| Electronics > Beginners |
| what causes this noise |
| (1/1) |
| hsn93:
hello, I'm having Voltage regulator (5V to 3V3) [TI LMS8117A] recommended 10uF tantalum cap on input and output (with low ESR). i didnt follow that because i thought i will compensate with the number of capacitors: [CL_tps1] (i find it close enough?) so i only placed 100nF near the regulator [C23] although, CL_tps1 is only 0603 package 10uF capacitor (i dont know about it as its turnkey by PCBA). anyway, you can see on the output I've [C24] which is again very low for what TI recommends. but that's because I've nearby MCU [SAM4E16E] that have too many decoupling caps *VDDCORE is just internal 1.2V of mcu regulator so it doesn't count.. so I've: 7x 100nF [MCU_VDDIO] 1x 10uF [MCU_VDDIN] 1x 100nF [C24] which is more than 10uF recommended by TI (not tantalum and supplied by turnkey) in my design i forgot to connect MCU_VDDIN to 3v3 and i just made jumper for it. now when i probe the 3v3 power i get noisy output: you can see this oscillating 3v3 rail is scary as the micro-controller has absolute maximum of 4 volts. any i already have reset from mcu every 40-60 seconds also (discovered this because of usb cdc). (thats where i thought the mcu has brown out detect but i can see the voltage is exceeding the 3.3 not going below) (in your opinion) - why do you think there is this oscillation in 3v3 rail? and what causes mcu to reset you think? |
| JohnPen:
Are you using the normal scope probe earth lead or not? It is very important to use a minimum length to avoid extraneous pick up. Use a spring coil on the probe's concentric earth to connect directly to an adjacent earth point at the regulator to confirm that your scope signal is a genuine problem. If it is no longer visible the problem of resetting the MCU is caused by something else. If still present the Texas data sheet does recommend placing the 10ufd directly on the 3.3v Vout for the the regulator to be stable. They also mention that the value can be increased up to 22ufd so increasing the value should not matter. I would suggest replacing C24 with a 10ufd Tantalum or adding one in parallel with the existing capacitor if more convenient. There have been a number of comments on the forum about these types of regulator being unstable if not decoupled properly. Hope this helps. |
| David Hess:
From page 10 of the datasheet: The ESR of the output capacitor must be greater than 0.5Ω and less than 5 Ω. The relatively large output capacitor swamps the effects of the various small decoupling capacitors which should *not* be located directly at the output of the regulator. The output capacitor should not be a film or ceramic or polymer part because the ESR will be too low. The regulator requires a minimum load current which may be as high as 5 milliamps. Is this requirement being met? If the load current is low, like maybe while the CPU is in reset, then the output voltage will rise if the minimum load current requirement is not met. |
| Ice-Tea:
3V3 -> 4V sounds a lot like a diode drop to me... Maybe you have an input signal being fed to your uC @ 5V? Which is clamped by an ESD diode which feeds into the 3V3 rail? |
| hsn93:
--- Quote from: Ice-Tea on January 08, 2019, 11:32:16 am ---3V3 -> 4V sounds a lot like a diode drop to me... Maybe you have an input signal being fed to your uC @ 5V? Which is clamped by an ESD diode which feeds into the 3V3 rail? --- End quote --- that is correct there was can bus and rs485 drivers that are powered from 5v rail .. this noise was 100hz which is me toggling a pin in that frequency (i dont know how this pin is shorting the Rx from Rs485 driver to ground?) --- now the 3v3 is stable but i still have the issue of (reseting) which i guess its in code (although its blink code) .. i will have play with different codes. thank you. |
| Navigation |
| Message Index |