Platform forum

Voltage regulation for analog data

Mahsheed , 09-05-2019, 05:37 AM
I am using single Li ion battery to provide power to GSM Module, micro-controller and accelerometer. When GSM module transmits data, voltage for micro-controller drops and it gives wrong accelerometer values ( analog pins). Usually, Lithium ion Battery gives 4 Volts and I am unable to use linear regulator for fixed 3.3V output ( difference between output and input voltage is very less). One option that I have is to use Zener Diode. Any other option or suggestion would be highly appreciated. Thanks in advance.
robertferanec , 09-08-2019, 09:40 AM
I would not be using linear regulator on battery powered device - linear regulators are wasting a lot of power. Maybe, try to find an accelerometer what works well on wide range input (I randomly googled for "accelerometer chip low voltage" and it looks like they exists: https://www.st.com/en/mems-and-sensors/lis3dh.html). Of course, you may need to then figure out how to

However, I would not be surprised if the problem is not only "power drop" but the whole kind of noise generated on power rail by GSM. Try to filter power going to GSM module (maybe use ferrite bead + big and small capacitors) and do the same on the circuits with microcontroller and accelerometer (ferrite beads, small decoupling capacitors). Be sure, voltages and currents going to GSM module are not going to influence microcontroller and accelerometer (e.g be sure you do not route your board like Input POWER pins - GSM MODEM - MCU - Accelerometer ..... you may want separate the circuits, do for example like GSM - Filter - Input POWER pins - Filter/Decoupling + MCU & Filter/Decoupling + Accelerometer).
Mahsheed , 09-08-2019, 11:24 AM
Yes, GSM is creating noise in power rail. That is the real cause behind the wrong calculation of analog values. I actually checked the accelerometer values by plugging the analog data to arduino powered by PC. These values were perfectly fine. Only microcontroller on board that is powered by same source, miscalculates the values.

I can not use ferrite bead due to size limitation of my product, but I am using capacitors on VCC of microcontroller as well as GSM. Unfortunately that does not help.

Originally posted by robertferanec
Be sure, voltages and currents going to GSM module are not going to influence microcontroller and accelerometer (e.g be sure you do not route your board like Input POWER pins - GSM MODEM - MCU - Accelerometer ..... you may want separate the circuits, do for example like GSM - Filter - Input POWER pins - Filter/Decoupling + MCU & Filter/Decoupling + Accelerometer).
Can you please explain this type of circuitry? I mean any reference link or video to this procedure. I never seperated power, so I do not have knowledge about this thing. Thank you for your fruitfil response.
robertferanec , 09-09-2019, 11:02 AM
Very simple explanation: What I meant is, that you maybe do not want to have your circuit connected in series like this Battery - GSM module - MCU - AM (Accelerometer). This means, that the power going from Battery to AM will be on its way affected by GSM and MCU. Instead what you may want to do, you may want to route GSM directly to Battery (nothing on the way), and maybe you may want to also route MCU directly to Battery and AM directly to Battery. So currents going between Battery and GSM are not flowing through the same tracks which are used to power MCU and AM. And on these tracks between Battery - GSM, Battery - MCU, Battery - AM you may want to have filters.

GSM modules are often very noisy - they need to be well filtered.

Layout Option 1 - BAD:

Battery ---- Track ---- GSM ---- Track ---- MCU ---- Track ---- AM

Layout Option 2 - Better:

GSM --- Filters --- Track --- Battery ---- Track --- Filters --- MCU
-----------Track --- Filters --- AM
Mahsheed , 09-09-2019, 01:59 PM
OK, So do I need to use 2nd order low pass filter consisting of Inductor and capacitor? Because Capacitors used on VCC pins does act like first order filters.
Also I assume that voltage drop due to GSM is actually considered noise. So, If voltage drops then being everything in parallel ( tracks may be separate), it will effect the voltage of other ICs. Only filters will do their work.
robertferanec , 09-10-2019, 12:40 PM
Well designed filters will stop spreading the noise (and voltage drop) from GSM to the other circuits.

