CPLD EPM240T Serious Burn down!!

[Altera_Forum]

Hi,I met a very bad problem when I use CPLD EPM240T. 

 (1) Environment : in this project,it have 10 IO board with EPM240T,all these board connected through one back board.Between these EPM240T,there is one sigle connection without any RESISTOR,this sigle connection is used EPM240T PIN-77,and all EPM240T PIN-77 are set as input.   (2)Problem:when insert these IO board into back board,EPM240T will burn down after power-on delay,not all EPM240T will burn down,but at least two nearby,sometimes 3 or 4,the rest board are work well.After burn down,there is short circuit in EPM240T between VCC and GND.It's confuse that this problem is not happen everytime,and cannot reappear.   (3)Question:There are at least two board burn down nearby shows that should be something cause this problem with connection between these EPM240T.All connected pin are set as input,is this a reason cause the burn down problem??Why this problem happen?Is there any one meet this problem?Could anyone help me?

[Altera_Forum]

Even a short circuit of one or more output pins, although violating the maximum ratings, isn't usually causing catastrophic failure of the device. Thus I assume, that a serious problem of violated supply or input voltage ratings is present in your design.

[Altera_Forum]

This problem is not usually come out,and I don't know how to reappear it,it work well almost time.

[Altera_Forum]

Are the system and operator properly grounded when the boards are inserted? There could be an ESD when the connectors touch. You can avoid it by having everything grounded before insertion, or ensuring that the boards ground touch before any other contact. 

Other than that I think too that there could be a problem with the voltage supply.

[Altera_Forum]

The system is grounded properly,and power supply seems Ok,no avoltage surge. 

 Today I found that there three pins connect to CPU pin which define as output,but not assign these pin as input or output in Quartus II.And I cut the connection between CPLD and CPU,and measure voltage: 3.3V at CPLD pin,2.0V at CPU pin.Someone says that the pin without assignment will be stuck to GND by Qaurtus II automatically.How Quartus deal with the pin without assignment??stuck to GND or stuck to VCC??Is this the reason cause CPLD damage for over current flowing from CPU to CPLD?

[Altera_Forum]

--- Quote Start ---  How Quartus deal with the pin without assignment?  --- Quote End ---   It's set in Quartus under device options/unused pins. I would always select input with weak pullup or hold circuit.

[Altera_Forum]

I check it in Quartus : Device--Pin Option--Unused pin : As input tri-stated with weak pull-up resistor. 

 So,It seems that pin without assignment is not the point.

[Altera_Forum]

--- Quote Start ---  Are the system and operator properly grounded when the boards are inserted? There could be an ESD when the connectors touch. You can avoid it by having everything grounded before insertion, or ensuring that the boards ground touch before any other contact. Other than that I think too that there could be a problem with the voltage supply.  --- Quote End ---   There should not be have ESD problem when insert the board,because the board only power on when delay 5 second after insert the board.All the board are in one metal case and case grounded well.

[Altera_Forum]

I'm just telling that because I ran into this problem a few years ago on an industrial equipment. The box was metallic and well grounded, but not the operator. If the operator has a high static charge, when he/she takes the board it will be charged too. When inserting the board, both the operator and the board will suddenly discharge into the ground through the box, using the path of least resistance. 

Did you check the power supply with a scope at power on, just to be sure that you don't have a spike at power on?

[Altera_Forum]

Thks Daixiwen,as you told,there is a surge at poweron,and the surge up to 6~7V,this is the reason caused CPLD burn down.So,we get the point and try to fix it.Do you have any good idear about dealing this problem??

[Altera_Forum]

--- Quote Start ---  there is a surge at poweron  --- Quote End ---   Overvoltage had been also my assumption. Basically, the bad performing power supply should be replaced by something reliable.   It's not easy to limit the supply voltage at the board level. A 3.3V TVS diode won't considerably reduce the overvoltage because it's characteristic is too soft. A thyristor crowbar circuit could act more effectively but needs supply disconnect, if trigerred. And it possibly causes more harm with a badly designed power distribution system.

[Altera_Forum]

So,the power system is very important,in this such of structure,hot-swap chip in every hot-swap board is necessary!

[Altera_Forum]

Hot-swap circuitry is only necessary when you actually do hot-swapping. From what I understood from the first post, the system is powered on after the boards have been inserted, so a hot-swap circuit won't help. 

As FvM said (and it was his assumption that was right, not mine ;) ) the only viable way is to fix the power supply. Any other correction such as a soft start on each board is just like a patch made with strings and duct tape. It may work on the short term but it's just asking for trouble on the long run.

[Altera_Forum]

How can I deal with the surge when insert the IO board to back plane if a hot swap controller can't fix it?

[Altera_Forum]

I'm taking as granted, that you don't intend hot swap, insertion of boards with powered backplane. Under this assumption, I completely agree with Daixiwen "the only viable way is to fix the power supply". The power supply must have efficient means to strictly avoid overvoltages that are at risk to damage the connected equipment. We don't know however the energy involved with the surge. If it's low, TVS diodes at the boards may help. A "V"LDO ("very" low drop out voltage regulator, some have drop out voltages of 50 to 100 mV) can be an option, otherwise a special designed supply disconnector with a FET switch.