Helpful ReplyHot!MCP2561 production failure rate

Author
EdG
Starting Member
  • Total Posts : 9
  • Reward points : 0
  • Joined: 2015/01/05 09:42:13
  • Location: Leicestershire, UK
  • Status: offline
2017/04/10 14:02:49 (permalink)
0

MCP2561 production failure rate

Having just moved from MCP2551 to MCP2561 for our CAN transceivers I was expecting an improvement in board assembly yield at manufacture; however one of our modules (PIC18F25K80 with MCP2561) currently has a 75% failure rate for CAN communication. I'm suspecting this to be a batch problem, perhaps with an incorrect solder reflow profile, but I cannot find any mention of the recommended profile in the datasheet or online documentation (aside from a maximum pin temp of 300degC for <10s).
 
Has anyone else had issues with this transceiver, specifically around soldering temperature and times? I haven't tested them all yet but it seems a transceiver change fixes the problem so it looks like the IC itself. Is there anything I can test directly to confirm an issue perhaps? Any and all information gratefully received!!
 
Thanks,
 
Ed
#1
CinziaG
morite
  • Total Posts : 3144
  • Reward points : 0
  • Joined: 2016/12/07 14:20:36
  • Location: Wien
  • Status: offline
Re: MCP2561 production failure rate 2017/04/10 14:19:47 (permalink)
0
A customer just started using it so... have not much info about that board...

mi fate schifo, umani di merda.
#2
Jerry Messina
Super Member
  • Total Posts : 325
  • Reward points : 0
  • Joined: 2003/11/07 12:35:12
  • Status: offline
Re: MCP2561 production failure rate 2017/04/10 17:34:45 (permalink)
+1 (1)
I was expecting an improvement in board assembly yield at manufacture

Why? What was wrong with the MCP2551 assembly that you expected using an MCP2561 would fix?
 
I think you need to find the real failure here. Unless you have a completely incompetent board house it's likely not the assembly that's the issue, esp. if you're using the DIP or SOIC package.
 
What's your setting for the CIOCON.ENDRHI bit? Many times setting that to 1 (instead of leaving it to the default of 0) helps.
#3
qhb
Superb Member
  • Total Posts : 7175
  • Reward points : 0
  • Joined: 2016/06/05 14:55:32
  • Location: One step ahead...
  • Status: online
Re: MCP2561 production failure rate 2017/04/10 20:11:07 (permalink)
+1 (1)
egummow
...
one of our modules (PIC18F25K80 with MCP2561) currently has a 75% failure rate for CAN communication. 



Does this mean 75% of boards are totally dead?
Or the board runs, but CAN doesn't work at all on 75% of them?
Or you get 75% errors when using the CAN interface on these boards?
 
 
#4
EdG
Starting Member
  • Total Posts : 9
  • Reward points : 0
  • Joined: 2015/01/05 09:42:13
  • Location: Leicestershire, UK
  • Status: offline
Re: MCP2561 production failure rate 2017/04/11 00:11:21 (permalink)
+1 (1)
75% of the boards show no CAN communication at all when plugged into the test rig, but program up ok. I need to get a scope on the pins to confirm that the issue is with the transceiver.
#5
EdG
Starting Member
  • Total Posts : 9
  • Reward points : 0
  • Joined: 2015/01/05 09:42:13
  • Location: Leicestershire, UK
  • Status: offline
Re: MCP2561 production failure rate 2017/04/11 00:17:52 (permalink)
+1 (1)
We had some failures with the MCP2551, exact problem undiagnosed but swapping out the CAN transceiver resolved it. So you'd think an external issue perhaps but all modules use the same transceiver running at 5V so we decided the most likely cause (looking at forums and other people's experiences) that ESD could have caused it, which is one of the things that is improved on the MCP2561. Having reviewed the Microchip MCP2551 to MCP2561 migration application note and confirming we should have no issues we made the decision to swap over.
My theory about the reflow temperature profile is that we don't seem to have had any issues when the ICs were hand soldered (prototype boards) by myself but only when delivered fully assembled. Of course ESD could still be the cause or at least a factor, but then (as per the original post) I'd have expected a better yield.
#6
EdG
Starting Member
  • Total Posts : 9
  • Reward points : 0
  • Joined: 2015/01/05 09:42:13
  • Location: Leicestershire, UK
  • Status: offline
Re: MCP2561 production failure rate 2017/04/11 00:19:25 (permalink)
0
I'll check the CIOCON.ENDRHI bit setting too - thanks for the tip. And it's the SOIC8 package we're using.
 
