Problems with ESP8266, IRF510N and circuitboards

Hey thanks for the reply. Yes it is, I have a 12v PSU from which I power 
the strip and the ESP. The ESP is connected via Vin and GND to GND of 
the PSU. So it shares the common ground.

40/60 wrote:
> where are the current-limting resistors for the LEDs?

Oh I'm sorry, I totally forgot it in my drawing. The resistors are 
located on the LED-Strip itself. Each LED hast its individual one.

stefanus wrote:
> What does the oscilloscope show? What are the measurement points?
>
> Are your GND lines connected to star centre as shown in the image?
>
> The image comes from https://www.ranecommercial.com/legacy/note151.html
> . This article is worth to read.

I connected all the GND lines to the GND Terminal of the PSU.

The Probes of the oscilloscope are connected to the gate of the Mosfets. 
Channel one (the yellow one) is connected to the setup on my breadboard.

Channel two ( the blue one) is my soldered setup. The Probe is also 
connected to the Gate pin.

For simplicity reasons, I only posted one picture, where I connected the 
Osci probes to the Mosfet for the Red colour on my soldered board and on 
the breadboard. I probed all of the Mosfets, to make sure its not a 
single dead one, and all results look the same. Like I mentioned in my 
original post, the background noise is about 1V higher on my soldered 
board than on the breadboard.

Both I provide a LOW signal ob both of the Mosfets in the Oscilloscope 
picture. So they should be off at this moment.

argos wrote:
> Is the switching noise at the gates still present when you turn of
> all 3
> channels? Are you using switching mode power supplies for the LED strips
> and/or the ESP? Have you applied proper decoupling/filtering for the ESP
> and LED supply?
> The IRF510 is also not the optimal choice for 3.3V PWM signals.

Thanks for the quick reply.
 The noise does not seem to be present when I switch the Mosfet on. At 
least I can not observe the characteristic I've shown you in my picture.
At the moment I'm using my lab bench power supply. What do you 
specifically mean with decoupling/filtering? I've planned to place a 16v 
1000uf capacitor between Vcc and GND, but I have to wait for them to 
arrive. I used the last one a while ago.
Can you make a suggestion for a better Mosfet? It should be able to 
handle at least 5 meters of LED strip at 12V.

stefanus wrote:
> I hardly wonder why you have so much noise in both circuits.
> Something
> is going very wrong here.
>
> Please show us an oscilloscope picture of the power supplies (3,3V and
> 12V).
>
> Please show us pictures of your setup.
>
> When you switch off an LED, does your Microcontroller output an active
> LOW level or do you switch the pins into High-Z (input) mode?

I will post a picture tomorrow. I have to get up early tomorrow morning.
The ESP will produce the 3.3v itself. I will make you an osci picture 
from my 12v supply. By pictures of my setup, do you mean my circuit or 
my equipment? Or both? I control the Mosfets by supplying a PWM signal. 
Wenn I switch them low, I will set analogWrite(pinR, 0 );

argos wrote:
> Is there any noise on any of the gates when all 3 strips are turned off?
> Is there any noise when you switch the strips on/off without PWM? How
> long are the gate lines?

I tried to switch the Strip on and off manually, but with the same 
result. The noise is still there.


> At least some capacitors (electrolytic+ceramic) on the supply line close
> to the ESP and LEDs. The 5m strips make a formidable antenna spreading
> the PWM everywhere.

I added a few capacitors to the beginning of my circuit. I switched 
multiple in parallel to get a total capacity of 400uf. I don't have any 
bigger ones lying around at the moment.


> "5m LED strip" doesn't tell anything about the current flowing. IRLZ44N
> was already suggested. FQP30N06L, FQP13N10L, IRL640A or something like
> IRLML6344 or STN4NF03L in SMD. You can also do a search for logic or
> super logic MOSFETS on Digikey/Mouser.

I will definitely try that, to get the maximum out of my controller.

I added some pictures to this thread. I have taken a picture from my PSU 
( which also outputs noise), from my setup on my breadboard, from the 
LEDs manually switched on and off, and a shot with capacitors engaged.

stefanus wrote:
> Fabian W. wrote:
>> The ESP will produce the 3.3v itself. I will make you an osci picture
>> from my 12v supply.
>
> And check also the 3,3V supply. With such huge problem you should not
> simply trust that the bought hardware is Ok.

I probed the 3.3v regulator and could not observe any faults. It seems 
to work ok.

stefanus wrote:
> The picture from the PSU is very very bad.
>
> Please measure again directly at the output of the PSU (not somewhere on
> your breadboard). If the noise is still visible then you should consider
> to return the PSU.

I measured directly on the terminals of the PSU and the noise was still 
there. Returning the PSU is a bit late bc its my lab bench supply. I 
have it for about 2 years now and never saw such a negative side effect 
like now.

>
> Breadboards are not suitable to supply ESP modules. They have too much
> resistance in their contacts. Also, the Dupont wires are often thin or
> made of cheap materials (not copper).
Maybe this is the reason, it works on the Breadboard because the 
resistance of the parts is so bad, it cancels out the negative side 
effects of my PSU.

