Tracking down a noise on my prototype

[eTobey]

Hi there,

i have an AP66300 (A) that gives me a 5V from 14V. After that i got a regulator (R)(MCP1703) that puts out 3.3V, but these 3.3V have some kind of noise (pretty exact 400Hz) (is that terminology right?).
On that 3.3V i got a few components, that are on the pictures. Not to forget the uC, but which i disabled via reset.

I think i can rule out the can (C) unit, since i do not measure a corresponding signal over R727, just a regularly signal. The same applies to pin 1 to gnd on that unit.

For the Hallsensor AH1913 (H), i put a magnet over it, which did not change anything.

I got also an oscillator (O) on the 3.3 rail with 24Mhz, but i dont think it would make a noise that has 400Hz?

That EEPROM (E) is on the other side of the board, and i cant access it.

When i turn off the board (not disconnect plus or minus), the noise is gone.

The AP66300 and the regulator are 15mm apart, and the rest is on average 50mm away. In between there is nothing.

Probing around with a spring loop, did not show a source of this noise. Nor did a holding the tip on those.




Docs:

https://www.diodes.com/assets/Datasheets/AP66300.pdf
https://ww1.microchip.com/downloads/en/DeviceDoc/22049f.pdf (MCP1703)
https://ww1.microchip.com/downloads/aemDocuments/documents/MPD/ProductDocuments/DataSheets/24AA32A-24LC32A-32-Kbit-I2C-Serial-EEPROM-DS20001713.pdf
https://ww1.microchip.com/downloads/en/DeviceDoc/20001667G.pdf ( CAN MCP2551)
https://www.diodes.com/assets/Datasheets/AH1913.pdf

[dobsonr741]

Seems current spikes pulled by something on the 3.3V rail. I can think of 3 actions:
1. Live with it, few mV rms should not matter for the components you told about.
2. Find the source of the current spikes, by pulling off components from the board.
3. Try a footprint compatible better LDO.

[eTobey]

Found it! Suspected the right one: It was the hall sensor.

[coppercone2]

you can try to add more decoupling or a bit of series resistance with the hall sensor to see if it quiets down


I wonder what would happen if you connect the hall sensor to the regulator with a wide strip too? and does it have a ground plane?

[eTobey]

Yes i could try a series resitor. It cn go down to 1.6V.  Bigger cap will be the easiest way for now.

Why do you think a wide strip would help with this? It has a solid GND plane underneath.

That supply trace is not so good, as its jumps from layer 1 to layer 3.

[coppercone2]

mot sure, I think a series resistor has the most probability of working

[eTobey]

I just had a look at some of the other noise. I thought it came through the CAN bus, but no, it only is there when the MCU sends.

If there would be so much noise, then it would not stand out at all. See the arrow.