[Solved] Servo movement range limited to 90 degrees with NodeMCU V2 - potentiometer control

Hello. I apologise in advance for my inexperience but this is my first attempt at servo control with the NodeMCU V2. Right at the start I hit a problem that I don't understand. With simple code I want to control the servo with a potentiometer, but the servo only moves around 90 degrees (roughly from 45 to 135). The same thing happens when I upload the "Swap" example from the Servo.h library.
I don't understand what I've done wrong, the current is rather not missing, the grounds are common, I've connected the potentiometer to 5V as well as 3.3V, and I've also controlled the servo via a transistor and connected it to different pins, I've copied examples and schematics from the internet.... still the same. On the serial monitor the values from the potentiometer show me 13-1024, which is rather ok. The cat variable also operates between 2 and 180.

I also connected the MG90 servo and it behaved identically.
After plugging everything into an Arduino Uno and uploading the same code, everything works as expected so it's more of a NodeMCU thing.

I was wondering if maybe it's the Servo.h library that works badly with the ESP8266 but in all the examples I've found people are using it successfully. I installed older versions of Servo.h but that didn't help either.

I am using a NodeMCU V2, an MG995 servo and a 10k potentiometer.

I have included a link to the video and a screen shot of the code below:

https://www.youtube.com/watch?v=UPR1Ny_D6Fo

.

khoam wrote:

Krismir102 wrote:

I was wondering if maybe it's the Servo.h library that malfunctions with the ESP8266

.
Which specific library is this referring to? Link?

.

https://www.arduino.cc/reference/en/libraries/servo/

I also noticed that ESP isn't there, but all the tutorials on connecting servo to Node that I found use it. At least it seems that way to me because in every code (on ESP too) it is included:
#include <Servo.h>
If this is the incorrect library then I have no idea what other library to use.

khoam wrote:

Arduino Core for ESP8266 already has its own Servo library bundled and there is no need to install an additional one in the Arduino IDE (best to uninstall it).
Link

.

There is indeed a library for ESP with servo. I removed the one there. Unfortunately after uploading the example from the library you are talking about the servo behavior did not change.

khoam wrote:

Try running the "Sweep" example from this library: Link .
You need to match the GPIO number to your configuration.

Unfortunately, it continues to do the same thing. It does not rotate more than within 90 degrees. It appears that I have the library correct.

khoam wrote:

How is the servo connection via this transistor? What kind of transistor is it? Maybe post a schematic, even a handwritten one.

.

The transistor was an act of desperation and I have already removed it. I now have the signal wire from the servo directly connected to a pin in the Node.

But if you want to know it looked like this:
.

khoam wrote:

The MG995 requires a PWM signal with an amplitude of 5V at the control input, from the ESP8266 pin is 3V3.

I checked with a logic state analyser the PWM signal that the Node provides. It is exemplary. I have now routed the control signal through the optocoupler and checked, it is 5V. On the power supply when the servo moves, the voltage drops from 5V to 4.8V.
Unfortunately, the servo still moves in a limited range. Everything is working and it is not working.... Maybe some interference?

Ha! Problem solved. Something amazing. It turns out that everything works as it should just that the servo library from the ESP environment is feeding the PWM signal TOO ACCURATE. I discovered this with a logic state analyser, with which I checked the PWM signal fed through the Node and the Arduino Uno. I recall that on the Node the servo moved in a limited range and on the Uno in a full range. The library from the Node generated a downright exemplary PWM signal, with the shortest fill time of 1 ms and the longest of 2 ms, and all to the nearest hundredth of a ms. The library from Uno generated a PWM signal with a shortest fill time of 0.5 ms and a longest of about 2.4 ms!!!! At least both agree on the 20 ms period. I installed the older ESP8266 environment library and suddenly everything started working! After reading the PWM waveform, it appeared to me that now my Node, on the older library, generates a signal with a fill range of 0.5 ms to 2.4 ms over a 20 ms period. In this fill range, 1 ms for my servo is about 75 degrees, and 2 ms is about 150 degrees. And indeed, that is about the extent to which it was moving for me. It is a pity that the documentation I came across gives the correct fills as 1 ms and 2 ms.... I don't know about other servo models, but the Tower Pro SG90 and MG995 servos require the ESP8266 environment library version 2.7.4 or older. Versions 3.0 and later generate an inappropriate PWM signal for these servos. Below is a graphic showing how my servo works. Probably in a 200 zł servo and some of the latest ones this problem will not occur, but mine for three grand from allegro has such requirements . . .

khoam wrote:

There is a version of the attach () function in the Servo library with additional arguments that allow you to specify a minimum and maximum duty cycle. Actually, the default is 1000us and 2000us in Core 3.0, but this can be changed just with the attach() function:

Code: C / C++

Log in, to see the code

A version of this function is available from Core 2.7.3.
Link .

Krismir102 wrote:

It turns out that everything works as expected only that the library for the servo from the ESP environment is giving the PWM signal TOO ACCURATE.

.
This is the value specified by the author of the code, as " uncalibrated default ". The aforementioned attach() function is used for calibration.

.

O good to know about this function I will definitely test it. Thanks for your help and time, I close the topic .

khoam wrote:

There is a version of the attach () function in the Servo library with additional arguments that allow you to specify a minimum and maximum duty cycle. Actually the default is 1000us and 2000us in Core 3.0, but this can be changed just with the attach() function:

Code: C / C++

Log in, to see the code

A version of this function is available from Core 2.7.3.
Link .

Krismir102 wrote:

It turns out that everything works as expected only that the library for the servo from the ESP environment is giving the PWM signal TOO ACCURATE.

.
This is the value specified by the code author as " uncalibrated default ". The aforementioned attach() function is used for calibration.

.

O good to know about this function I will definitely test it. Thanks for your help and time, I close the topic .

Added after 2 [minutes]:

Ha! Problem solved. Something amazing. It turns out that everything works as it should just that the library for the servo from the ESP environment is giving the PWM signal TOO much.
I discovered this with a logic state analyser, with which I checked the PWM signal fed through the Node and the Arduino Uno. I recall that on the Node the servo moved in a limited range and on the Uno in a full range.
The library from the Node generated a downright exemplary PWM signal, with the shortest fill time of 1 ms and the longest of 2 ms, and all to the nearest hundredth of a ms. The library from the Uno generated a PWM signal with a shortest fill time of 0.5 ms and a longest of about 2.4 ms!!!! At least both agree on the 20 ms period.
I installed the older ESP8266 environment library and suddenly everything started working! After reading the PWM waveform, it appeared to me that now my Node, on the older library, generates a signal with a fill range of 0.5 ms to 2.4 ms over a 20 ms period. In this fill range, 1 ms for my servo is about 75 degrees, and 2 ms is about 150 degrees. And indeed, that is about the extent to which it was moving for me.

It is a pity that the documentation I came across gives the correct fills as 1 ms and 2 ms....

I don't know about other servo models, but the Tower Pro SG90 and MG995 servos require the ESP8266 environment library version 2.7.4 or older. Versions 3.0 and later generate an inappropriate PWM signal for these servos.

Probably in a 200 PLN servo and some of the latest ones this problem will not exist, but my 3 grand servo from Allegro has such requirements

EDIT
On newer libraries, the problem is also solved by using the command that a colleague recommended above:
servo.attach(pin, min, max);
Where for min we enter 544 and for max 2400