| FORUM

FEDEVEL
Platform forum

What is Noise?

SockThief , 06-28-2022, 05:34 AM
Hi!

I have a post (https://designhelp.fedevel.com/forum...lter#post19868) which is talking about RC filters, however the genesis of this was this schematic from @phils-lab's Mixed Signal course



specifically, for now, R100 + C102 forming an RC filter for "noise" - and this has lead me to a rabbit hole on noise. I have discovered that there seems to be a lot of hand-waving when it comes to noise - its a blanket term that feels like it gets thrown around a lot "this component is noisy and so we add a filter" and that's it.

I am eager to dig into this in more detail, as there are, to my knowledge a lot of different types of noise - with reference to the above

- noise is anything that is not the signal we are interested in; so if we expect a 5V DC voltage, then anything that causes that voltage to not be 5V anymore is considered noise? i.e.: we might say, "the VBUS signal is noisy" if it's not a constant 5V; we don't say anything about what causes that noise
- the VBUS signal is coming from a USB port - we don't know for sure how nice the 5V from that port is, it might be nice it might not
- the USB cable might be long, it could be a good quality cable with some ferrite baluns or it might be cheap and nasty and we will get RF and EM radiation coupled into the cable
- L100, C100 and C101 for a Pi LC filter to clean this up? (what sources of noise are we concerned about here?)
- the 5V line after the Pi filter is then used to feed the LDO and the Buck converter
- if whatever is connected to the LDO draws current, we can expect that there might be higher than expected voltage drops on the 5V line and the voltage might droop

the above I am fairly confident in; but please correct me if i made a mistake, what follows I am less confident in

- in the lesson (lesson 5 @ 14:00) @phils-lab mentions the the "switching noise" will not only go to the output but will feed into the other power supplies - what is this switching noise? is this due to the output switching and causing voltage drops at the switching frequency? is this due to high speed transients causing EM/RF coupling between components and traces or a combination? what does the voltage at "+5v" label look like? is it 5v + some ripple due to noise, and it's the noise component we are filtering using R100/C102?

Ps. I am currently watching this: https://www.youtube.com/watch?v=49dfOQbmKvY - hopefully some insights; also please see this related post based on a series of videos from @robertferanec on noise (https://designhelp.fedevel.com/forum...clarifications) and also this post on Ferrite beads (https://designhelp.fedevel.com/forum...-ferrite-beads)

P.P.S looking at this circuit and this question in particular i dug deeper into RC filters and looking at current flows, which resulted in this post (
https://designhelp.fedevel.com/forum...ow-pass-filter) - leading to the question - is the reverse current that RC filters can generate in issue in this case?

thanks!
qdrives , 06-28-2022, 03:15 PM
Originally posted by SockThief
- noise is anything that is not the signal we are interested in; so if we expect a 5V DC voltage, then anything that causes that voltage to not be 5V anymore is considered noise?
I would say no. Noise is an AC element. So if the voltage is 5.5V, but constant, it is not noise. It may be incorrect (5.5V vs 5.0V) but stable.
the VBUS signal is coming from a USB port - we don't know for sure how nice the 5V from that port is, it might be nice it might not
The 5V coming from a PC (USB) is pretty good as far as I know. However, USB chargers, that is a completely different story.
with some ferrite baluns
Those won't help with differential noise, i.e. switching noise from the regulator.
L100, C100 and C101 for a Pi LC filter to clean this up? (what sources of noise are we concerned about here?)
Both incoming and outgoing. Do also consider incoming ESD. The chance you touch a USB cable/connector is very high.
if whatever is connected to the LDO draws current, we can expect that there might be higher than expected voltage drops on the 5V line and the voltage might droop
If the LDO draw more than 170mA, the 3.3V cannot be maintained (considering a perfect 0.000V drop LDO) with the 10 ohm resistor supplying 5.0V.
USB (1.1, 2.0) is specified for 100mA or 500mA ((non)powered hub).
Heidi Barns and Steve Sandler made an analysis of a power supply. At a specific frequency, the voltage swing was bad. With the inductor L100 and capacitor C101, if you switch at the resonance frequency, it will not filter, but make it worse!
what is this switching noise?
The noises are:
1) Input current (U101, pin 4). Due to the motor draining current.
2) Voltage drop due to the motor starting from stand-still. Turning motors cause back-EMF and then draining less current as there is less voltage to push it. This is why there are motor controllers.
3) High frequency noise from the motor brushes. This is largely radiated, whereas the others are only conducted. A local filter capacitor and common mode choke work best to suppress the noise.
is the reverse current that RC filters can generate in issue in this case?
Not that I know of.
SockThief , 06-30-2022, 07:11 AM
"I would say no. Noise is an AC element. So if the voltage is 5.5V, but constant, it is not noise. It may be incorrect (5.5V vs 5.0V) but stable."
- yes, you're right. a small distinction i overlooked

I just found this page which actually seems to address a lot of my questions: https://andybrown.me.uk/2015/07/24/usb-filtering/

qdrives , 07-01-2022, 02:49 PM
That is only 150mV peak-to-peak.
Just running a simple simulation on his filter solution.
BLM18PG221 is an 0603, 1A 220 ohm model. I pick the Wurth 0805 200 ohm 2A version that is in LTSpice.
100nF X7R
10uF 16V electrolytic
Result: No problem. It filters from about 1MHz.

However, that cap has 1.9ohm ESR and is an electrolytic. Now lets replace it with a ceramic cap.


If either of them would be switching at 62kHz (a quite common frequency for SMPS), then it would not filter but make it worse.
You test and measure with a electrolytic with 1.9 ohm ESR. You think you improve the design by switching to ceramic.....
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?