Unconnected pins - What to do for EMI and EMC performance

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

Hi guys, I am developing the final version of a PCB board for medical applications and I need to have an extremally good EMI and EMC performance. I have searched but I did not find anything related to unconnected pins in this situation.

 

How should I proceed with unconnected pins? Should I ground then? External or internal pull?

 

I am not really concern about power savings.

 

Ps: I am using Atxmega-A and Atxmega-E microcontrollers.

 

Thank you. 

 

 

Regards!
FS.

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

tying pins directly to power or ground could be risky if a software bug accidentally changes them to outputs.

 

I am not sure if the internal pull ups would be sufficient as I am no expert on EMI ( Also it is just as Ilkley for a software bug to witch them off.) so I suspect an external pull up/down would be best

 

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

IPguru wrote:

tying pins directly to power or ground could be risky if a software bug accidentally changes them to outputs.

 

I would think that a bug in a medical product causing an output to die is the least of the OP's problems!!!

 

As the default state is INPUT then you certainly want to do something. Tying them low is probably the more common as long as you know 100% that you'll never need that pin for a later mod. If you think you might nmeed a field mod then tie it low with a resistor. Although I suspect field mods of a medical device will invalidate the certification.

 

I've also seen, and done it myself, where an unused pin is set in software as an output and left low.

 

I assume that you are getting theses things certified? In that case, pick a method, get it tested, and if it passes then you're good to go.

"This forum helps those that help themselves."

"How have you proved that your chip is running at xxMHz?" - Me

"If you think you need floating point to solve the problem then you don't understand the problem. If you really do need floating point then you have a problem you do not understand." - Heater's ex-boss

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

Brian Fairchild wrote:
I assume that you are getting theses things certified?

In which case - especially as it's a medical device - shouldn't you be hiring an expert to help with this ... ?

 

 

Top Tips:

  1. How to properly post source code - see: https://www.avrfreaks.net/comment... - also how to properly include images/pictures
  2. "Garbage" characters on a serial terminal are (almost?) invariably due to wrong baud rate - see: https://learn.sparkfun.com/tutorials/serial-communication
  3. Wrong baud rate is usually due to not running at the speed you thought; check by blinking a LED to see if you get the speed you expected
  4. Difference between a crystal, and a crystal oscillatorhttps://www.avrfreaks.net/comment...
  5. When your question is resolved, mark the solution: https://www.avrfreaks.net/comment...
  6. Beginner's "Getting Started" tips: https://www.avrfreaks.net/comment...
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

That's only the tip of the iceburg---what else is on the board?  A 1000W switching drive would likely completely obscure the puny micro.

Are all of your I/O's esd protected?  Using a gnd plane?  You can use a dithering xtal osc to blur out the peaks in the xtal spectrum (easier to pass radiated emissions).

When in the dark remember-the future looks brighter than ever.

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

I usually leave unused pins unconnected, and init all port pins to be outputs and then set them low. 

This way they are less likely to radiate, or pick up any unwanted radiation.....

YMMV,

 

Jim

Mission: Improving the readiness of hams world wide : flinthillsradioinc.com

Interests: Ham Radio, Solar power, futures & currency trading - whats yours?

 

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

Brian Fairchild wrote:

I assume that you are getting theses things certified? In that case, pick a method, get it tested, and if it passes then you're good to go.

 

Yes, I am. I have already made one test, the board has passed, but pretty close to IEC60601 limits. I am now re-projecting the board and all filters (because I know in which frequencies my board irradiates.

 

I will do a second test, but I don't wanna miss any aspect now.

Regards!
FS.

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

awneil wrote:

Brian Fairchild wrote:
I assume that you are getting theses things certified?

In which case - especially as it's a medical device - shouldn't you be hiring an expert to help with this ... ?

 

As it matters, I am the expert at this point. However, Atmel does not have a specific document describing the procedure of what to do with unconnected pins - ST is the only manufacturer [that I found so far] that completely describe what to do in this case, they even describe the value of the resistance that I should connect to unused pins.

 

avrcandies wrote:

That's only the tip of the iceburg---what else is on the board?  A 1000W switching drive would likely completely obscure the puny micro.

Are all of your I/O's esd protected?  Using a gnd plane?  You can use a dithering xtal osc to blur out the peaks in the xtal spectrum (easier to pass radiated emissions).

 

I know, I already used multiple capacitors, choke inductors, inductors, ferrites... But like I said, I don't wanna miss any point of it.

We have had no problems with ESD.

Regards!
FS.

Last Edited: Fri. Dec 29, 2017 - 04:44 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

This way they are less likely to radiate

I wonder if it makes any difference...high, low, in/out...it all ties into the noisy guts of the micro (in very close internal chip proximity). Gnd inside the chip might not be the same good RF gnd as the external copper gnd plane.  Things like not accidentally creating loop antennas with trace routings is probably 100x potentially more problematic. 

