• AVR Freaks

Hot!EMI problems using MCP2515

Page: < 12 Showing page 2 of 2
Author
atestani
Starting Member
  • Total Posts : 13
  • Reward points : 0
  • Joined: 2003/11/07 12:49:52
  • Status: offline
Re: EMI problems using MCP2515 2019/11/14 20:07:12 (permalink)
+1 (1)
I opened a case with Microchip:
 
Case Number :  00477184
Subject :  MCP2515 EMI Issue
 
 I don't know if anyone else can see the case but I will post any responses here as (if) I get them.
#21
ric
Super Member
  • Total Posts : 27654
  • Reward points : 0
  • Joined: 2003/11/07 12:41:26
  • Location: Australia, Melbourne
  • Status: online
Re: EMI problems using MCP2515 2019/11/14 20:47:05 (permalink)
+1 (1)
atestani
I don't know if anyone else can see the case but I will post any responses here as (if) I get them.

No. Anything you post there can only be viewed by yourself and the Microchip employee assigned to it.
 
I have a suspicion that many of these devices are being used by people who have side-stepped the EMC approval requirement.
 

I also post at: PicForum
Links to useful PIC information: http://picforum.ric323.co...opic.php?f=59&t=15
NEW USERS: Posting images, links and code - workaround for restrictions.
To get a useful answer, always state which PIC you are using!
#22
GMAprogrammer
New Member
  • Total Posts : 12
  • Reward points : 0
  • Joined: 2019/11/10 10:01:39
  • Location: 0
  • Status: offline
Re: EMI problems using MCP2515 2019/11/15 03:40:59 (permalink)
0
Hi Nikolay,

here's my findings on the ferromagnetic properties of the chip.
https://youtu.be/_u03B6Gfo2M
 
My Chip ist from 1748. The Chip i tested is the one i desoldered so one that i defenitivly experience the problem from. I also tested a brand new chip and it has the same properties.

While the leads itself seem less attracted to the magnet, the chip itself definitivly is.

-Alex
#23
Antipodean
Super Member
  • Total Posts : 1857
  • Reward points : 0
  • Joined: 2008/12/09 10:19:08
  • Location: Didcot, United Kingdom
  • Status: online
Re: EMI problems using MCP2515 2019/11/15 03:58:37 (permalink)
+1 (1)
I take it this part of the PCB is the MCP2515.
 
One problem I see is that the fill under the chip has only one via to ground. I would be wary of this acting as an aerial at the frequency area you are looking at, especially if 160MHz is your area of concern.
 
I would drill another hole at the top end and use a bit of tinned copper wire through to GND and possibly another at the other end.
 
May not make a difference, but strange things do happen.
 

Attached Image(s)


Do not use my alias in your message body when replying, your message will disappear ...

Alan
#24
GMAprogrammer
New Member
  • Total Posts : 12
  • Reward points : 0
  • Joined: 2019/11/10 10:01:39
  • Location: 0
  • Status: offline
Re: EMI problems using MCP2515 2019/11/15 04:14:51 (permalink)
0
So, to investigate the suspicion that it is a package related problem further i desoldered a MCP2515 from a CAN-Bus Shield i had laying arround and did the same test as yesterday. The Chip is marked 17100TW (Former Chip with the Problem was marked 1748PBS).

Please note the difference in amplitude from the radio stations is not related to an attenduation. I guess they vary their transmission power over time of day.

As you ca see the Spur at 64MHz is no longer passing the 30dBuV. Strange..., the Board functions perfectly fine.
 
Maybe it's a bad batch? Or they changed something in production that causes the issue?

-Alex

Attached Image(s)

#25
GMAprogrammer
New Member
  • Total Posts : 12
  • Reward points : 0
  • Joined: 2019/11/10 10:01:39
  • Location: 0
  • Status: offline
Re: EMI problems using MCP2515 2019/11/15 04:41:01 (permalink)
0
Here is a direct comparison between two boards that are 100% identical and just differ in the MCP2515 beeing used. The Magenta Line is the 17100TW MCP2515. You can see noise performance is improved by a good amount. Also picked up Emissions from Radio stations seem to have been reduced. I will send this board to the test chamber for a pass/fail test.
 
Maybe it's time to order some "newer" MCP2515 Chips to investigate.


 
-Alex
 

Attached Image(s)

#26
atestani
Starting Member
  • Total Posts : 13
  • Reward points : 0
  • Joined: 2003/11/07 12:49:52
  • Status: offline
Re: EMI problems using MCP2515 2019/11/15 09:47:48 (permalink)
+1 (1)
That is *very* interesting. The chip I tested was date code 1811 and it wasn't magnetic at all! I checked a board that was built in the same batch as the two units at the test lab and it was also 1811. I have asked the test lab to open a unit and check the date codes there. It is certainly possible I had two date codes in that board build.  EDIT:  The test lab confirmed the date code in the unit I reworked to 4MHz was also date code 1811.  It definitely had the problem with a 16MHz crystal. 
 
Does anyone have a theory about why the magnetic cases are causing or allowing this emission?
 
 
 
post edited by atestani - 2019/11/19 12:36:28
#27
GMAprogrammer
New Member
  • Total Posts : 12
  • Reward points : 0
  • Joined: 2019/11/10 10:01:39
  • Location: 0
  • Status: offline
Re: EMI problems using MCP2515 2019/11/15 09:57:19 (permalink)
0
I would not blame the magnetic IC Case to be in charge of the EMI Problem. Infact the older 17100TW that i tested succesfully was also magnetic and as you saw above it has not failed the precompliance test i did. But it is infact true that ferromagnetic components in the chip itself could lead to worse EMI performance.