For GSM, usually Battery - Ferrite Bead + Big capacitor + Small capacitors + GSM is good an ok filter, but you can also use something like Battery + Middle capacitor + inductor + big capacitor + small capacitor + GSM .. that is a better filer.

I do not know exact values, but often it is like small cap (100nF), middle cap (10uF), big cap (100uf or more), inductor e.g 2u2. But the exact values for specific board may need to be calculated and filters tested or simulated.
Mahsheed , 09-11-2019, 12:51 AM
Do I need to add these for every IC ( GSM, Microcontroller, accelerometer). I mean would it be enough to have it for GSM (ferrite bead or 2nd order filer).
Lastly what other things can be included in noise due to GSM?
Thanks in advance for the detailed response
robertferanec , 09-11-2019, 11:12 AM
If you have problems with noise, I would add a good filter on GSM power, simple filter on CPU, simple filter on AM

good filter: cap - inductor - caps
simple filter: bead - caps

PS: What you may want to do, add all the footprints into PCB, so you can play with exact filters and in final board you only fit the components you need.
Mahsheed , 09-11-2019, 10:08 PM
Perfect. Thank you very much
Mahsheed , 09-17-2019, 01:57 AM
Respected Sir,
I tried using filters on bread board. Initially, I used 0.1uF and 470uF Capacitors on battery pins and then coil inductor and 0.1uF, 470uF Capacitors on GSM VCC as well as Microcontroller VCC seperately. I left Accelerometer with out filter because I am using full module which has linear regulator in it. With this combination, I could not succeed getting improved results. Then I added 100 uF capacitors with other capacitors as well. The results improved enough but there is still some error. I think it is due to the use of bread board and high impedance paths. I just wanted to have your opinion on that.
Also I can make single layer pcb with the resources I have, but that would not have ground plane. Would you suggest me that for testing of Filters. (I get my final PCB made from China, so it wont have that issue. But I can not take a risk on my final PCB.)
Your help would be much appreciated.
robertferanec , 09-17-2019, 02:39 AM
Maybe try to make a cheap 2 layer PCB for testing (or 1 layer in house). When using breadboard, the results may be even worse - placing components close to each other and keeping leads as short as possible is important (this is not possible when using breadboard).

PS: Large capacitors are slow and it may take to long to react. If needed, it may be better to use multiple smaller caps (e.g. 3x 100uF may be better than 1x330uF).
Mahsheed , 09-17-2019, 03:19 AM
Ok. Thank you very much. I will check the results of 1 layer in house pcb.
ypkdani , 11-11-2019, 02:27 PM
Hello Mahsheed,

i have already design a device with a GSM module and you need to put some attention to:
  • the voltage required by the module, for example for the SIM800 you need a voltage from 3.4 - 4.4V (so higher than the standard 3.3V)
  • as said by Robert you need to put some big capacitors near the GSM module (>330uF) and near the SIM, some times if the module don't have enough power the sim voltage drop down
  • put a ferrite on the power net for the module
  • use large nets for power supply and gnd, when transmitt the module can absort up to 1.5A. If your battery is not good itself can not be able to give all the power required.
An application note https://www.electrodragon.com/w/imag...sign_V1.08.pdf
Ashish1710 , 11-13-2019, 10:59 PM
@Mahsheed. The power supply range of SIM800 is from 3.4V to 4.4V. Recommended voltage is 4.0V. The transmitting burst will cause voltage drop and the power supply must be able to provide sufficient current up to 2A. For the VBAT input, a bypass capacitor (low ESR) such as a 100µF is strongly recommended. For the VBAT input, a 100uF Tantalum capacitor (CA low ESR) and a 1uF~10uF Ceramics capacitor CB are strongly recommended .The 33pF and 10pF capacitors can effectively eliminate the high frequency interference. A 5.1V/500mW Zener diode is strongly recommended, the diode can prevent chip from damaging by the voltage surge.

In your design please you can use , Lithium ion Battery more than 4V. for better result


Use our interactive Discord forum to reply or ask new questions.
Discord invite
Discord forum link (after invitation)

Didn't find what you were looking for?