufacturers and my laptop :/

Are you suggesting you want to design a 'project' and put it on a PCB or whatever, but that you are unable or unwilling to actually test it ?
http://www.50dollarsat.info/
http://www.loratracker.uk/

gpssignal

Are you suggesting you want to design a 'project' and put it on a PCB or whatever, but that you are unable or unwilling to actually test it ?
(Sorry for my english, I'm a spanish speaker.. I am doing my best effort writing in english D:)

I mean... I'm designing a PCB prototype version 2. I made the first version and then I paid for complete service PCB & PCBA in China (currently is in manufacture). I said I dont have element to test like : protoboard, arduinos, resistors, "anything", so I have to design based on modules with reference circuits of datasheet or similar, because if all my routing are okay the circuit/PCB/Project should be okay.

So summarizing, I am simulating the circuit in my mind only :S
For this reason maybe my questions are stupid doubts for you

I dont have two opportunities to pay twice PCB assembly.
I can do it one time and I cant fail :S


wvmarle

Sucks to be you, then. Totally untested, obviously without much if any experience of building such projects, going for PCBA right away, and then having basically only one chance to get it right! I'm used to 2-3 iterations to get it right. I'm doing the assembly part myself, not that hard, especially not now my local makerspace has a reflow oven. Saves heaps in setup costs, easily a couple hundred USD for a simple PCB.

For circuit testing: buying the components costs a fraction of the PCBA cost. You should at the very least test the different parts, make sure it works, that you have the connections right, that the sensor works for your specific purpose.

To see if something fits in memory, just start programming. You'll quickly enough find if it does or not. A BLE module mcommunicating over the Serial interface takes very little extra memory.
Quality of answers is related to the quality of questions. Good questions will get good answers. Useless answers are a sign of a poor question.

gpssignal

hey I am not agree with you. I design my first PCB a back to back converter years ago using cpld for logic section without test the circuit before. I used reference circuits of datasheet. I didnt have problem, only a linear regulator was overheating. Also I made others basic PCB and never I found a funcional problem. Also in my previous Job in a microelectronic lab they never tested the circuit, only use a reference circuit to make PCB. The Main reason is In my country there is not budget/money for development and to find components is very very difficult. So the cost of assembly in China is a bit more of trying myself. But they have tools and Experience to solder tiny components and MCU. The circuit used in Arduino Module are not complex, the most circuit are reference circuit of data sheet and to integrate with arduino you need digital connections SPI/I2C/UART. No complex to integrate. Also PCB will not be a very very fast signal. I think a good design in four layer PCB is enough. IF I need to design Analog Electronic I think I need test before offcourse. For example now I am having doubts because I need modify to adapt to a battery powered project removing unnecessary parts, And I don't know how will works. I dont need to know programming to make a circuit integration and make a PCB. Finally like I dont fine a answer I will make all using main 5V power supply only, now I dont mind  to save battery :/

Go Up