Netburner 5234 for OPtical Keyboard - Keep Killing Netburner

[mpalladino]

Hello All,

We have designed an optical keyboard which is transmitting keys over fiber to a host computer which is able to view the packets and decode the key which is down/struck.(It's military...not sure why they want fiber..they won't tell us) The netburner MOD5234-200IR is the P/N we are using. We mount the netburner onto our PCB which serves as a carrier board. It has some logic to enable the netburner and disable it depending on input signals into the core processor. It's mainly a 5V system on the netburner carrier board. I'm using LM1117 to regulate 3.3V for the netburner to be powered up correctly.


PROBLEM:

We load the F.W we wrote into the netburner module and everything works great. Keys come flying out over fiber and we have also written a program which sees these packets and displays the key which was down onto the screen. After using it for a while(a day of two) and then the netburner stops sending keys. We have reloaded another netburner which works for a while and then dies too. So, after burning up two netburners and our wallets are hurting I'm thinking we are doing something bad to the netburner which is causing it to fail. We're not sure why. Can anyone shed some light as to the possible things I could be doing to the netburner which would cause this?

Below is the carrier board schematic. The 3 modules in the top left of the schematic are the netburner PCB.

Thanks!!

-mike

[rnixon]

Are there any pins connected to the netbuner that have more than 3.3V on them? Any 5V logic or other IC's?

[Ridgeglider]

I can't really make out the schematic from the jpg, but it looks like you have a few 5V powered comparators that drive inputs to the 5234. The comparators outputs are pulled up to 3V, but it seems like the outputs might easily exceed 3V and go up close to the 5V rail when driven hi by the comparator...?

[mpalladino]

Ridgeglider/rnixon,

First thanks for your replies!

There are no pins that are directly connected to 5V. That said we do have comparators(and other circuitry on the PCB) which are powered by +5V and are pulled up to 3.3V.(that seems a little silly in hindsight!) Sorry for the crummy schematic. I'm going to try to upload a better jpg. I'm going to get into the lab and see if the comparators are shooting up to +5V then settling down to +3.3V. I'll let you know results. Also, it appears that if the comparitors are driving +5V even for a short time we may be blowing the netburner. It might be a stretch but the netburners don't initially die. I think I mentioned it above but they run for about a day or so then die. Perhaps the stress of +5V blips are slowly killing the netburner. hmmmm

I'll post back this AM

-mike

[mpalladino]

Update,

You guys are correct! We do indeed spike up to +5V for about 3mS. I need to look into the datasheet and/or call netburner to see if this is illegal or not. Doesn't seem like you would want to do this to a 3.3V part!

Thanks

-mike

[Ridgeglider]

No need to call... 5V on a pin w/ 3.3V max is ALWAYS "illegal"!

[rnixon]

I second RidgeGliders comment. That is a problem, and it is a distinct possibility that is what is killing the unit. The processor needs to do something with the extra power it is seeing due to the higher voltage. I don't know exactly what the silicon looks like, but there will be current generated by the 5V to 3.3V differential, which gets shunted by the silicon in the processor, which gets hot, degrades the part, and it eventually fails. Maybe one pulse isn't bad, but when the circuit becomes very active it just too high a voltage for too long. Just my guess. If you wanted to test that theory, and can cut into the traces in the board, you could try an experiment and put in series resistors between your logic and the processor. I would not ship a product this way, but it could be a way to test the theory before a pcb modification/relayout. I'm not sure what value, maybe 120 ohms?

[Chris Ruff]

A reasonable way to do this is to put a 220 ohm (or so) resistor in series and clamp the processor side of the resistor with a schottky diode to 3.3Volts. The processor board can handle the 3.3 volts + the diode drop. Any and all lines that can deliver more than 3.3 volts need to be clamped to 3.3 volts. (connect the anode to the resistor)

It is not necessary to go haywire with level-shifting chips when the timing constraints are not being violated with a resistor/diode .

Chris

[jsampson]

Also check the data sheet for your comparator. Some parts are open collector (or open drain) output. If you are seeing 5V at the output then I assume this particular part is not open collector. So possibly picking a different pin compatible part will solve the problem. (I can't read the part number on your schematic.)

Also, is this keyboard being used in a high static environment?