When in the dark remember-the future looks brighter than ever.

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

avrcandies wrote:

This way they are less likely to radiate

I wonder if it makes any difference...high, low, in/out...it all ties into the noisy guts of the micro (in very close internal chip proximity). Gnd inside the chip might not be the same good RF gnd as the external copper gnd plane.  Things like not accidentally creating loop antennas with trace routings is probably 100x potentially more problematic. 

 

I partially agree. But why, then, some manufacturers say that you should do one or other thing related to those pins?

 

Regards!
FS.

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

F Schneider wrote:

...the board has passed, but pretty close to IEC60601 limits.

 

I would be surprised if unconnected pins were your problem. As the default state is an input the bigger problem when they float is increased supply current caused by the input stage sitting in a linear regions around VCC/2.

 

Do you have access to measuring equipment without going back to the test lab? A few simple firmware tweaks should help eliminate the pins as the cause.

"This forum helps those that help themselves."

"How have you proved that your chip is running at xxMHz?" - Me

"If you think you need floating point to solve the problem then you don't understand the problem. If you really do need floating point then you have a problem you do not understand." - Heater's ex-boss

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

I, also, suspect that the pins, themselves, do very little. In my experience, most emissions come from traces. 

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

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

Brian Fairchild wrote:

Do you have access to measuring equipment without going back to the test lab? A few simple firmware tweaks should help eliminate the pins as the cause.

 

The test equipment that I have here would not describe well any worsening or improve in this context...

Regards!
FS.

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

Did you say you used a gnd plane?  That is one of your best friends.

Don't create loop antennas (hidden or otherwise)

Don't have a super-sharp clock (lots of harmonics).

All busy pwr fets should have some gate resistance (soften the switching)

When in the dark remember-the future looks brighter than ever.

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

avrcandies wrote:

Did you say you used a gnd plane?  That is one of your best friends.

Don't create loop antennas (hidden or otherwise)

Don't have a super-sharp clock (lots of harmonics).

All busy pwr fets should have some gate resistance (soften the switching)

 

I am using a 4 layers PCB, so I have a GND plane.

I am avoiding antennas, as much as I can. All signal cables leaving the board has filters.

I am using the Xmega feature to limit the rising and falling of digital pins.

My fets have an RC filter in their gates.

Regards!
FS.

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

FS,

 

Glad to hear you have a product going into production.

Congrats!

 

It wasn't clear from your OP if the concern is EMI/EMC radiated by your device, or that generated from other devices being picked up by your device and interfering with its operation.

Both are a major concern for the health care environment, obviously.

 

There are entire books written on the topic of reducing radiated emissions.

 

The issue is if your device is rolling down the hall past the radiology department's X-Ray power supplies and X-ray machines, or past an MRI machine, (or, heaven forbid, trying to get certification for operation within the immediate vicinity of the MRI machine...), every trace on the PCB will act like an antenna picking up EMI.

 

If your enclosure will have a Faraday cage surrounding the micros' PCBs then you have clearly reduced, significantly, the impact of the small differences in some of the options expressed above.

 

For unused pins, I seem to recall making the output, low, with no trace at all connected to the pin's pad, which is still present for mechanical integrity reasons.  YMMV.

 

Obviously, with the wide range of Xmegas available, select one which meets your requirements, but doesn't have tons of extra pins to even have to deal with. 

 

Good luck with your project!

 

JC

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

The product can't fail in the next pre-test, so the reason for my concern. I am looking for and eliminating every single point of possible EMI or EMC.

 

Like you guys have mentioned, I will set the pin as an output and clear it. It should be enough.

 

Thank you all for the tips.

Regards!
FS.

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

Part of susceptibility testing is seeing how the product reacts to "noise". Remember, software can be your friend....rather than just polling the "apply X-rays" switch & respond to the first "hi", you could require the switch has to be held continuously for 0.125 sec (much longer than typical debounce).    If you update a display whenever a knob is rotated, maybe consider updating all the time instead---so if noise garbles the display it immediately refreshes & recovers.  There are a number of things that can be done to "harden" the system against failure (of course some of those might make it more failure-prone).

 

Remember, widest traces are most desired (ultimately, such as a full plane), to reduce inductive effects.  Conversely, in a metal housing, lots of small holes are much better than fewer large holes.  Imagine a strand of RF, "spaghetti" signal trying to jump out.  A long very thin  (or thick) slot is just the thing it needs to escape (or get in)...keep the longest escape dimension as small as possible.  Joining two metal plates requires lots of screws/rivets/bushings to avoid a long "crack" ...of course welding is better!

  

When in the dark remember-the future looks brighter than ever.

