Ok, I'm completely puzzled by what is happening to my LM317 power regulator adjust pin voltage. Here is the schematic of the power supply section of my PCB that distribute the 3.3v (VDD) to the microcontroller while other components use the 5.0v (VCC) directly from the connector (J14 and J13 are test pins on the PCB).

enter image description here

So far so good. I get 3.3v distributed to my microcontroller no problem and 5v to my leds and LCD.

Trouble arise when I connect a MAX 485 chip to VDD or VCC:

  • When connected to VDD point, the LM317 outputs about 1.6v and gets super hot.
  • When connected to VCC point, the LM317 outputs about 3.8v not getting hot but not either returning the proper 3.3v.

Here is the part of the schematic that handles the MAX485 chip:

enter image description here

The 10k pull up resistor aren't soldered so I crossed them out. As soon as the pins 5-6-7-8 are soldered in place, the LM317 gets out of wack according to my previous stated observation. Please note that the DMX in and out plug aren't connected at this point.

What I think I understand is that the MAX chip circuit introduces new resistance path on the adjust pin of the LM317 through VCC and GND but I that's just a gut feeling. Anyone has any idea on why this MAX chip seems to affect my power supply section?

  • \$\begingroup\$ Which package of MAX485 are you using? \$\endgroup\$ – Ale..chenski Sep 10 '17 at 5:39
  • 2
    \$\begingroup\$ Sounds like the way you are connecting up the MAX485 leads to serious loading (bad part, incorrect wiring, etc.) If you load the 3.3 V, it just pulls down extremely hard and the LM317 gets hot for obvious reasons. If you load the 5.0 V, it droops as well and that fact removes the margin that the LM317 requires. Why don't you measure current into pin 5 of the MAX485? Might be a bad part? Or some other problem related to its surrounding circuit that you don't otherwise recognize right now. What's Vcc when you load it directly? \$\endgroup\$ – jonk Sep 10 '17 at 5:40
  • \$\begingroup\$ I use soic8 package. I know it is not a bad part since I get the same result on 2 different pcbs. \$\endgroup\$ – Kévin Isabelle Sep 10 '17 at 5:44
  • \$\begingroup\$ Vcc is 5v mesured both when the max chip is connected or not. Vdd though will get to 1.6v or 3.8v depending if max receives its power from vdd or vcc. Which is 3.3v when the max chip is not connected. \$\endgroup\$ – Kévin Isabelle Sep 10 '17 at 5:48

According to MAX485 datasheet, for DIP/SOP packages the ground goes to Pin5, and +5Vcc goes to Pin8. You seem to have the chip powered upside-down.

| improve this answer | |
  • \$\begingroup\$ Gosh I think you are right. My kicad footprint seems to be wrong! I'll verify this tomorrow with my PCB and give updates. Thanks pal! \$\endgroup\$ – Kévin Isabelle Sep 10 '17 at 5:53

Your design with the LM317 is a bit stretched for the application of lowering 5V to 3.3V. In other words it could be unreliable.

In fact, according to ON semiconductor LM317 datasheet the minimum current needed for regulation is 3.5mA (and R6 provides that), but that's typical. The worst case value is 10mA. Unless your circuit always draws at least 6.5mA and/or your LM317 is hand selected, you might lose regulation, perhaps when some other parameter changes (e.g. temperature).

enter image description here

Another limit of the LM317 is its needed dropout voltage. Your design assumes that Vin-Vout=1.7V will be always sufficient to maintain regulation, but this is not the case. Look at fig.10 in that datasheet. The LM317 could need as much as 2.5V dropout voltage to ensure correct regulation under all conditions.

enter image description here

I hope you took all this into due consideration, otherwise your design could have reliability problems once in the field (if it is a one-off project, with hand selected parts, it could be OK, though).

| improve this answer | |
  • \$\begingroup\$ Awesome thanks for the info! The lm317 only drives an xmega64a3u microcontroller and the max chip. 5v from the power supply is used for the leds, LCD screen and rotary encoder. Is this separation of power distribution enough to garantee reliability? \$\endgroup\$ – Kévin Isabelle Sep 10 '17 at 15:41
  • \$\begingroup\$ As you may see from the graph of the I/O diff voltage, from that POV you could be safe if the junction temperature is above 25°C (fair assumption if your device is used indoors) AND you don't draw more than 200mA from it. \$\endgroup\$ – Lorenzo Donati -- Codidact.com Sep 10 '17 at 19:09
  • 1
    \$\begingroup\$ This still won't guard you against the first point I made, i.e. the min current needed for regulation. If there are times when the load is disconnected or draws less than 6.5mA AND you have a specimen that needs at least 10mA of output current to regulate properly, then the output could rise above 5V. From this point of view your design is flawed, unless you can guarantee that min 6.5mA of current draw from the load (3.5mA are taken care of by R6 already). \$\endgroup\$ – Lorenzo Donati -- Codidact.com Sep 10 '17 at 19:09
  • \$\begingroup\$ I'll confirm that from the microcontroller datasheet but I doubt that it will draw less than 6.5ma since I run it at full 32mhz speed all the time and doesnt use sleep modes. \$\endgroup\$ – Kévin Isabelle Sep 10 '17 at 22:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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