Thermal Impact on Threshold Voltage

[Pack34]

I'm having an odd issue and I was hoping someone here might be able to help me out. I believe I'm having some thermal issues and I was hoping to get some sort of confirmation here before I start carving up my board.

I'm working with an ENC28J60 Ethernet Controller. It's a basic SPI Ethernet controller by Microchip. Now, I'm getting some repeated results in which when the board heats up after a while I'm loosing my UDP connection. After I let everything cool down overnight, everything works without a hitch again. I skirted some requirements a bit due to success I was having on my dev board. It's a 3v3 device, but the input pins are 5V tolerant. Having it plugged straight to my PIC's SPI bus allowed everything to go off without a hitch.

I'm thinking that it's the connection between the ENC28J60 and my microcontroller (PIC18F4680) because the Ethernet chip seems to be working fine. I'm getting the correct status LEDs on my RJ45 connector and the PIC seems to be going on it's way normally, it just isn't communicating with the Ethernet chip anymore.

The PIC handles the 3v3 signal from the ENC fine under normal conditions, but it seems like I will need to hack in a level shifter to get it working in all conditions. Does this sound plausible to anyone else, or do you guys think I should be looking elsewhere?

ENC28J60
http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en022889

[chickenHeadKnob]

With high speed digital typically heat effects timing first, that is some part of your circuit which is fast-clocked is no longer meeting spec.

[mazurov]

You said you were losing UDP. Are you loosing anything else, like is ICMP still functional (can you ping it ?). The controller works at datalink layer, it would take everything above with it if it died.

[Pack34]

You said you were losing UDP. Are you loosing anything else, like is ICMP still functional (can you ping it ?). The controller works at datalink layer, it would take everything above with it if it died.

The initial UDP loss is typically is due to a network failure. The network is wireless and the radios can move too far away from one another. This would cause a data connection failure. I'm trying to get everything to fail safely and then reconnect. I know that the transmitter is not doing it's job because the data lights on the receiver (my board) are no longer going off until I reboot the transmitter.

Once I've finally been able to run some tests again, I'm getting consistent results. Measuring the voltage lines, I'm losing the 5V and 3.3V rails, so it looks like LDOs just aren't handling the heat of the enclosure. I'm going to wire in a temporary switching regulator module to see if the issue goes away. I really wanted a switching regulator on this guy anyway. I'm dropping 12V down to 5V on a battery.

I'm going to try this first before I see if doing a proper level shifting is necessary. Since it looks like I'll need to do a revision on this board, I'm going to add one in on the next spin anyway.