AT90USB high level noise 96MHz

Go To Last Post
10 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hello,

Does anyone know solution for high level noise in AT90USB series? There's 96 MHz noise as high as it can be found by FM radio station at 96MHz frequency(I'm using AT90USB646 in vehicle) Maybe it's from USB? (2*48MHz)

I'd like to bug report on atmel page but can't find any tool for it..

Best regards
Arthur

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Noise is not really a bug, it's largely dependant on the device emitting it. Atmel may have made the chip but they didn't make the consumer electronics emitting the interference. Is that noise from a project of yours or from some device you are using?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It's definitely from working chip. I tested supply circuit, changed filters etc. Finally I disconnected all peryph on board (only at90usb646 left) and noise is still at 96MHz. What important, it's exacly at 96.0 MHz, not 95.9 and not 96.1 - just 96.0 and can be easy find as radio station. Only when I disconnect XTAL noise quiten down. Tomorrow I'll try to disable all peryph builded in chip (usb, timers, adc) but I think it's propably from USB.

We offered this device sucessfull for about year, just one customer found out, when he plug device he can't hear his favourite station (there's only one FM station in his small village - at 96.0 MHz ;) We tested it and .. he was right!

Last Edited: Wed. Jan 25, 2012 - 08:59 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Such noise is almost always radiated from the board traces or connected wiring. Its very likely the harmonic of an oscillator in the system.

How do you get rid of it? Lots of bypass caps. GOOD ones (SMT, short traces). Lots of ferrite beads. Metal box. Ferrites on board I/O and power. Common mode choke on power. And, lots of magic dust, which you will need in great quantity.

Atmel cannot do diddly about it. It is up to you to solve. There is at least one AVR app note on EMI which is work reading.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

To save some space and such, there are monolithic pi filters (C-R-C) with TVS or ESD diodes that one can use for the I/O lines.
Sources are Texas Instruments, On Semiconductor, etc.

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

OP, post a schematic of your oscillator and power supply sections and we can probably help a bit more...

Personally I find it hard to believe it's emitting to the extent of scrambling a FM radio station with oh say 25KW ERP a few miles away... If it had even a minimum of bypass caps and a semi-decent layout it might at worst sound like an annoying hum or static.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What CAN happen is that a signal is so strong that it makes the receiver turn down the RF "gain", making the proper signal appear weaker. This is called "desensing" and is a common problem where many radios are used in close proximity, even if they are not on the same frequency.

A very important question is: how far is your board from the radio?

But, yes, it does seem a bit odd that the signal should be so strong. On the other hand, I have noticed that circuit boards (of almost all sizes) seem to be worst in the range of 80MHz to 120MHz, even with much lower frequency clocks (8MHz for example). There must be something about board physical sizes or trace lengths that tend to do their worst in that frequency range.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
There must be something about board physical sizes or trace lengths that tend to do their worst in that frequency range.
And the bypass capacitor network.
Done incorrectly this can result in harmonic resonances in the planes and power traces.
One digital guru, Dr. Howard Johnson, tries to keep the L/C of regulator bulk capacitor and sum of bypass capacitors roughly similar; but, that is one technique of at least several (a difficult topic).
Sometimes even need to increase the ESR of a small bypass capacitor by adding a resistor or the ESR of an regulator input capacitor to prevent over-voltage or reduce EMI.
Probably quicker to verify the problem with a scope then simply try educated/experienced guesses or rough estimates.
Ground planes help a lot for low impedance circuits (confines the EM field).

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have lots of respect for Howard Johnson. For an academic, he has provided some very useful and practical techniques.

Part of the problem with the case at hand is that there is so little to go on. To make anywhere close to useful suggestions, a schematic, a board layout, and several photographs (at varying zoom values, including external wiring) would be the very minimum. Without this, the only thing that can be offered are generalities that might not have any bearing on the case at hand.

Another important question: does the interference occur with the USB cable removed?

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

There is at least one AVR app note on EMI which is work reading.


It's either AVR040 or AVR042 - all h/w designers should read both anyway.

(EDIT: 040 in fact)