Can Voltage Regulators partially fail?

HI- I’ve got an AMS1117-3.3 on a home-made circuit board, and things were just ducky with my little ESP8266 project, but then I tried to flash some code, same way that has worked in the past, and I got the timeout error. I rebooted my device (the 3.3vreg is fed from a 5vdc wall supply), and didn’t get any flash from the ESP8266 LED. Figuring I bricked it, I began to explore the PCB, and all checks are good- resistance is where it belongs on my pull-ups and -downs, no shorts, only thing anomalous is that my Vreg is putting out only 1.82V. Again, I suspected a short, so I DMM’d around again, and no shorts. Leads me to suspect the VReg, but I don’t have enough experience to know if they can fail this way (I assumed it would be 0V (open) or 5v (closed)).
I’m going to desolder the component tonight, and will try to power the board from an external supply to see if I can get it working before I replace…but all I have are VRegs from the same batch from China ($0.80 for 10 or something ridiculous). I know, I know, get what you pay for!

Here’s my schematic and board:

DHT11_v1.2_BRD.PNG

It can also oscillate which makes it appear to have a lower DC voltage. Do you have a scope?

I do...sorta...Hantek BE60xx USB scope, but will give it a try! And now that you mention it, I think I'd heard that before, but couldn't pull it out of my memory banks. Thanks! I'll measure the output on the scope to see where I am. If it is oscillating, I should look at my bypass caps of the 1117? To small...or too big?

Hi,
Try putting 0.1uF caps in parallel with C1 and C2.
Also your gnd tracks need to be thicker, especially around the reg IC and the power input socket and the track to the pins that feed the sensor. You appear to have the space to do this.

When you design a PCB remember you are paying to have it made, not remove as much copper as possible.

5e4af9c7d2cdf83913c5aff252e6f517000582d9.png
467024d0e20f79725e127911d5a14065692e099b.png

Tom… :slight_smile:

Wrong ESR is the usual cause for low-drop-out regulator oscillation.

To quote datasheet:

Input Capacitors
To ensure stable operation, the input supply must be low impedance up to a frequency of a few MHz. This requires a closely placed input decoupling capacitor of 4.7μF minimum. This can be either ceramic or solid tantalum. The ESR of this capacitor must be less than 0.5Ω.
Output Capacitor
The AP1117 also requires a closely placed output capacitor as part of the device frequency compensation. As part of its improved performance over industry standard 1117, the AP1117 is suitable for use with MLCC (Multi-Layer Ceramic Chip) capacitors. A minimum output capacitor of 4.7μF ceramic X7R or 4.7μF solid tantalum is required. Aluminum electrolytic can be used but a minimum of 47μF is required. When using Aluminum electrolytic it is still recommended to also use a 1μF MLCC in parallel. The ESR of the output capacitors must be less than 0.

Looking at the PCB, power traces between regulator and capacitors should be low inductance (short, wide), not thin/long. Same for between a chip and its decoupling capacitors. Wide traces for power and ground, nothing to do with the resistance unless high current (amps), everything to do with the inductance.

The ESR of the output capacitors must be less than 0.

Wow! They are looking for some expensive components there!

TomGeorge:
Hi,
Try putting 0.1uF caps in parallel with C1 and C2.
Also your gnd tracks need to be thicker, especially around the reg IC and the power input socket and the track to the pins that feed the sensor. You appear to have the space to do this.

When you design a PCB remember you are paying to have it made, not remove as much copper as possible.

5e4af9c7d2cdf83913c5aff252e6f517000582d9.png
467024d0e20f79725e127911d5a14065692e099b.png

Tom… :slight_smile:

Thanks for the advice! I’m actually fabbing this myself, have gone through a couple iterations with photosensitized copper, and I am quite happy with the output! I originally did have a ground plane in on the back, but took it out because I thought (maybe incorrectly?) that it might interfere with the antenna?

MarkT:
Wrong ESR is the usual cause for low-drop-out regulator oscillation.

To quote datasheet:
Looking at the PCB, power traces between regulator and capacitors should be low inductance (short, wide), not thin/long. Same for between a chip and its decoupling capacitors. Wide traces for power and ground, nothing to do with the resistance unless high current (amps), everything to do with the inductance.

I’m probably going to get booted for saying this, but I’m using the autoroute function on Eagle, and am still figuring out classes! I’m sure I can manually re-route those though! I know it’s the wrong way to go but I’m just a biologist…

Thank you both!

I don't think you want any traces or ground plane under the antenna trace.
I would rip up any traces near the antenna and route them differently.

Alright, here's what I got off my scope:

This isn't what I was expecting to see if my reg were oscillating (I pictured a sine wave with the average being 1.82v, since that's what my meter gave me), but we all know I'm new to this...mean anything?
FWIW, I also looked at the incoming supply, and it's steady-steady, couldn't find a ripple anywhere.

Did something wrong and my attatchment got stripped, then I got locked out for 5 min…sorry. Here is the scope view (I hope).

CrossRoads:
I don't think you want any traces or ground plane under the antenna trace.
I would rip up any traces near the antenna and route them differently.

That's great to know, I'll do that on the next version. Can I ask though, if I did do a ground plane on the bottom, just left a square where the antenna is, would that work, or is it best to keep ground planes away from antennas in general...so if a design has one, just do without? I'm sure there's so much more to this than meets the eye (I briefly considered getting a HAM Technician license, but have to learn a lot more).

if I did do a ground plane on the bottom, just left a square where the antenna is, would that work,

Yes.

Just giving this a little bump...got a lot of early advice but could really use some help with the scope trace...am I looking at a bad V Reg, or is this one good just doing some oscillation that I'm not able to pick up?

Hi,
I think that is the TX pulses getting into the regulator because of the proximity of the aerial.

Move your reg down closer to the U1 power pins and away from the aerial, beef the tracks up and include the 0.1uF caps will help immensely.

The aerial is on the edge of the sub board, so move the sub board assembly to keep the aerial on the edge of your board.

Tom... :slight_smile:

TomGeorge:
Hi,
I think that is the TX pulses getting into the regulator because of the proximity of the aerial.

Move your reg down closer to the U1 power pins and away from the aerial, beef the tracks up and include the 0.1uF caps will help immensely.

The aerial is on the edge of the sub board, so move the sub board assembly to keep the aerial on the edge of your board.

Tom... :slight_smile:

I can do that, but I"m wondering: A) If the reg is only putting out 1.82v, is the 8266 even transmitting? B) would the proximity to the antenna pull down the voltage of the reg to only 1.82?
The whole thing worked fine for several days running on it's own, and only went south when I tried to reflash some code...

Well, I replaced the regulator, and everything seems to be fine. Will keep an eye on it, and have already reworked my board file so I can take the recommendations here into count if it happens again. Thanks for the advice so far!

MarkT:
Wrong ESR is the usual cause for low-drop-out regulator oscillation.

To quote datasheet: "The ESR of this capacitor must be less than 0.5Ω."

MarkT:
Looking at the PCB, power traces between regulator and capacitors should be low inductance (short, wide), not thin/long. Same for between a chip and its decoupling capacitors. Wide traces for power and ground, nothing to do with the resistance unless high current (amps), everything to do with the inductance.

The required low 0.5 Ohm equiv series resistance for the capacitor is to minimise series RC transient effects is it?

And minimising inductance is to try avoid LC resonance oscillations? Or at least in the region where we don't want oscillations? I thought I'd ask while the subject is going on here. Thanks!