DC converter vs voltage regulator

[asap3210]

Can anyone please tell me if I can safely replace the voltage regulator and capacitor with a DC-DC converter MP1584EN 3A?
Also, is the MPU6050 reading more stable with the capacitor?
The supply voltage is 12 v (LIPO battery)

 

[betwixt]

Yes you can but make sure the output voltage is set before connecting to the ESP board.
If you still want to measure the supply voltage on 'D34' you should note that the maximum voltage allowed at the input to the ESP32 is 3.3V so if you start with a voltage higher than the original battery you have to alter the values of R2 and R3 to scale it to a safe range.

Capacitors on the supply rails are mandatory, you will get all kinds of problems if they aren't fitted.

Brian.

 

[asap3210]

Yes you can but make sure the output voltage is set before connecting to the ESP board.
If you still want to measure the supply voltage on 'D34' you should note that the maximum voltage allowed at the input to the ESP32 is 3.3V so if you start with a voltage higher than the original battery you have to alter the values of R2 and R3 to scale it to a safe range.

Capacitors on the supply rails are mandatory, you will get all kinds of problems if they aren't fitted.

Brian.

"so if you start with a voltage higher than the original battery you have to alter the values of R2 and R3 to scale it to a safe range." -

You are right! I completely forgot about that. Yes, I need 3.3v max at D34. Thanks a lot.

 

[KlausST]

Hi,

Yes, I need 3.3v max at D34

Indeed even 3.3V is too high.

The power supply has tolerance. Let's say +/-5%, so worst case the supply is 3.135V. Expecting accurate 3.3V is not realistic.
So it may survive 3.3V ... but it can't decode it.
Additionally the ESP32 ADC is rather nonlinear at both ends of it's input range...
Thus I recommend to leave a lot of headroom.

Klaus

 

[betwixt]

I think the maximum ADC input at the ESP32 module is 1V but there is already a divider on the 'nodeMCU' board. Klaus is correct, there is a flat response at both ends of the scale and to make matters worse, the WiFi transmitter starting up tends to make the ADC reading 'wobble'. I use lots of ESP32s but now always use an external ADC.

Brian.

 

[KlausST]

Hi,

I also used the ESP32 ADC.
I can't remember clearly, but I also read somewhere that the range is 0..1V only, but experienced differently.
I din't think there is a voltage divider in the nodeMCU .. if there was the pins could not work a digital IO anymore.

Too bad I can't remember ... but I guess there is a reference setting to choose the ADC input range.


Update:

Analog to Digital Converter (ADC) - ESP32 - — ESP-IDF Programming Guide v4.4 documentation

According this there are "ESP32 internal attenuators" to adjust the range.

And yes, the ESP32 ADC performance is really poor. It has accuracy problems, linearity problems, noise problems.
If you need some accurate results you need to use an external ADC.

Klaus

 

[danadakk]

Some intesting data measuements and corrective approaches to the ESP32 ADC :

Do You Know About ESP32 ADC Adjustment?

Do You Know About ESP32 ADC Adjustment?: Today, I'm going to talk about a more technical issue, but one I think everyone who works with ESP32 should know: the issue of ADC (analog-to-digital converter) read adjustment. I find this important because when doing a "measurement," especially wi…

www.instructables.com

Of course one can always do this at manufacturing / programming time -

Above a simple cal routine generating a table of corrections for use in table form, or
polynomial coefficient generation, least squares.....


Regards, Dana.

 

[betwixt]

Fixing the calibration isn't the issue on those devices, as long as the RF isn't switched during an ADC conversion, they are quite linear. The problem is that both ends of the range are quite flat, it takes several mV above ground before the ADC measures anything but zero and at about 97% of FSD it reads constant maximum. As long as measurement is taken 'mid-range' they are fine. No amount of calibration will return a useful value when the ADC has flatlined. It may not be a problem at the top end because scaling the voltage can keep it within the bad zone but at the bottom the results can be significantly out.

Brian.

