1
\$\begingroup\$

I have a circuit in which there is a mcu lpc1768 and a gsm module sim800l.

For power I use DC/DC converter and made a 4.2v using lm2576 for sim800l. Used that voltage, I made a 3.3v using ams1117 and powered mcu.

My problem is,sometimes not always,mcu keeps resetting for seconds. The time is when sim800l wants to find a network or send data.

I tried to capture signal on reset pin and 3.3 using osciloscope when reset happens.

Here is what I could capture

enter image description here

And here is one of the pulses with t/d on 2.5 ms

enter image description here

Here is some information may help

  1. 3.3 node doesn't have this noise
  2. 3.3 node is actually 3v becuase of drop out caused by linear ams regulator
  3. reset circuit has 100nf cap connected to ground and 10k ohm res connected to 3.3 node
  4. there is a 2200uf cap on 4.2 v beside sim800l
  5. sim800l doesn't reset and find its network

Does anyone have any idea what may the problem be? Let me know if you need more info about my circuit

More Info

Here is my board power scheme

enter image description here

enter image description here

and here is reset and 3.3 node together !

enter image description here

In fact I was wrong that those pulse are only on reset pin . They are on 3.3 node too. but they drop more on reset pins

More Info After More experiment

The problem is completely related to antenna! when I close it to my micro it starts to reset and when I close it to my battery wires, my protection circuit shut down the power!!it all started when we wanted to place the board into the case!

Do you have any suggestion how I can clean my circuit from this problem?

here is my reset layout as @Sunnyskyguy EE75 told that it may happen because reset loop is too large and itself is antenna

enter image description here

\$\endgroup\$
19
  • \$\begingroup\$ Please add the schematics as well \$\endgroup\$
    – Huisman
    May 5, 2019 at 18:40
  • 1
    \$\begingroup\$ You mention all sorts of capacitors but you mention no decoupling capacitors on anything. 2200uF caps are large but also have too much inductance and too slow to respond to the types of current surges the module pulls. \$\endgroup\$
    – DKNguyen
    May 5, 2019 at 18:42
  • 1
    \$\begingroup\$ This is clearly not a noise. Noise is not a slogan you can use each time you don't understand an interference. This does look like a short time load that causes the voltage to dip until the power supply slowly restores it. Do you have any LED, motor, whatever you gurn on and off? Show your scheme, we will look for such load. It feels like a bigger capacitor will not be enough. \$\endgroup\$
    – user76844
    May 5, 2019 at 18:45
  • 2
    \$\begingroup\$ Your issue is far more likely brownout than noise. GSM, especially when meant literally has very high pulse currents during transmit, and a lot of attempted power supplies cannot handle this. Typically the modules are designed for direct connection to a suitable battery. \$\endgroup\$ May 5, 2019 at 18:45
  • 1
    \$\begingroup\$ Then start pushing it methodically. Use the scientific method: declare a hypothesis (which component causes the problem), define an experiment (what should you do to either prove or disprove it), then go to your next step. \$\endgroup\$
    – user76844
    May 9, 2019 at 19:59

2 Answers 2

1
\$\begingroup\$

This regulator is rated for 3A output. You are showing a drop of 1V but unspecified load current. If GSM load current is 1A then 2200uF cap ESR could be 1 ohm ( V/I ) which is too high.

2200uF caps range in ESR from 18 mohm to 1 ohm depending on part.

The same effect can occur with high DCR in the choke. But if using 100uH, DCR ranges from 19 mohm to 2 ohm

What ESR and DCR values do you have in your design?
What are your design specs and how did you validate it?

\$\endgroup\$
11
  • 1
    \$\begingroup\$ Then go fetch ESR & DCR specs! \$\endgroup\$ May 5, 2019 at 20:17
  • 1
    \$\begingroup\$ Is there a DC OK signal tied to reset, then maybe the DC is not OK, if not what is the open drain or collector switch tied to reset! \$\endgroup\$ May 6, 2019 at 2:29
  • 1
    \$\begingroup\$ Then like ESR but you have an ESL problem with loop inductance of layout of wires. How big an area roughly is the power ground loop to reset? with 100nf cap connected to ground. Is any ground or power wire in the reset loop with cap > 1/10 wavelength. Then that's your problem. 10nH/1cm of wire inductance. Show your layout. if you have a camera. \$\endgroup\$ May 6, 2019 at 12:58
  • 1
    \$\begingroup\$ Ground plane is assumed when you have Tx nearby. So antenna effects increase with ground, power and signal wire length/width ratio creating inductance. But high ESR not using multiple caps causes supply sag during burst current of GSM sync to tower. SO you probably have classic EMI layout issues everywhere,. \$\endgroup\$ May 6, 2019 at 13:02
  • 1
    \$\begingroup\$ Then your Reset loop is also an antenna. Make it small \$\endgroup\$ May 9, 2019 at 19:06
1
\$\begingroup\$

The output capacitance of the AMS1117 is too less, causing the 3V3 to be unstable. The datasheet recommends 22μF:

The circuit design used in the AMS1117 series requires the use of an output capacitor as part of the device frequency compensation. The addition of 22μF solid tantalum on the output will ensure stability for all operating conditions.
When the adjustment terminal is bypassed with a capacitor to improve the ripple rejection, the requirement for an output capacitor increases. The value of 22μF tantalum covers all cases of bypassing the adjustment terminal. Without bypassing the adjustment terminal smaller capacitors can be used with equally good results.
To further improve stability and transient response of these devices larger values of output capacitor can be used.

Do also read: Capacitors for AMS1117 for 5V to 3.3V regulation

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.