High noise spikes when connecting Teensy 3.2

Status
Not open for further replies.
K

Kaitain

Member
Hi there

I'm using a Teensy 3.2 to make potentiometer readings for a MIDI controller. The idea is that the controller is USB powered.
I know USB power is very noisy, that's why I've tried everything in my hand to filter that noise:
- In this FTDI application note recommends to use a filter just before the usb 5v pin.
- This filtered voltage goes to a LDO to get 3.3v, and from there to the potentiometers and also to the AREF pin.
- The cursor of the potentiometers have a capacitor to AGND. The pots are also AGN grounded.
- Also I'm using some software filtering to smooth the readings.

After the LDO the noise is reduced a little bit, but it looks like the poor LDO can't handle the spikes. Trying to figure out where this noise is coming from, I was that without the Teensy connected to the circuit the USB power, although noisy, is "quiet" compared to when the Teensy is plugged in (see attachments). First file with Teensy unplugged, second with Teensy plugged. The yellow measurement is taken directly from the USB 5v pin (1), and the pink one is taken after the passive filter (2).

So, is that normal? Are this spikes coming from the Teensy itself, or maybe they come from the computer, when a load is connected? I tried with 3 different Teensy 3.2 boards, and results are the same, so maybe is the second option...
Anyway, any ideas on how to improve this?

Thank you!
 

Attachments

  • NewFile3.png
    NewFile3.png
    48.1 KB · Views: 89
  • NewFile3 4.png
    NewFile3 4.png
    59.1 KB · Views: 70
  • 20190115_120831.jpg
    20190115_120831.jpg
    94.9 KB · Views: 67
Instead of potentiometers, use Alps rotary encoders. These are cheaper and less bulky than potentiometers. Especially for Midi stuff, where almost everything is quantified in 0 to 127 steps, using analog potentiometers and 12bit ADCs is probably overkill.
 
Theremingenieur said:
Instead of potentiometers, use Alps rotary encoders. These are cheaper and less bulky than potentiometers. Especially for Midi stuff, where almost everything is quantified in 0 to 127 steps, using analog potentiometers and 12bit ADCs is probably overkill.
Click to expand...

Thank you Theremingenieur. Yes, using encoders is an option, but before changin all the design I'd like to try with potentiometers (I like much more the analog style:).

The main issue here are the spikes that shows up when turning on the controller. They go through the filter and the LDO almost intact, and I don't know why...
Maybe is there any filter configuration I haven't tested yet?
Cheers!
 
All sorts of measurement artifacts can create these sorts of spikes, including a possible ground loop through the scope's and compter's earth ground connection.

The sort of filtering you're using should be very effective, so my guess is you're seeing some sort of artifact due to the way the measurement is made.
 
Status
Not open for further replies.
Back
Top