Last Edited: Fri. Dec 29, 2017 - 08:56 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

Here is a point that I have used...

 

Vcc wiggles around, relative to ground, because of the current spikes on every clock edge. You can add capacitors and ferrite beads until a certain part of the realm freezes over and you can change the amplitude but you can never totally eliminate it.

 

Now, if an unused pin is set to be a logic high, it, also, will wiggle around, pretty much like Vcc (output PMOS pull up to Vcc is on, output NMOS pull-down to ground is off). So, if you make that unused pin a low output, then, internally, there will be a low resistance (NMOS) to ground and a quite high resistance (PMOS) that does almost nothing. In theory, then, those logic output low pins should have less power supply noise on them.

 

Whether or not this makes any quantitative difference is unknown to me (I've never been in a situation where I could measure it). BUT, that is what I try to do.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

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

Is it just me or should active medical equipment really be designed by someone just learning on the job? Sorry if that sounds harsh but I'd hope the electronics I might be hooked up to would have been designed by an experienced engineer skilled in that area. If you don't have that skill then please think about hiring someone who does and then not only use them for this design but also to learn from so you develop the same skill set for next time.

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

clawson wrote:

Is it just me ...

 

All medical devices from the humble bandage to things like ECG machines fall under the Medical Devices Directive in the EU. This requires a bit more than the simple self-certification route that you can use on other electrical equipment before a CE mark is attached. There's a bit of an explanation here...

 

https://www.gov.uk/guidance/medi...

 

Class 1 devices are regarded as low-risk but even they need approval from a notified body once they start to be 'active' in any way. Plus registration with the relevant national authority.

"This forum helps those that help themselves."

"How have you proved that your chip is running at xxMHz?" - Me

"If you think you need floating point to solve the problem then you don't understand the problem. If you really do need floating point then you have a problem you do not understand." - Heater's ex-boss

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

F Schneider wrote:
How should I proceed with unconnected pins?
Is the PDI clock (i.e. reset) signal always near the ground plane?

Reason: EMC (ESD, EFT, lightning)

If PDI is on a cable or at a PCB edge then will need some series resistance to decrease injection current (reduce risk of reset, clock upset, latch-up)

One LUFA AVRISP2 PDI interface has 100R series; IIRC, Dean recommended 220R.

megaAVR has a 330R series between the reset switch and reset per AVR042 (AVR hardware design)

Some tactile switches have a shield to direct the ESD strike to ground.

XMEGA do not have an ESD limit in the datasheets.

 

Art of Electronics has a section on how to harden digital signals against ESD and likewise with analog signals for some lightning.

Analog signals can typically have an order of magnitude greater series resistance with some capacitance; some op amps and comparators have EMI filters.

 

http://ww1.microchip.com/downloads/en/appnotes/atmel-2521-avr-hardware-design-considerations_applicationnote_avr042.pdf (AVR042)

http://www.microchip.com/wwwappnotes/appnotes.aspx?appnote=en590907 (AVR040, EMC)

http://www.cambridge.org/us/academic/subjects/physics/electronics-physicists/art-electronics-3rd-edition#lDEyvsCCTJlFMDDF.97 (Art of Electronics, third edition)

 

"Dare to be naïve." - Buckminster Fuller

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

Doesn't take much T&E instruments and tools to get a warm fuzzy for EMC and EMI; same for ESD.

T&E for EFT and lightning takes some thought.

 

A form of instrument is the device-under-test within some reasonable range of a spark gap generator (i.e. wide band with some power) (an ESD strike within a room can upset or destroy electronics in that room)

Ideally the application is operationally continuous with I/O tests and C asserts; even better is additional continuous built-in-test (BIT)

 

Risk = probability of failure * consequence of failure

Each issue goes through risk analysis (risk computation, risk evaluation)

Proposed risks, actualized risks

Hypothesis, design-of-experiment, experiment, data -to- information -to- knowledge

 

Failure probability may need adjustment due to instrumentation dwell (sample time duration / period)

 

High Frequency Measurements Web Page
Douglas C. Smith

Technical Tidbit - March-April 2014

http://emcesd.com/tt2014/tt040114.htm

Troubleshooting Radiated and Conducted Immunity Problems in the Development Lab
Abstract: Tackling radiated and conducted immunity problems can be difficult because of the high cost of the equipment and chamber normally used to perform these tests. An inexpensive test bench setup that can effectively find radiated and conducted immunity problems is presented.

...

Microchip Technology Inc

Microchip

AVR1610: Guide to IEC 60730 Class B Compliance with XMEGA

http://www.microchip.com//wwwAppNotes/AppNotes.aspx?appnote=en591027

via http://www.microchip.com/wwwproducts/en/atxmega128a1u

 

"Dare to be naïve." - Buckminster Fuller