But i am not an expert on that. I would like to hear what Nikolay intended with his guess. Any News on the Ticket you submitted? I read in the forum that Microchip is not really quick with there support. So i have little hope that we actually get some usefull informations in the next couple of days.

Anyhow i just ordered a different batch from another supplier hoping to get another production batch to test with. Unfortunately i have ordered some for production from the same batch that are facing the radiation problem. The important Information i got from all this tests were that it's not in my hands to change somenthing on the PCB side but to find a chip batch that works for the first production run.

-Alex
post edited by GMAprogrammer - 2019/11/15 09:59:57
#28
atestani
Starting Member
  • Total Posts : 13
  • Reward points : 0
  • Joined: 2003/11/07 12:49:52
  • Status: offline
Re: EMI problems using MCP2515 2019/11/15 11:03:00 (permalink)
0
The only news I have on the ticket is a confirmation it has been received.
 
IMHO, I would be concerned about trying to control the date codes of a part that goes into production product.  Without knowing the actual root cause, it isn't clear what a good date code is vs a bad one. In fact, there could be good and bad ones in the same date code.  You could end up with non-compliant product in the field... unless, of course, you test every part.  Again, in my opinion, until this is clear and when (if) we get some information from Microchip, I recommend that you do what I did: change the clock to 4MHz and change the appropriate register on chip initialization.   
 
 
#29
GMAprogrammer
New Member
  • Total Posts : 12
  • Reward points : 0
  • Joined: 2019/11/10 10:01:39
  • Location: 0
  • Status: offline
Re: EMI problems using MCP2515 2019/11/15 11:17:01 (permalink)
0
That's true,

would you mind sharing some of your config? What Crystal, Load Capacitance and Register Settings?

-Alex
#30
atestani
Starting Member
  • Total Posts : 13
  • Reward points : 0
  • Joined: 2003/11/07 12:49:52
  • Status: offline
Re: EMI problems using MCP2515 2019/11/15 12:24:27 (permalink)
0
Sure...
 
The 4MHz crystal I am using now is this one:
https://www.digikey.com/product-detail/en/ECS-40-20-18-TR/XC587CT-ND/191237/?itemSeq=306539952 
with 18 pF caps.
 
I thought you said you were driving this from an Arduino pin?  
 
I can't help much with the register settings as we are driving from a Raspberry Pi and the Linux driver allows the setting directly.  However, best I can tell from the datasheet it is: CANCTRL: CAN CONTROL REGISTER (ADDRESS: XFh) where 
   bit 1-0 CLKPRE[1:0]: CLKOUT Pin Prescaler bits
       00 = FCLKOUT = System Clock/1
       01 = FCLKOUT = System Clock/2
       10 = FCLKOUT = System Clock/4
       11 = FCLKOUT = System Clock/8
 
This is actually for the CLKOUT prescaler setting but the only thing I could find.  This register bits 1-0 defaults/resets to 11 and the default crystal is 16MHz so one could assume that it is using the divide by 8 prescaler and that the "system clock" is 2MHz.  If any of that is true then, for a 4MHz crystal,  I would set bits 1-0 to 01 = FCLKOUT = System Clock/2.
 
#31
Nikolay_Po
Super Member
  • Total Posts : 1940
  • Reward points : 0
  • Joined: 2012/04/01 13:49:27
  • Location: Russia, Novorossiysk
  • Status: offline
Re: EMI problems using MCP2515 2019/11/16 02:56:04 (permalink)
+1 (1)
About ferromagnetic IC frame response. I've knew that the components with ferromagnetic leads are producing more non-linearity, more harmonics in analog domains. Not as much as percents but there is a difference in high precision circuits. Also I know that ferromagnetic materials are successfully used as magnetic antennas up to 70+ MHz range. So I believe that ferromagnetic materials, used in IC package, will change both radiated EMI harmonics envelope and level. It's probably the factory producing the chips is changing the frame alloy by material availability and this leads to differences in EMI performance.
Also Alex's repeatability problem of EMI level should be investigated. May be there is some critical step in initialization sequence or timings, in operating modes, etc. The problem may be different than chip frame material change.
-=-
I have an idea! Try to measuring EMI while magnetize the IC by strong magnet different ways! Depending on magnetic induction remainder in IC frame meterial after magnetic field application. I'm predicting that largest EMI will be after IC frame demagnetization, because of higher permeability and lowest EMI will be under strong magnetic field because of lowest permeability closer to magnetic induction saturation.
So, my theory is the external magnetic field and magnetic induction remainder of ferromagnetic IC frame will change EMI level. A joke: place a DC coil around the IC and put constant current though the coil. Make current source programmable to let end-user to choose a trade-of between an EMI level and power consumption. :)
post edited by Nikolay_Po - 2019/11/16 02:57:13
#32
atestani
Starting Member
  • Total Posts : 13
  • Reward points : 0
  • Joined: 2003/11/07 12:49:52
  • Status: offline
Re: EMI problems using MCP2515 2019/12/06 10:33:38 (permalink)
0
Alex
 
Did you ever get the design information you asked me for?  
 
Also, Microchip has been communicating with me on the ticket i opened and are looking for at least the trace code (in addition to the date code) on the parts we found that have the problem.  I can't tell what the trace codes are since all of my hardware is at a test lab.  Can you supply trace code information on some bad chips you have so I can forward the data to them?  Thanks 
#33
Page: < 12 Showing page 2 of 2
Jump to:
© 2020 APG vNext Commercial Version 4.5