PDA

View Full Version : 16F876A chip failures



badrad
- 14th March 2005, 04:41
Has any members here experience pic chip failures?

We have been experiencing an occasion failure of the 876A pic, where the lcd winds up displaying on the top line - black bars. the pic might have been working for months perfectly, and this failure suddenly happen. we haven't been able to identify particular circumstances, but the pic is "toast".

the pic is in a board the basically reads an resistance ladder value and displays the information onto an optrex lcd 2x16 display. the hardware and software have been relatively stable and we have around 50 or so out in the field.

we have had a couple fail over the past few month while with this problem. while not overly alarming, it is of concern since we don't really have a handle on the problem.

just wondering if any other members have had seen or experience this problem?

mister_e
- 14th March 2005, 07:37
More than often this is caused by the unused pins or power supply spikes.

If you have unused pins set them to input and tie them to grounds.

badrad
- 16th March 2005, 09:03
More than often this is caused by the unused pins or power supply spikes.

If you have unused pins set them to input and tie them to grounds.
thanks for the assist. i'm reworking the pcb, and there are a few unused pins that i had left floating. i tied them to ground, and will rework the software to make them inputs.
i'll have take a look at how to eliminate power spikes.

titanvision
- 28th July 2005, 07:05
Well, I have same problems like you have Badrad , I am totally agree with Mister e but my case all unused pins set it up input and tide up the ground, however among 200 PCS PIC 16F876A 20/P 40PCS was still totally broken After some time,

And strange thinks that some of them look like works but some ports are defected.


Finally we trace the problem from EMI interferences chip affected, at the moment it looks okay, but still worry? May be also it is good idea to look some other problems.
Also for your information I changed controller from previous one to PIC 16F876A I/SP

regards

badrad
- 28th July 2005, 15:38
Well, I have same problems like you have Badrad , I am totally agree with Mister e but my case all unused pins set it up input and tide up the ground, however among 200 PCS PIC 16F876A 20/P 40PCS was still totally broken After some time,

And strange thinks that some of them look like works but some ports are defected.


Finally we trace the problem from EMI interferences chip affected, at the moment it looks okay, but still worry? May be also it is good idea to look some other problems.
Also for your information I changed controller from previous one to PIC 16F876A I/SP

regards
So, I am not totally alone with my problems...
Did you rework your design or shield your PIC?

As for my problem, since then, I had to rework my pcb and included the following:
1. left the unused pins alone, but defined them as outputs. (now this might sound contrary and I did initially ground them and assign as inputs, but the tech people at Microchip recommended this to me).
2. put in sufficient ground plane.
3. installed MOV at my A/D input as well as at the power input.
4. increased my A/D input resistor from 1K to 20K.
5. keep my fingers crossed...

All my PICs have been the 16F876A I/SP.

mister_e
- 28th July 2005, 23:49
As for my problem, since then, I had to rework my pcb and included the following:
1. left the unused pins alone, but defined them as outputs. (now this might sound contrary and I did initially ground them and assign as inputs, but the tech people at Microchip recommended this to me).
2. put in sufficient ground plane.
3. installed MOV at my A/D input as well as at the power input.
4. increased my A/D input resistor from 1K to 20K.
5. keep my fingers crossed...


I agree with the ground plane idea, i should said it before.

For MOV at the A/D input... yes/no/toaster it depend of where the signal come from but some 1n5817 could do the job... it's always case by case anyways.

Increasing the A/D input resistor... i can't see how it will help and it's totally out of the Microchip spec... well in the datasheet.

But for the left pin open and set them to output... OOOOOOH no i totally disagree. I did it in my first projects as this is what Microchip sugested to me too. I had a loooooooooad of unit that comes back to me after few months... PIC burn.... Solution, set to input, tie them to ground. Most of those unit were dedicated to do some test in car engine... YEAH you want noise... that's one of the best place to test it. After the modification, i place some units
2 X close to the alternator
2 X close to the sparks plug
2 X close to distributor

in my cars, my friends car and run with them for 1 months. Those unit had no metal shielded box... only sit on a plastic Ziploc bag... with a touch of silicone. After the test... no more problem with those modified, most of those unmodified still had the same problem after the test.

I still produce those units. I never, never, never see the same problem. now exactly 5 347 are on the market. If i look to my warranty/exchange list and i remove those from the first edition, i have only 8 which come back for a 7 segment problem... Big deal !!!

I never use internal pull-up for final product... always place an external pull-up 1-4.7K... yeah more power hungry but save my butt in many case.

I'll be curious to see your actual schematic and board layout to see what i can suggest to you.

mister_e
- 28th July 2005, 23:56
Just in addittion... chances are that you use a 16F and not a 18F.... Those 18F or some 16F SERIE A are more sensitive to EMI and all other noise.

A good PCB, hardware, cabinet design combination is not as this easy to do.

I can't talk for them but i'm pretty sure that Ralph and Melanie (those i know who design stuff for industrial noisy environement) will agree with that.

Melanie
- 29th July 2005, 02:15
When the 'A' series PICs started to come through, then they along with the 18F series got a lot of bad press. So much so, that rather than risk wasting precious time with possible problems, I delayed their introduction. Since then, the Purchasing Department forced my hand because they argued that the 16F876A series (and others) were much cheaper than the original non-A. As stocks ran out of the older parts I switched across... and guess what?.... Nothing! Zippo. No rejects, no returns, no feedback... they go out the door and it's like they've fallen into a Black Hole and are never seen or heard from again.

The only major problems I've ever encounterd across the last four years were with a particular brand of op-amp in a voltage doubler circuit...

http://www.picbasic.co.uk/forum/showthread.php?t=245

After a lot of investigation, I discovered the failure was always with National Semiconductor as the manufacturer, and checking deeper discovered that the internal schematic of their part was flawed. Banned the use of Nat Semi and use TI or ST instead - no more problems.

mister_e
- 29th July 2005, 05:00
mmm interesting point about National. Well i'm happy to hear that and i'm proud now to have more TI in stock than National.

Any bads with Analog Device on your side??? As now, not here yet.

NavMicroSystems
- 29th July 2005, 11:47
...Any bads with Analog Device on your side??? As now, not here yet.

Steve,
I have loads of Analog Device parts out in the field
(in really harsh environments)
and I have never had any problem.

mister_e
- 29th July 2005, 17:11
Great. Thanks Ralph.

titanvision
- 2nd August 2005, 04:53
So, I am not totally alone with my problems...
Did you rework your design or shield your PIC?

As for my problem, since then, I had to rework my pcb and included the following:
1. left the unused pins alone, but defined them as outputs. (now this might sound contrary and I did initially ground them and assign as inputs, but the tech people at Microchip recommended this to me).
2. put in sufficient ground plane.
3. installed MOV at my A/D input as well as at the power input.
4. increased my A/D input resistor from 1K to 20K.
5. keep my fingers crossed...

All my PICs have been the 16F876A I/SP.

Hi badrad

Well yes I did re-design board again and, I kept unused pin as still input, but I take care
Ground plane very beginning but my mistake was I did not set unused pin as input and tide pins to ground, additionally I improve power supply for reduce ripple noise and more proper casing design.

Well so far 11 months I do not have any problem but who knows what it comes next

Theory for all of us engineers or beginners some time we have iron wall front of us due of undocumented informationís or poor documentations. Thatís the reason why we have to keep researching for tomorrow.

Regards

Joe