USB Port to Port noise variability on the same machine

Other Hardware Sensors
1

The amplitude of USB noise varies significantly from port to port on my machine, seemingly independent of USB version (2.0 vs 3.0).

Any idea what is causing this?
And what measures can be taken to reduce USB noise?

celem's post may suggest this behavior results from powering and serial port communication over the USB cable.
crocket's post may suggest this behavior results from other USB devices residing on the same USB bus.

I have tested multiple USB cables and compared across USB versions; no 60 hz line noise is showing in the frequency spectrum.

Thanks in advance!

2

boogie:
The amplitude of USB noise varies significantly from port to port on my machine, seemingly independent of USB version (2.0 vs 3.0).

Any idea what is causing this?
And what measures can be taken to reduce USB noise?

celem's post may suggest this behavior results from powering and serial port communication over the USB cable.
crocket's post may suggest this behavior results from other USB devices residing on the same USB bus.

I have tested multiple USB cables and compared across USB versions; no 60 hz line noise is showing in the frequency spectrum.

Thanks in advance!

Give us a little hint of what you are writing about!!! What machine? What do you consider "noise"? Why is it a problem?

Paul

3

Paul_KD7HB:
Give us a little hint of what you are writing about!!! What machine? What do you consider "noise"? Why is it a problem?

Paul

Sorry Paul :sweat_smile: I have a Pro Micro hooked up to a custom PCB, which filters and amplifies the incoming sensor (pulse oximeter) data.

The attached image has plots of a recording from two USB 2.0 ports on my machine (Windows 10 Dell desktop 16 GB i7) -- in the interest of simplification I removed the pulse ox sensor from the PCB for these recordings.

To my surprise, one signal (boxed in red) has much more "resting noise" than the other, which I am attributing to the port because everything else in the recording setup is seemingly the same. I have noticed each USB port on my machine has a unique "resting signal". Variability has also been seen across PCs (Windows 8 and 10).

Unfortunately this resting noise is hiding the actual pulse ox signal on some ports, which is obviously not ideal. Any advice on what could be causing this or how to reduce the noise would be greatly appreciated.

port_comparison.png
port_comparison.png1822×604 46.9 KB

4

boogie:
Sorry Paul :sweat_smile: I have a Pro Micro hooked up to a custom PCB, which filters and amplifies the incoming sensor (pulse oximeter) data.

The attached image has plots of a recording from two USB 2.0 ports on my machine (Windows 10 Dell desktop 16 GB i7) -- in the interest of simplification I removed the pulse ox sensor from the PCB for these recordings.

To my surprise, one signal (boxed in red) has much more "resting noise" than the other, which I am attributing to the port because everything else in the recording setup is seemingly the same. I have noticed each USB port on my machine has a unique "resting signal". Variability has also been seen across PCs (Windows 8 and 10).

Unfortunately this resting noise is hiding the actual pulse ox signal on some ports, which is obviously not ideal. Any advice on what could be causing this or how to reduce the noise would be greatly appreciated.

Oh, well then that is easy. The problem is you have a DELL!. Just kidding. Dell has a very old reputation of having junk and perhaps you found one. We had DELL problems years ago and never bought another.

Suggestion: Try a HP machine. Or other manufacturer.

IF the USB ports all work properly with printers, mice, and other normal USB devices, then there is not really a problem with the computer.

A PC is really full of noise on all signals. The design has to allow for that and mostly that is done by using clocking signals to tell a chip when to look for a matching signal and to ignore all others.

Each of the USB circuits is a separate circuit and control. They could be scattered to different available space on the mother board. And they can pick up all kinds of noise form adjacent devices.

More specifically, are you just using the USB for the 5 volt power or are you communicating with programs on the PC? Why is noise on the 4 USB wires causing noise on you board? And how is it getting to your circuit?

Paul

5

The pictures are missing units. Are these µVs? I doubt they are mV as then the red USB port would be differing in voltage of almost 1V within less than a second. I definitely don't have a computer with such a lousy USB voltage. And BTW, how did you measure this?

6

Paul_KD7HB:
Oh, well then that is easy. The problem is you have a DELL!. Just kidding. Dell has a very old reputation of having junk and perhaps you found one. We had DELL problems years ago and never bought another.