#7
Jerry Messina
Super Member
  • Total Posts : 325
  • Reward points : 0
  • Joined: 2003/11/07 12:35:12
  • Status: offline
Re: MCP2561 production failure rate 2017/04/11 03:40:21 (permalink)
+1 (1)
we decided the most likely cause (looking at forums and other people's experiences) that ESD could have caused it

I suppose anything's possible, but if that's the problem I'd be looking for a different board house. And, ESD aside, even I can solder an SOIC package so I doubt that's the issue either.
 
Since you were seeing this with the old board design, does your board and/or test fixture have the req'd termination between CANH/CANL? If you're running at low freqs you might get by with only one, but both should really be there.
#8
EdG
Starting Member
  • Total Posts : 9
  • Reward points : 0
  • Joined: 2015/01/05 09:42:13
  • Location: Leicestershire, UK
  • Status: offline
Re: MCP2561 production failure rate 2017/04/11 04:52:48 (permalink)
+1 (1)
Yes, 120R at each end of the bus. We're only running at 100k too.
#9
EdG
Starting Member
  • Total Posts : 9
  • Reward points : 0
  • Joined: 2015/01/05 09:42:13
  • Location: Leicestershire, UK
  • Status: offline
Re: MCP2561 production failure rate 2017/04/20 05:38:52 (permalink)
+3 (3)
Ok, in case anyone finds this useful, we found the problem.
We found that changing the microcontroller, at first, resolved the problem. Then it didn't. Turns out the first few replacements were older PICs, the later ones were the most recent Version (same as originally fitted). We noted that when it didn't work the on-board regulator got *hot*, as did the MCP2561. My conclusion was that the PIC was sinking current from the CAN transceiver and an analysis of the code showed an error where the PIC CAN TX was set to an input and the CAN RX to an output (and had always been so) - previous hardware versions clearly ignored these settings but the latest version does not.
Anyway, a bit of stupidity on our part but at least we managed to work out the cause.
#10
CinziaG
morite
  • Total Posts : 3144
  • Reward points : 0
  • Joined: 2016/12/07 14:20:36
  • Location: Wien
  • Status: offline
Re: MCP2561 production failure rate 2017/04/20 06:03:07 (permalink)
0
Well, good! Smile

mi fate schifo, umani di merda.
#11
almec
New Member
  • Total Posts : 1
  • Reward points : 0
  • Joined: 2017/05/04 13:18:34
  • Location: 0
  • Status: offline
Re: MCP2561 production failure rate 2017/05/10 13:06:41 (permalink)
+1 (1)
We found a behavior similar to that described. This problem was solved twisting the CAN pair.
#12
tolliug
New Member
  • Total Posts : 3
  • Reward points : 0
  • Joined: 2018/08/07 02:50:57
  • Location: 0
  • Status: offline
Re: MCP2561 production failure rate 2018/09/12 09:44:04 (permalink)
0
We experiment a lot of failure using PIC18F46K80 and MCP2561 especially after lot of wild switching off/on power supply. Changing the transceiver is the only way to get  proper board again. Thus we need to protect or prevent something like ESD. Do you have any clue? 
#13
Antipodean
Super Member
  • Total Posts : 1647
  • Reward points : 0
  • Joined: 2008/12/09 10:19:08
  • Location: Didcot, United Kingdom
  • Status: offline
Re: MCP2561 production failure rate 2018/09/12 10:00:16 (permalink)
0
tolliug
We experiment a lot of failure using PIC18F46K80 and MCP2561 especially after lot of wild switching off/on power supply. Changing the transceiver is the only way to get  proper board again. Thus we need to protect or prevent something like ESD. Do you have any clue? 


