FEDEVEL
Platform forum

GuyMcTavish , 01-29-2025, 01:40 AM

**Intro:**
First off, thank you Robert for your YouTube courses. I followed this course to do this: https://www.youtube.com/watch?v=_x2fzKEjUGQ. I'm a software engineer who has dabbled in Arduino-based hardware for quite a while. Well, this time I needed something compact, and purpose-built, so convinced myself I needed to learn PCB design.

**Purpose**
The end goal is to create a prototype for a more sophisticated board that will fly in high-powered rockets to track flights, manage communication, and someday, do active control.

**My problem**
I designed a PCB using EasyEDA and had it manufactured by JLCPCB. I had them mount most of the components except for two. However, after manufacture, straight from the factory, I have a dead-short to ground on my 3V3 plane. My 5V and 3V3LA (low amps) planes are fine. I've gone over it for hours, I don't see any issues with the copper planes touching what they shouldn't in the design. The components seem to be the correct ones in the correct places. I'm hesitant to blame manufacturing since it's my first board, and it's the same issue on all 5 boards straight out of the box.

How do you know you have a dead short? Well, my resettable fuse got hot enough to toast my finger. I thought maybe I undersized it and jumped it with some copper wire. Magic smoke escaped from my diode that controls battery vs USB power input. Also, when using the continuity functionality I can touch a ground pad to any 3V3 line and it will have continuity with very low resistance. I also confirmed continuity on all boards, not just the one I smoked 😛 .

My 3V3LA power plane and my 5V power plane do not have this behavior, and they are working as expected. However, the important power plane is the 3V3 one.


**Screenshots**


**Summary**
I doubt the screenshots are enough, I can share Gerber files or any other files you all think are best to share. There's nothing propriety going on here, I'm just trying to learn where I went wrong.

GuyMcTavish , 01-29-2025, 01:47 AM

4 layer PCB, data paths are on the bottom layer, didn't include them in screenshots since I didn't think it was too relevant. Data lines don't seem to have issues when I probed their paths. Not sure how much can be debugged here with what I provided anyway, but hoping someone can let me know what more I should share or is willing to have a screen share, and I can get you an AMZ gift card or something as a thank you.

FuzzyOtter , 01-29-2025, 01:50 AM

Post a clear, high resolution photo of the front and rear of your board

FuzzyOtter , 01-29-2025, 01:56 AM

What is your setback between your power plane and what looks like a plated through-hole that looks connected to the green "ground" plane? It looks painfully small.

QDrives , 01-29-2025, 02:31 AM

Are you sure this is ok?
TI datasheet does not agree - https://www.ti.com/lit/ds/symlink/lm1117.pdf

GuyMcTavish , 01-29-2025, 07:15 AM

Here is are some photos. I honestly have no idea what the setback is. I did wonder if those had anything to do with it. But if they did I would imagine the issues would be present on the 5v plain and 3v3la plain, but those appear to be fine.

GuyMcTavish , 01-29-2025, 07:15 AM

GuyMcTavish , 01-29-2025, 07:16 AM

Certainly a great question and reasonable observation. I'll confirm.

GuyMcTavish , 01-29-2025, 07:40 AM

I think you might win the brownie points here. So I was trying to read the docs for the specific copy-cat part I got, which is a different manufacturer, but I imagine it's the exact same part: https://www.lcsc.com/product-detail/Voltage-Regulators-Linear-Low-Drop-Out-LDO-Regulators_HGSEMI-LM1117MP-3-3-TR_C498153.html. The TI one Robert used in his vid wasn't in stock at the time. The one I chose has documentation that is as clear as mud to me. That said, I'm still not sure why I decided it was best to wire it that way. The TI documentation you linked and Robert F. used is quite clear.

TIL: I should probably try to find the documentation from a different manufacturer that has better documentation like TI, even if I'm using a copy-cat part.. Which seems to be what I'm doing as far as I can tell.

It's late now, basically midnight, but tomorrow I'll see what can be done to jump/rewire/salvage this so I can test the rest of the board.

GuyMcTavish , 01-29-2025, 07:46 AM

Curiousity, got the better of me, it seems I need to unsolder the 'tab' portion of the LDO, bend the LDO up a bit, and then jump that tab back to VOUT. I'll try that tomorrow, probably around noon and report back.

GuyMcTavish , 01-29-2025, 07:46 AM

GuyMcTavish , 01-29-2025, 07:52 AM

I think my image is wrong though..... Vout should be the middle one. I'll confirm tomorrow

QDrives , 01-29-2025, 05:08 PM

I do not think it would be possible to bend the tab without damage. Besides, it is much easier to cut the two traces to the side of the tab that connect to Gnd.

QDrives , 01-29-2025, 05:11 PM

As for your datasheet -- it is common, if not always the case, that the center pin of SOT-223, DPAK (TO-252), D2PAK (TO-263), TO-220 and TO-247 is connected to the "TAB".
Since there is no mention what the tab is, it may be assumed that it is the same as pin 2 -- Vout.

GuyMcTavish , 01-29-2025, 05:27 PM

Good call, probably easier

GuyMcTavish , 01-29-2025, 05:27 PM

I'll try to commit that to memory, I appreciate the time and insight.

GuyMcTavish , 01-29-2025, 05:59 PM

It looks like that was the issue. I cut those traces/tracks, no more short, and I read the right voltages in the right areas. Fuse no longer overheated as well.

GuyMcTavish , 01-29-2025, 06:01 PM

I'll need to clean up the margins on the cut tracks and jump the tab over later (not sure if there is any issue with leaving it unconnected). Then try out it out with the microprocessor and other sensor added.

QDrives , 01-29-2025, 09:25 PM

There is no need to connect the tab. It is hard connected internally to pin 2 and that is your 3.3V.

GuyMcTavish , 01-29-2025, 09:47 PM

Yeah tried it out and it seemed to work. Much appreciated. Now we'll see what the next issues are I run into ha.