Suggestion: Try a HP machine. Or other manufacturer.

IF the USB ports all work properly with printers, mice, and other normal USB devices, then there is not really a problem with the computer.

A PC is really full of noise on all signals. The design has to allow for that and mostly that is done by using clocking signals to tell a chip when to look for a matching signal and to ignore all others.

Each of the USB circuits is a separate circuit and control. They could be scattered to different available space on the mother board. And they can pick up all kinds of noise form adjacent devices.

More specifically, are you just using the USB for the 5 volt power or are you communicating with programs on the PC? Why is noise on the 4 USB wires causing noise on you board? And how is it getting to your circuit?

Paul

Haha good point Paul, here are results from the machines I have tested thus far:

Asus Laptop Windows 8 #1 - low noise on all USB ports
Asus Laptop Windows 8 #2- low noise on all USB ports
Asus Laptop Windows 8 #3 - low noise on all USB ports
'Azza' Desktop Windows 10 - variable USB port noise
custom Desktop Windows 10 - significant noise on all USB ports
Dell Desktop Windows 10 - variable USB port noise

The USB powers the board, which communicates print statements & current signal amplitude. Removing the print statements and all other communication besides the signal amp did not reduce the noise in testing thus far.

The PCB amplifies the sensor data coming in but does so in a fixed manner, which shouldn't differ per USB port. My PCB + Arduino setup obviously has a resting or "idle" signal amplitude range, but I am not sure why it changes per USB port or where the noise is originating from. celem's post may suggest this behavior results from powering and serial port communication over the USB cable. crocket's post may suggest this behavior results from other USB devices residing on the same USB bus.

On my to do:

  • test a cable with a ferrite choke when it arrives
  • test some code that sends a 2 Hz sine wave into the USB port instead of sensor data
  • test on more machines

pylon:
The pictures are missing units. Are these µVs? I doubt they are mV as then the red USB port would be differing in voltage of almost 1V within less than a second. I definitely don't have a computer with such a lousy USB voltage. And BTW, how did you measure this?

Pylon,
Each recording has a time series and frequency spectrum plot of the incoming sensor (pulse oximeter) data over ~ 1 minute. Pulse oximetry is unitless, and the frequency spectrum is power spectral density (PSD) by Hz.

Thanks guys, really appreciate the help!

7

Very interesting!

You wrote: On my to do:

  • test a cable with a ferrite choke when it arrives
  • test some code that sends a 2 Hz sine wave into the USB port instead of sensor data
  • test on more machines

Are you going to add a ferrite choke or is there one on the cable already? You my need to get a sample of various ferrite mixes in order to find one that works with you problem PC.

Paul

8

Paul_KD7HB:
Very interesting!

You wrote: On my to do:

  • test a cable with a ferrite choke when it arrives
  • test some code that sends a 2 Hz sine wave into the USB port instead of sensor data
  • test on more machines

Are you going to add a ferrite choke or is there one on the cable already? You my need to get a sample of various ferrite mixes in order to find one that works with you problem PC.

Paul

The cables I have tested do not have a ferrite choke, I am going to test out this cable when it arrives. Do you have any cables/mixes you recommend?

As celem mentioned, the mostly likely impact the USB port is having on my circuit is probably the power supply affecting the gain.. perhaps there is some variability in USB port powering across machines when serial communication is involved.

9

When you talk about USB noise, do you measure the 5V power supply level or do you measure the D+/D- signals?

10

pylon:
When you talk about USB noise, do you measure the 5V power supply level or do you measure the D+/D- signals?

pylon,

Thus far I have only analyzed the signal communicated from the Arduino+PCB setup to my PC.
The signal is a buffer of sensor values read from the A1 pin, which is sent over periodically.

The "resting signal" amplitude read by the PC varies by USB port significantly on some machines, while other machines have a more consistent low "resting signal" amplitude on all ports.

My current suspicion is that USB power supply from a computer is inherently noisy according to what I have read-- the mostly likely impact the USB port is having on my circuit is probably the power supply affecting the gain. Testing thus far suggests certain brands (like Asus), which may have better components show a more consistent low "resting signal" amplitude on all ports.

11

boogie:
The amplitude of USB noise varies significantly from port to port on my machine, seemingly independent of USB version (2.0 vs 3.0).

