Spartan 3A温度依赖性可重复失效该怎么办？

我有一个基于Spartan 3A的设计。
我的客户退回了一块板（600多块产品中有1块），当温度达到45摄氏度左右时，它会失效。温度非常具体，44度工作，46度失效，这是完全可重复的
。
在成品中，电路板强制冷却通常不会超过35度，因此其他电路板可能具有类似的不稳定性。
FPGA旁边有一个温度传感IC，因此我有一个相当好的温度指示。
当我对FPGA设计进行非常小的改动时，不会影响设计的操作部分并加载该熔丝图，那么温度问题就会发生变化。
我只是通过输出缓冲器将逻辑高电平门控到未使用的引脚，现在器件在65摄氏度左右失效，再次非常可重复。
我不确定FPGA中是否存在故障节点，或者我的FPGA设计是否存在问题，导致延迟导致不稳定。
我的想法是，当我改变温度以消除设备时，我能以某种方式以某种方式测试Spartan 3A节点上的所有节点吗？
客户认为他可能有另一块板在54摄氏度左右失效。
任何输入赞赏。
问候
戴夫



以下为原文

I have a design based on Spartan 3A. I have had a board returned (1 out of over 600 produced) by my customer which fails when the temperature reaches around 45 degrees C. The temperature is very specific, at 44 degrees it works and at 46 degrees it fails, this is completely repeatable.
In the finished product the board is force cold usually would never get above 35 degrees so other boards may have similar instability.
There is a temperature sensing IC alongside the FPGA so I have a fairly good indication of temperature.
When I made a very small change to the FPGA design, something that does not affect the operational part of the design and load this fusemap then the temperature problem changes. I simply gated a logic high through an output buffer to an unused pin, now the device fails around 65 degrees C, again very repeatable.
I am not sure whether there is a faulty node within the FPGA or if I have a problem with my FPGA design with delays causing instability.
My thought is can I somehow test all on the nodes of the Spartan 3A in some way while I change the temperature to eliminate the device?
Customer thinks he may have another board which fails at around 54 degrees C.
Any input appreciated.
regards
Dave