 

[danadakk]

Using cal depends on the actual signal path response, and how its
non linearity is generated. If non linearity is literally an on off switch
in signal path no amount of cal will fix that. But if it is a continuous,
causal, albeit non linear f(), then cal can work. Of course if non linearity
dynamic range is huge, small perturbations exceed dynamic range of
measurement ADC system, that would eliminate correction. But if
its a "soft" non-linear response it might be possible to use a cal routine.

We have to know what the root cause of distortion is before we can
draw any conclusions. Or just implement a correction system and see
if its repeatable and meets some accuracy goal.

Not to forget we have the whole issue of reference generator, especially
if we care about absolute accuracy. And of course statistical issues like
noise, although we have excellent methods of handling that, eg. CDS
as example.


Regards, Dana.

 

[danadakk]

Here is some work that was done using linear equation -

ESP32 Analog To Digital Conversion Accuracy - IoT Kits

The ESP32 is a powerful microcontroller with many input/output ports. Specifically, it contains two 12-bit multiplexed analog to digital converters (ADCs) for a total of 18 channels. ADC1 is attached to eight GPIOs from 32 to 39. ADC2 is attached to 10 GPIOs (0, 2, 4, 12 to 15 and 25 to 27)...

w4krl.com

Would be interesting if it was done with least squares or a polynomial curve
fit.


Regards, Dana.

 

[Alan8947]

Unless there's a good reason, why change the regulator of all things? 78x12 is analog, not generating any noise. You can use switching regulators, but you need to know what you are doing and what kind of circuit to make sure noise is not an issue.

I did not look into your circuit, just general comment. I use switching regulators more often, BUT I work with SIGNAL INTEGRITY and RF stuffs and watch out the layout of the DC to DC converter(we designed ours) and circuit layout of our circuit to minimize interference. Unless you know what to do, WHY CHANGE?

I would be very careful if buying DC to DC converters. Don't count on they know what they are doing(assume they don't).

JMHO

 

[danadakk]

One caution you can exercise is look at these low cost DC/DC, their
parts visible, and layout, and compare to datasheet recommendations.
If they seem to match up manufacturer recommendations you might
have a possible solution.

Many (buck) based on this part, and schematic looks like :

Datasheet has extensive layout recommendations.

Of course the 2596 could be a Chinese knockoff, or Schottky a poor choice, but at these
prices ($1) buy one and evaluate it.

This part supported on TI Webench as well.

Regards, Dana.

 

[Alan8947]

OP has not been back for a while. I have no idea his level of knowledge and the reason for changing to a DC DC converter. He claim he's "newbie" and from the wiring diagram drawn instead of schematic, I take that he's a newbie. I would NOT change the 78x12.

Obviously, this is a project to play with the circuit(which I never even look deeper into), there are more important things to do and learn than the regulator. Play with the circuit first, then worry about the regulator. Don't want to add one more unknown to the equation. AND YES, using a DC DC converter will add an known.

Just from looking at the circuit diagram, there are so many LOOPS between the circuit board and each of the "Nidec"(which I don't know what). If it is wired like that, there are a lot of good LOOP ANTENNA(like UHF tv antenna). Those are good in picking up radiated noise. One has to be careful and aware of all that using SMPS(DC DC converter).

Do NOT count on the converter has UL or CE mark. I lead my team to pass quite a few CE test for our system. It is NOT fool proof. Don't count on whoever design the DC DC know what they are doing even if it is not China knockoff. Even a lot of EE are not interested in SIGNAL INTEGRITY. I remember long time ago when company actually hired an expert to give a class on this, all my friends(EE) kept falling asleep, ended up mostly just me and the instructor talking to each other. Later on, my friends said to me "You can have it all if you are interested!!". Ha ha, I was one of the strange dude that gets excited chasing the current loop!!!

Just don't mug with it if possible. Even the ATmega328 I am learning, it has option to turn off the main clock when using ADC to prevent noise. They are the one that design the chip, still they won't count on it when doing critical reading with ADC.

JMHO