Any idea what is causing this?
And what measures can be taken to reduce USB noise?

Yes, USB 5V power is for digital devices, its noisy, sometimes very noisy, basically if its within 4.5V to 5.5V
range at all times its good enough for digital devices.

For analog use you have to filter the supply noise out yourself, as with any poor quality supply rail (such
as automotive 12V). As a rule of thumb I'd expect USB to carry maybe 300mV of broadband noise from
audio through to 50MHz or more, since it comes straight off a computer motherbaord. I'd also expect the
voltage to fluctuate on a large range of timescales as parts of the host computer power up and down.

The simplest way to deal with this is use a 3.3V LDO linear regulator and use a 3.3V circuit (although this
is very cramped for analog circuitry).

A powered USB hub may be much cleaner than power from a computer, but I'd expect digital hash to be present as the hub itself has digital circuitry on it.

12

MarkT:
Yes, USB 5V power is for digital devices, its noisy, sometimes very noisy, basically if its within 4.5V to 5.5V range at all times its good enough for digital devices.

Thanks for the insight Mark. Do you have any thoughts on what hardware,etc is responsible for this noise? The following options come to mind:

  • machine manufacturer
  • OS
  • port version (2.0, 3.0)
  • power supply plugged or unplugged
  • desktop vs laptop
  • CPU
  • USB cable
  • port manufacturer
  • front panel/rear panel
  • motherboard type

Paul mentioned that certain brands with junk components may be causing this issue -- in fact, testing thus far suggests certain brands (like Asus), which may have better components, show a more consistent low "resting signal" amplitude on all ports. I am hoping to narrow it down past the machine brand if possible (i.e. why are the Asus machines showing better results). Test results thus far:

  • Asus Laptop Windows 8 #1 - low noise on all USB ports
  • Asus Laptop Windows 8 #2 - low noise on all USB ports
  • Asus Laptop Windows 8 #3 - low noise on all USB ports
  • Asus Desktop Windows 10 Pro - low noise on all USB ports
  • 'Azza' Desktop Windows 10 - variable USB port noise
  • Custom Desktop Windows 10 - significant noise on all USB ports
  • Dell Desktop Windows 10 - variable USB port noise
13

Wound not the best solution be to filter the noise on your project end?

Paul

14

Paul_KD7HB:
Wound not the best solution be to filter the noise on your project end?

Paul

Totally, but I have quite a few of these already made and would like to determine the machine requirements for this version :confused:

15

boogie:
Totally, but I have quite a few of these already made and would like to determine the machine requirements for this version :confused:

boogie:
Totally, but I have quite a few of these already made and would like to determine the machine requirements for this version :confused:

Understand!

That is what one of my customers discovered when they redesigned their light aircraft accessory to run off the 5 volt phone charging port on newer aircraft.

Way too much noise on the port to allow their device to operate. Had to redesign to have room for filtering.

Paul

16

Paul_KD7HB:
Understand!

That is what one of my customers discovered when they redesigned their light aircraft accessory to run off the 5 volt phone charging port on newer aircraft.

Way too much noise on the port to allow their device to operate. Had to redesign to have room for filtering.

Paul

Paul,

At least I am not the only one :smiley: A redesign is eminent, no doubt. My question at this point is why are the ASUS machines showing better results than the others? Answering this will hopefully allow minimum machine requirements to be set for this version.

17

boogie:
Paul,

At least I am not the only one :smiley: A redesign is eminent, no doubt. My question at this point is why are the ASUS machines showing better results than the others? Answering this will hopefully allow minimum machine requirements to be set for this version.

You keep asking the same question that ONLY you can answer by taking the boxes a[art and tracing the noise with your oscilloscope.

Paul

18

There are 2 sources of 5V in most pcs.
One is a small capacity 5V supply in the main power supply case that runs all the time and provides the necessary voltages to allow the PC to start and to provide power to the Ethernet card if the PC supports wake up on Lan.
Sometimes this power supply also provides power to the USB ports, but not always.
The other 5V supply is only present when the PC is switched on and has a large capacity to power all the 5V devices on the motherboard and anything else that needs 5V.
Both of these power supplies are switch mode and are inherantly noisy, but the standby one is usually less noisy than the main one .
If you need a clean source of 5 V , then you can only get it from a linear regulated supply.