The most important elements in processor selection.

[ertansuluagac]

Hello to everyone
I want to something ask. What should I pay attention to when choosing a microprocessor? Does anyone have any idea about this?
Yeah, I know some basic things. But I want to take advantage of your experience
Respects.
_________________
Es

[Ttelmah]

There are so many factors this is almost impossible to answer. Some
general comments:

1) Don't choose the 'latest'. Don't choose really old. Look for a chip that
has been 'out' for perhaps a year or more, but is not antique.
2) Voltages. If you want to interface to things like an SD card, choose a
3.3v processor.
3) Calculate the peripherals you actually 'need'. Make this a minimum
list for the selection.
4) Then generally better to have more pins than too few.
5) Then assuming you do have PCH, the PIC18 chips are much more capable
than the PIC16 family.
6) Don't be afraid of using extra chips. For example, currently on a project
that needs a stepper control for five motors. It is cheaper and easier to
send an isolated control signal to a board controlling these near to the
motors, than to bring these high power signals down to a single board.

[ertansuluagac]

thanks
_________________
Es

[temtronic]

Would like to add to Mr. T's comments

3a) Select a PIC with more memories and pins ! While the cost is a little more it WILL reduce R&D time and overall costs.

I use the 18F46K22 for 99% of my products. Yes, it costs a dollar more BUT I've not run out of codespace, RAM, or peripherals. Yes, I've come close(used all but ONE pin, once). I consider it to be a 'seasoned' part, well tested. There is the smaller sister, a 28 pin device, might be better for you ?
There's no such thing as a 'perfect PIC' but over the years I've created a library of known, working drivers and functions for the 46K22. THAT has saved days if not weeks of coding and testing. Most don't consider 'bench time' to the overall cost of a product, especially for home use but spending an extra dollar 'up front' can save 100s if not 1,000s in the long run.

Also, I no longer use any PICs with internal USB (like the 4550). For $2, I get a USB<>TTL module which has connector, LEDs, 5/3V operation, USB code inside and well it works ! Actually by the time you price parts to get a 4550 to a PC, the premade module is only 50 cents more ! Less when you consider R&D time, PCB layout, etc.

I don't make itty, bitty sub micro sized PCBs. Maybe because I'm 66 and did PCBs with donuts and tape, but I like 4x6 inch PCBs. Lots of room for clean, easy access to parts and 'room to expand'. Yeesh, I've spent half my life 'playing with PICs'...

Jay

[ertansuluagac]

Well, I understand. How to create a circuit for the voltage applied to the input of the processors? So that he won't be affected by the noise. So if I need to drop 12 volts or 24 volts to 5 volts or 3.3 volts, what kind of circuit should I install?
_________________
Es

[Ttelmah]

Problem again is that this covers so many possibilities.

Start with a circuit that has enough input impedance to not significantly
affect the voltage involved. Now for a typical voltage this can just be a
resistive divider, but for some signals something with much higher input
impedance is involved. For example, I use some sensors where the source
impedance is over 1000M ohm. The input used to buffer these has over
1E12 ohm impedance, and has to use PTFE insulation. Then step the voltage
down or up to the level required (so op-amp, or resistive divider). If the
voltage is offset compared to where is has to be for the PIC ADC, use
op-amp addition/subtraction to shift the signal into range. As early as you can
have circuitry to limit the voltage to safe levels. Transorbs, diode clamps,
zeners etc.. However in each case be aware of the limitations. Each can
only handle particular amount of power, introduces loads and is limited
in the actual voltage/current capability. Remember too, that signal
filtering can be used (within the limits of what bandwidth you actually
want to measure), to reduce the 'noise' arriving at the ADC. However
remember that this too has 'consequences'. Phase shift typically.
Each extra stage of processing, itself introduces 'noise' (though typically
only at very small levels). However if you are trying to measure small
signals to high accuracy you can find that simple thermal noise from
resistors is enough to introduce significant errors.
You could spend a lifetime designing input circuits, and still be learning.
Especially when manufacturers launch something new....

[ertansuluagac]

thank you very very much Ttelmah
_________________
Es

[newguy]

Not mentioned yet: always consult the data sheet of the processor you are considering *and* its errata as well. There's been times where I've thought "this processor will do nicely" and then I look at the errata and slowly turn tail and run.

[Ttelmah]

Give this guy applause....

Vital.

[ertansuluagac]

I will use this circuit by applying a 12 volt input voltage. What do you think.
https://1drv.ms/u/s%21AskLNX_iFag3gfIi4mDc-1CLkmMGng?e=P9P9tc
_________________
Es

[Ttelmah]

As a basic circuit, fine. BUT. Layout and component selection are vital.
Low ESR (OSCON or tantalum capacitors for the two main ones. The
current sense resistor needs to be a non inductive type, and the board
layout needs thick traces kept very short. Also depending on what else
is involved you need to consider that the magnetic field generated by
the inductor, can couple into other traces nearby on your board.
The manufacturer will usually have a 'recommended' layout shown in
the data sheet. Follow it. Don't even consider trying to use a switcher
in a breadboard. 99% likelyhood of not working....
If you want to breadboard, instead buy an assembled module. For
example:
<https://power.murata.com/oki-78sr-3-3-1-5-w36-c.html>

Beware with these though 22uF max output capacitor or they can go
unstable.

[ertansuluagac]

Well, is there a circuit diagram you can suggest to me ? If you're sure you're okay, if there's a circuit diagram that works well, would you share it?
_________________
Es

[Ttelmah]

What you have posted, is quite capable of working well, _but_ requires
careful layout. Being a switching circuit, you have very high instantaneous
current changes in the traces. If these are not going to cause issues, good
layout is essential. For a basic breadboard or something being built on
veroboard, much safer and easier to just use a basic 7805. Downside
is of course that it is much less efficient.

[ertansuluagac]

No, I'm going to use it on the system by pressing the pcb.
_________________
Es

[Ttelmah]

In which case just lay the board out properly.

However 'be aware' of limitations here. Switchers will always introduce
noise. If you are working with low level analog signals these may become
important. So for example it is common on instruments using such signals
to find that the switcher is used to generate 5v, and then a linear regulator
is used to generate the 3.3v feeding the processor. Helps to reduce the
noise on this supply.