2
\$\begingroup\$

I have a circuit with two supplies, a 19 V supply for stepper motors, and a 5 V supply for driving the ATMEGA644, 3 stepper motor driver daughter boards, a Bluetooth LE daughter board, and two WS2812B RGB LEDs.

With two supplies everything worked, but I really wanted to get it down to just one supply. So I did another version of the board using just the 19 V supply and a VX78-500 switching regulator. I used their recommended input and output caps (but not the low noise version that include some extra inductors).

I have a 100 μF electrolytic on the 5 V rail. I'm not 100% sure if that's a good idea, but it's well below the max. capacitive load for the VX78-500 which is 680 μF and I thought it might help with noise.

I also have three 100 μF electrolytic on the 19 V line next to the stepper driver boards for the same reason.

When I first turned things on I noticed the WS2812b'd flashed slightly but everything was fine. That didn't happen on the two-supply version of the board. However, on maybe the 3rd or 4th time I turned it on the Bluetooth LE daughter board died, which is kind of surprising because it's got its own on-board 3.3 V LDO regulator.

I don't know that much about supply best practices/inrush current, etc. I knew my new setup would be noisier using a switching supply/being down stream from the same supply as the steppers. Any ideas what is going on? That flashing makes me think there's a bad spike happening at switch on time. It seems weird that the ATmega644 is still ticking but the BTLE daughter board died. How do I trouble shoot this?

Are those 100 μF caps just a bad idea in general? Does my circuit need some other form of protection?

Those daughter boards are $19 so I don't want to blow out too many of them in a search for the solution. Here is a pic of the PCB. I'm sure I'm just making some noob mistake, but I'd really like this one-supply version to work.

PCB With One Supply

Ignore where that says 12 V, it's really 19 V. blush

Here are the two sheets of the schematic as well:

The first sheet of schematic

Second sheet of schematic

\$\endgroup\$
11
  • 4
    \$\begingroup\$ Welcome to EE.SE! Please add the schematic, as this is more helpful than the PCB screenshot. \$\endgroup\$
    – Daniel
    Commented Mar 4, 2019 at 17:53
  • \$\begingroup\$ You PCB picture isn't very helpful here. As Daniel said, please, provide the schematic. 100 uF capacitor is ok, nothing wrong, I would add more closer to the logic ICs. Also did you use decoupling capacitors (about 0.1 uF)? What is the powers consumption on 5V line? Don't you think tat 500 mA from DC-DC converter could be not enough for such kind of schematics (only RGB leds need at least 50 mA each; may be even more)? What is you 19V supply output power? Is is possible, that it has bad ripples itself, which you didn't see, when used second power supply? \$\endgroup\$
    – cyclone125
    Commented Mar 4, 2019 at 18:51
  • \$\begingroup\$ "bluetooth LE daughter board" died - What is this card exactly? What is its power requirement, especially max input voltage? Did you check transient voltages on all rails (+5V most suspect) upon power-on process? \$\endgroup\$ Commented Mar 4, 2019 at 20:09
  • \$\begingroup\$ It's this adafruit bluetooth LE board: learn.adafruit.com/… \$\endgroup\$ Commented Mar 4, 2019 at 20:36
  • \$\begingroup\$ Which seems to be using the MIC5225 as a regulator that says it's ok up to 16v. Of course something else may have been killed like the level shifting chip, etc. \$\endgroup\$ Commented Mar 4, 2019 at 20:43

1 Answer 1

1
\$\begingroup\$

That flashing makes me think there's a bad spike happening at switch on time.

I think you are on right track. The problem is likely in start-up transients.

From the PCB layout: it is customary to place a de-coupling capacitor in close proximity to every Vin entry point to any sub-block, be it a IC or an external module. There is no cap at the entry to NRF8001, the module which has died.

I also tried to imagine the shape of bottom ground plane (I assume it is there). It looks like it is badly traversed by many signal traces, so the actual ground integrity looks very questionable.

So I am afraid this board has strong transients that killed your NRF module. Try to put a 10-uF ceramic 0805 capacitor between Vin and GND pins on the NRF connector, and maybe put some 5-V voltage suppressor.

\$\endgroup\$
1
  • \$\begingroup\$ I guess it's true I have caps by all the other daughter boards, but not that one. I always do that for chips, but I kind of assumed that since boards like that have their own local caps it would be ok. Good to know I shouldn't assume that. So you think the spike is on the 5v rail? What sort of 5v suppressor. Like a zener on the 5v rail? Sorry I'm just clue less about power rail protection stuff. Do I put that near the regulator, or down stream near the components? I guess I'll have to read up on those. Thanks! \$\endgroup\$ Commented Mar 4, 2019 at 21:53

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.