Interference ?

I got some PCB made for a ESP-01 to WS281x signal shield, I tried to minimize the size and maximize the multi-purposeness. So they can be powered by 12v and 5v. and in my first assembly i had no signal at all. It took me a while to figure out that the TTL chip i used for providing the signal output was not receiving a high enough logic-high level (i used a 74LVC1G17) It does work normally but before i had been using either a hex-inverter twice or a non-inverter, and i suppose the schmitt-trigger is changing things a little, but after fiddling with the resistor values a little i have i got it to go. then after soldering the resistors onto the PCB it doesn't work anymore. Now my suspicion is that because i used a SOT223 regulator (for the 3.3v) and made most of the copperfill the heatsink (which is the vOut) the resistors and diode is getting interference from that. I'll include the schematic and the Gerber files. Maybe someone can have a look. I never experienced something like this before. Just to get it clear. if i tap from the PCB and have the resistors on the breadboard, and then tap back to the TTL-chip there is no issue. I never thought resistors would be that sensitive to interference.

ESP_WS2812.zip (41 KB)

Hello,
I think it still is about the input level to the schmitt-trigger. If you have 3.3 and there is voltage drop over R2 and D2 :slight_smile: it might drop to much. Datasheet says 0.5xVCC so…
Try to power it with 3.3V.

Good luck!

What is R1 for?

What is R2 for?

What is D2 for?

What is R3 for?

Please do not attach .zip files. Most people (here) are trained never to open a .zip file and it is never appropriate.

Paul__B:
What is R1 for?

What is R2 for?

What is D2 for?

What is R3 for?

all of them are to protect the ESP GPIO pin, under extreme circumstance they can burn out. Not often but often enough.

Paul__B:
Please do not attach .zip files. Most people (here) are trained never to open a .zip file and it is never appropriate.

The gerber files are not accepted by the forum, the zip file is. Feel free not to open it.

802Lundgren:
I think it still is about the input level to the schmitt-trigger.

Yes one does suspect such, though how to explain that having the same configuration at a few centimeters difference i have no issue ?

802Lundgren:
If you have 3.3 and there is voltage drop over R2 and D2 :slight_smile: it might drop to much. Datasheet says 0.5xVCC so....
Try to power it with 3.3V.

Yes i think the Schmitt-trigger is partly to blame as well (i ordered 74LVC1g14's to check and see if it helps) when i used these actual values i had no signal at all, just putting a 470R between the GPIO & the & 7417 it worked ok, while i had it sticking out a cm at the side, but once mounted on the PCB i had occasional flashes.
on the breadboard stable i have R1 = 1K, R2 = Jumper, R3 = 4.7K, D2 = 4148. and on the PCB it's a big mess.
Using 3.3v for the TTL defies the purpose of having it there, i need logic-HIGH output level above 4.5v to drive the WS281x

Well, if it works on the breadboard but not on the pcb it is more likely to be something wrong with your pcb, maybe a gnd is not connected? Think about occham it is more often just a simple fault than something extraordinary.

Wish i could help more.
Good luck

Think about occham

Or more correctly of "Ockham's razor". Ockham is a proper name, sometimes spelt Occam, and his razor is
the principle of ruling out the simpler possibilities before considering the more complex ones, although I'd
suggest fault-finding is more about localizing the problem than checking lots of possibilities.

Or more correctly of "Ockham's razor".

You are correct sir.

About finding the fault.
If you are troubleshooting in an analytical way, for example using the Kepner Tregoe method. One phase is to exclude a number of more or less probable reasons the fault occur, in order to find the fault itself.

But I mostly agree with you anyway.

Mr. Deva_Rishi,
please describe the faulty behaviour some more, do you have an oscilloscope? and please post a picture of your installation.

802Lundgren:
Mr. Deva_Rishi,

Rishi will do Thx. So to describe the faulty behavior i will separate into 3 parts
-first the original setup with the resistor values as they are in the schematic. No functionality at all. I therefore went for excluding possible causes,

  • put the ESP-01 back into another shield i made out of prototype-experiment board. Clearly not the problem (happy about that i hate burning them although i buy them for just over a Euro a pop)
    *started measuring the connections on the shield and if i had properly connected the SMD chip, bypassed the diode and got some functionality, figured the diode may cause to much of a voltage drop, so pulled it off the PCB as so with R3 (which i don't need if there is no diode) But still experienced a lot of 'static' Partly patterns moving but generally not the correct colors, very whitish and flashy, similar to strip that is being controlled through a long (thick) cable where capacitance causes timing issues.
  • Pulled off R1 & R2 and figured maybe the TTL chip is broken, or not powered correctly, but as a last test put a 470R to the side of the PCB connecting only the tips of the Resistor having it sticking out about 1.5cm and got good functionality no real disturbances as far as i could tell.
  • conclusion TTL-chip works, ESP-01 is ok, something with the voltage drop, i'll sort it out later, i should at least put R1 back in (The only possible way i have could think off burning the ESP-01 is when the GPIO is in output and somehow even momentarily 5v is applied. It should not be possible with a TTL input and maybe it hasn't happened but i did fry one in a similar setup once without R1 and never with R1) But then i got occasional (white) flashes, so i decided to just solder R2 (470R) and leave the rest out, and when i soldered it to the PCB (on it) the flashes returned.