https://www.bourns.com/docs/technical-documents/portnote-solutions/de/Bourns_CANbus_Srg_Prot_ESD_PortNote.pdf?sfvrsn=410d2b2e_3
 
Other manufacturers offer similar products.
 

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

Alan
#14
tolliug
New Member
  • Total Posts : 3
  • Reward points : 0
  • Joined: 2018/08/07 02:50:57
  • Location: 0
  • Status: offline
Re: MCP2561 production failure rate 2018/09/12 11:37:27 (permalink)
0
We plan to add TVS however our power supply can vary from 9 to 36 volts, thus choosing the right one is not obvious. I will go on to investigate deeper and will let U know
#15
hexreader
Super Member
  • Total Posts : 999
  • Reward points : 0
  • Joined: 2008/05/04 03:06:55
  • Location: England
  • Status: offline
Re: MCP2561 production failure rate 2018/09/14 12:23:58 (permalink) ☄ Helpfulby Jim Nickerson 2018/09/15 09:02:22
+1 (1)
Long ago, I had loads of problems with CAN transceivers burning up on a simple test bench setup. I since learned that providing a ground connection between boards is needed, i.e. 3 wires - CANH, CANL, Ground (or more accurately - common). I had no further transceiver failures once I learnt to provide a common connection.
 
No idea whether this might be related in any way, but I thought I would mention it in case.
 

Experienced Hobbyist
#16
SpokaneNexus
Super Member
  • Total Posts : 386
  • Reward points : 0
  • Joined: 2013/02/05 14:53:25
  • Location: 0
  • Status: offline
Re: MCP2561 production failure rate 2018/09/15 10:18:16 (permalink)
+1 (1)
No CAN transceivers "burning up" here, but I can definitely second that comment about needing a ground reference. In my early days with CAN I made what I thought was a very understandable presumption, that "differential" meant no ground reference was required. But a scope examination of the CANH and CANL signals quickly shows they are NOT differential in the normal sense, they do not have the same dynamic range, etc. It's a shame Bosch didn't specify CANbus to be a true balanced differential pair, but here we are.
#17
tom_usenet
Super Member
  • Total Posts : 502
  • Reward points : 0
  • Joined: 2011/12/07 21:32:51
  • Location: 0
  • Status: offline
Re: MCP2561 production failure rate 2018/09/26 16:34:10 (permalink)
+2 (2)
> It's a shame Bosch didn't specify CANbus to be a true balanced differential pair, but here we are.
 
CAN is primarily for use in cars. Cars are a big metal box (well they were in 1981 when CAN was developed) so a common ground is assumed. "True balanced differential" needs transformers at both ends, like Ethernet has. It also needs "DC_Balanced" waveforms, which CAN certainly doesn't have.
 
If you need isolation, like working in a factory with large voltage differences between large pieces of gear, then you should add opto-isolators, or use isolated CAN drivers.
 
> I thought was a very understandable presumption
 
It is always a good idea to read the data sheets for the chips you're using. You might find out where your assumptions are about to cause expensive damage. The Data Sheets always specify the maximum differential range for operation, and the wider differential range where it won't work but it won't blow up. Get outside that range and you'll blow them up.
 
Test on the bench with a common power supply and the common earth is present, if not obvious. Test with two separate floating two-pin plug-packs and you can easily get 100V or more differential voltage between the units.
 
The development boards that have TWO PIN CONNECTORS for CAN should be ashamed of themselves...
 
Tom
 
#18
Jump to:
© 2018 APG vNext Commercial Version 4.5