3
\$\begingroup\$

I am using the bidirectional MAX3377 level translator by maxim. I am doing a prototype board and as I test this board piece by piece, everything works until I solder on this part, and my 3v3 rated microcontroller begins to fail.

Level Shifter

PCB layout

PCB layout

The low and high side of the boards are appropriately isolated, yet some damage is caused to the chip as soon as i solder on these two level shifters. This has occurred on 3 of my prototype boards already

Could anyone explain why this could be happening? Is it something to do with the 3 state pin? I only have one prototype PCB left, help would be greatly appreciated.

Thanks.

\$\endgroup\$
13
  • \$\begingroup\$ Vusb? You have a power rail tied to a pin? It's not a voltage regulator. Did you check for shorts? \$\endgroup\$
    – Passerby
    Sep 14, 2016 at 2:33
  • \$\begingroup\$ Could VUSB ever exceed the +5V supply (e.g. when USB is first attached and the +5V supply is not yet fully on)? \$\endgroup\$
    – MarkU
    Sep 14, 2016 at 2:35
  • \$\begingroup\$ The THREE-STATE pin is referenced to the VL logic supply, so connecting it directly to 3V3 supply (VL) should be OK. \$\endgroup\$
    – MarkU
    Sep 14, 2016 at 2:38
  • \$\begingroup\$ VUSB does not power the board, its just there to tell the microcontroller that it is plugged in. Also, no USB device was plugged in at any time as I have not reached that part of testing the board yet. \$\endgroup\$
    – Sam Bucca
    Sep 14, 2016 at 2:46
  • 3
    \$\begingroup\$ Not an answer, but some advice: now that you've isolated the problem, try to prototype with just the level translators (and power supplies) populated. You can simply touch wires to the inputs and read the outputs with a multimeter. This should help you figure out the problem. Since you only have one PCB remaining you may wish to remove everything from a failed board to do your testing. \$\endgroup\$
    – bitsmack
    Sep 14, 2016 at 3:38

1 Answer 1

1
\$\begingroup\$

If I understood your question and further comments to it properly, something in ATMega fries and then MCU pins, connected to the MAX3377, stop working.

It is logical to suppose that it happens due to some physical event affecting MCU pins. Most probably that MCU, at some point in time, has pin configured as output, and MAX3377 also tries to output some voltage, and MAX having better protection on the over-current than MCU. For example, MCU outputs logical 0, but MAX, for some reason thinking that it should translate from other side of its row, puts +3V3 onto the same line. Or vice versa. I can only guess, given that you soldered everything properly and there're no shortages.

Side note on the top drawing of MAX3377. Please try to avoid mirrored drawing of power supply symbol (you have 3V3 mirrored down) so that it looks like ground sign.

Conclusion: look into MCU's datasheet to see which I/O state its pins have on power on (the moment when level conflict may happen), and the sequence of port initialization to ensure that the first thing microcode does is to assign input state to input pins and output to output pins.

But in general, as The Photon advised, stop using bidirectional translators which use bias for its decision making; use unidirectionals or controlled bidirectionals (e.g. 74LVC_256 family)

Just to add: in my opinion Maxim's sheet for this device is lacking some vital information, looking more like marketing material, and I personally would not use this device in my designs, at least without close support from Maxim.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.