-Secondly i decided to go for a different setup with a differen TTL-chip recreate the whole thing on Breadboard but use a 74LS02 (NOR-gate) to see if i could remove the 'green or pinkish' flash that is put out by the ESP-01 at startup in DMA mode by blocking the output with one of the other pins. I had 'static' at first but then changed R3 to 4.7K and R2 to 4070R and it all worked !!

  • Great, then i decided i was going to exclude possible power issues, so in soldered wires onto the PCB (removing R2 again) and used the regulators that are on it. Mainly i wanted to make sure the SOT-223 package was capable of supplying the power required by the ESP-01 and that the heatsink is sufficient. And it is. no issues

-Thirdly I figured how about i try to use the 74LVC1G17 again and added some more wires to the PCB, and fiddled with the resistor values until i had a working setup.

  • I decided on R1 = 1K, R2 = Bypassed, R3 = 4.7K and a diode. It was either remove the diode or bypass R2 both worked.
  • So then i figured well lat's solder that up on another PCB , but then the 'static' returned !

I do have a 'cheap-ass' oscilloscope that i got a few weeks ago but i haven't used it yet and i am not quite sure how to (should manage though)

Deva_Rishi:
all of them are to protect the ESP GPIO pin, under extreme circumstance they can burn out.

OK, so that is complete nonsense. Remove them all and the circuit is far more likely to work properly.

74LS02 is essentially an obsolete part. :astonished:

Hi,
Can you post images of your PCB please.

Thanks.. Tom.. :slight_smile:

Paul__B:
OK, so that is complete nonsense. Remove them all and the circuit is far more likely to work properly.

How much experience with ESP8266 do you actually have ?

Paul__B:
74LS02 is essentially an obsolete part. :astonished:

i do agree the HCT version is much preferred, still i had this one in my box and didn't have to walk out the door for it, not only was delivery instant but completely free. Now the 74HCT02 may be considered obsolete as well, but they are still out there and fully functional, and cheap also in relatively small amounts. Just like last years Nike (or the pair from 10 years ago, though that may become a collectors item) In the end a logic gate is a logic gate and if the specs meet the minimum requirements for the circuit there is nothing wrong with actually using it. Throwing IC's into the rubbish on the other hand is wasteful, and semiconductors can not be recycled.

TomGeorge:
Can you post images of your PCB please.

What image do you want ? assembled ? copper layers ?, the gerbers are in the zip, i could try to convert them if you like ?

Deva_Rishi:
How much experience with ESP8266 do you actually have ?

Perhaps more than yourself for all I know. :roll_eyes: How much experience with digital electronics do you have?

I asked you to explain what "protective" function these components might have, apparently you have no idea and replied accordingly. There is clearly no requirement whatsoever for any of them and they appear to be the cause of your problems.

(Edit: Actually, R3 at a value of 47k or 100k may prevent spurious flashing of the LED strip at startup.)

Deva_Rishi:
In the end a logic gate is a logic gate and if the specs meet the minimum requirements for the circuit there is nothing wrong with actually using it. Throwing IC's into the rubbish on the other hand is wasteful, and semiconductors can not be recycled.

Actually, the situation now is that the carbon footprint of continuing to use obsolete parts that draw more power that their successors is simply not justifiable. There is a penalty for using the older part.

I am reminded of the situation with Bitcoin mining where the cost of the power required exceeds the value of the mined currency.

Deva_Rishi:
What image do you want ? assembled ? copper layers ?, the gerbers are in the zip, i could try to convert them if you like ?

Hi,
Post images of the top layer with silkscreen overlay and the bottom layer thanks.
You should be able to export images in jpg.
What are you using as a powersupply.
Thanks.. Tom.. :slight_smile:

TomGeorge:
You should be able to export images in jpg.

Well that was easier said than done, but OK, in the end i took screenshots of the layers and then laid the silk over the top in Photoshop at 25% opacity, and then cropped, The resolution is maximum to what i can fit on screen, due to the orientation of the board, there's 2 versions both from the PCB designer and from the Gerberviewer.front.jpg back.jpg


On the PCB designer you can more clearly see the vias, while in the gerber viewer the copper is a bit more clear.

TomGeorge:
What are you using as a powersupply.

I'm using a big-ass 300W 12v switching powersupply that is not waterproof and therefore ended up in my house for this purpose. I did have a 5v 30W powersupply on it at some point (i was powering 5v LED-strip with that and changed the jumper over) to exclude that as an issue.

back.jpg

front.jpg