>
> Please use soldered connections between PSU and the ESP module.
>
 I will update my connections and see where this brings me.

> Crocodile clips are also often not good enough due to their internal and
> contact resistance. I took a while until I could believe that when I
> discovered it the first time. You may be able to measure a significant
> voltage drop between the incoming and outgoing cables that are connected
> by these crocodile clips.

I measured through the cables and seriously could observe a voltage 
drop. I can not believe this is a thing.


stefanus wrote:
> One other thing: The big electrolytic capacitor in the right lower
> edge
> seems to have reversed polarity.

I double checked it, and its in the right polarity. The picture is just 
bad and the cap slightly turned.

stefanus wrote:
> Just to make clear:
>
> You main problem is that GND is not GND everywhere. You have a
> significant voltage drop on the cables and connections caused by the
> supply current of the ESP chip.
>



> You can check it. Connect the GND wire of the oscilloscope to the GND
> output of the power supply. Then check the voltage at any part where you
> would expect GND=0V  (e.g. at the 3 transistors and at the GND input of
> your ESP board).
>
> I am pretty sure that you will see much more than 0V.

I will definitely check that.

I see where this is going. My next best guess for this problem is, to 
decouple the esp from the less with a separate voltage regulator. This 
could probably reduce the noise in my circuit and can lead to better 
results on my gate pin. The problem is, that I can not opt for one 
single PSU in later use, because I want to use this device with 
different switching PSUs for a variable length of LED-Strips. Therefore 
I can not guarantee that the input for my device high quality. The best 
solution would be to completely decouple the esp from the rest of the 
circuit, Or maybe, as you mentioned before, use a separate transistor 
driver to isolate both circuits from each other.

argos wrote:
> Fabian W. wrote:
>> I measured directly on the terminals of the PSU and the noise was still
>> there.
>
> The noise from the power supply is awful. Instead of an additional
> voltage regulator, you could try a simple RC filter in the supply line
> of the ESP.

OK. I will first have to read a bit about RC filters. I've never built 
one. But I will definitely try it with that.

argos wrote:
> Fabian W. wrote:
>> I measured directly on the terminals of the PSU and the noise was still
>> there.
>
> The noise from the power supply is awful. Instead of an additional
> voltage regulator, you could try a simple RC filter in the supply line
> of the ESP.

Ok. I came up with this circuit. It reduced the noise drastically. I 
mean it works, but I'm not sure, if this is the right approach. Maybe 
you can tell me if this is the right way to do it. I probed the circuit 
with my oscilloscope, and it seemed fine.
As I said, I'm not totally into this topic yet.
I would be glad for any advice.

stefanus wrote:
> The right way is, to fix the problem at its source, which is your
> PSU.

I did understand what you Sayed. Nevertheless this isn't an option at 
the moment, because this is for test purposes. I do not know which psu 
I will use later on in my build. It could be, that the psu is better 
than this one. It also could be that it is equal or worse, I don't know 
yet , because I haven't bought it yet. The thing I want to do now is 
making my circuit bullet proof against this kind of problems. Therefore 
I want to know if the circuit I made is sufficient for that. I know that 
I have to switch the PSU but that is not a thing for today.

Is the circuit I made ok as rc filter ?

argos wrote:
> Fabian W. wrote:
>> Ok. I came up with this circuit. It reduced the noise drastically. I
>> mean it works, but I'm not sure, if this is the right approach.
>
> Yes, a LC-Filter is fine too. Keep in mind that with a working PSU you
> wouldn't need a filter at all. Are the gate voltages now clean and the
> LED strips working as expected?

Sorry that I couldn't answer earlier. I connected an improved version of 
my input filter and now have a very clean output voltage.
The only problem now is, that the random LED flicker is still there. I 
measured the Gate pin of my transistor, and I can not observe anything 
suspicious. NO RIppeling, no weird current spikes. Only thing I have is 
a random "Noise" of about 17mv peak to peak. But I think this could be 
ignored, since the voltage is so small, it should have no effect on the 
transistor. Since I solved my PSU issue with my input filter I have no 
idea what could cause this flickering behaviour.

The LEDs flicker only when they are off. When I turn them on, they are 
stable. I just tried it, but i can not observe any flickering during on 
time. I conected my input filter to my esp and my Led with the same 
result.

stefanus wrote:
>
> There must be a reason why the transistor switches on.
>
> Measure the voltage between Gate and Source of the transistor with your
> oscilloscope. You may use its trigger function to catch the moment where
> an unexpected voltage occurs. Also measure the voltage between Source
> and GND of the microcontroller (should be 0).

As you tole me, I measured between source and gate of the transistor. 
When I connect the GND side of my probe to source and the probe itself 
to the gate pin, nothing happens very much. I get some low background 
noise of around 12 -30 mv. Nothing that should trigger the transistor. 
However if I swap the probes and connect my probe side to source and the 
gnd side of the probe to the gate pin, I have frequent pulses of about 
300mv. I don't know why that's so different. I checked my Gnd with the 
Source pin of the transistor and it is about 0v.
By the way, my new transistors arrived. The ones that were suggested in 
this thread. I will try them out shortly and report if this changes 
something.