Choosing the right chip

[notbad]

Sorry if it’s a little off topic.
I was searching microchip website to choose a device based on specs and prices, but I am very confused.
Lots of new devices but I have a negative feeling about them because they’re unfamiliar to me. I was surprised to see that 16F873A is not in the list. Why do you think?
Also 16F84A’s price (which only has a timer) was twice the 18F26K22 (which has a lot of peripherals). That doesn’t make sense!

Can anyone please recommend some “general” devices with best tradeoff between price and features?
For example one 8pins, 18pins, 28pins and 40+ pins.

Thanks

[dyeatman]

The 16F87x series has been around since the "dark ages" :-) and
have been replaced by the 16F88x series.. Go to microchip and search
for the device, you will find their replacement recommendations.

Your other question opens a can of worms:

Which compiler(s) are you using (i.e. PCM vs PCH vs ??)?

What from factor (i.e. DIP versus TQFP, 28pins vs 40 pins etc.) can you handle? ( I try to always use 40 pin devices)

What voltage (3.3 or 5 or??)

What features do you consider essential beyond the standard
MSSP(SPI-I2C)/ADC/PWM (most chips have these in various
configurations)?

Would you normally require PSP, QEI, LIN, CAN/ECAN, CTMU, RTCC, DAC?

What speed would be minimum required?

First off, If it's a choice between the 16F and 18F series I would go
with the 18F.

The 18F46K22 is my "favorite" of the moment because of flexibility,
speed, 2 UARTS, lots of flash memory and RAM etc.
_________________
Google and Forum Search are some of your best tools!!!!

[temtronic]

I have to 'second' the 18F46K22 as my current favorite chip !!

Best 40 pinner so far, fast, tons of goodies, easy to use, cheap.

Actually though way 'overkill' for even simple projects, it is a 'one chip does all' solution.

I have some 16F648s here but 'man sized' PCB with 46K22 is easier for me. Smaller is NOT better....you always seem to need a couple more pins, a bit more RAM, etc.

hth
jay

[Ttelmah]

The key is production dates, quantities and competition.
Older chips use larger dies than many of the newer chips with dozens of times 'more' features. Production techniques have improved.
Historically, early chips have been made in good quantities, but nothing compared to modern designs. Some of the new chips have been sold as 'die' units, and fitted to things like smart cards. One customer of mine, is using over 100,000/month of one of the nano chips. No wonder they get a good price.
Other thing is competition. Old chips were priced aggressively when they were 'new', but now are produced as 'residual' designs for old applications only, hence pricing is much higher than for newer chips that are being sold to compete with the dozens of designs from other makers.

Best Wishes

[notbad]

Thanks Ttelmah. That really resolved the pricing dilemma for me.
All right, 46K22 for the 40 pinner. What about 18pin and 8pin.
@dyeatman: my requirements are pretty standard. You know… . pcd/pcm/pch, DIP & SO, 5v.
Important features are Timer, (E)CCP, ADC , UART, MSSP. The more, the better.

[dyeatman]

Haven't used 8 pin in any designs and haven't used 18 pin in many years so I can't suggest anything there.

However, the 18F26K22 is 28 pins with almost identical specs to the 46K22
and includes three 8 bit and four 16 bit timers.
_________________
Google and Forum Search are some of your best tools!!!!

[gpsmikey]

Yeah, the 18F26K22 is my choice for a nice chip with lots of goodies in it but does not take up the real estate of a 40 pin chip. And, if you do decide that you need to move to a bigger chip, then the 46k22 is an easy step (I guess you could also go smaller, but I don't think I have ever seen a design that got smaller ... ). And places like Digikey have them for $3.50 ea or $2.85 if you order 10 of them.
http://www.digikey.com/product-detail/en/PIC18F26K22-I%2FSP/PIC18F26K22-I%2FSP-ND/2480368

Here is a handy selector guide that you might want to look through
http://ww1.microchip.com/downloads/en/DeviceDoc/41438a.pdf


mikey
_________________
mikey
-- you can't have too many gadgets or too much disk space !
old engineering saying: 1+1 = 3 for sufficiently large values of 1 or small values of 3

[asmboy]

i choose the optimum chip for a design based on the
features i need. And there is no guarantee that
any one pic will have all you need.

for instance, only the 16f150x has an NCO , which is a wonderfully useful
feature if arbitrary frequency is a requirement. But not 2 EUSARTS.

In general the 18f family is great if the historic PIC basic features(timer/pwm/EUSART/ADC/ I_O.core) are all you need,
but it seems useful NEW stuff is appearing in the 16F family first.

decide what non-core I_O features your design needs, and let THAT decide your pic model. I use the conjunction of required functionality and $$ decide on what pic model.

if you use an 18f46K22 where a 12Fxxx
part will do the job, you are being lazy, silly and wasteful all at the same time.

I tend to design the application from the outside IN and choose the PIC based on only that. At the END of the circuit design.
just my 2 cents.

[John P]

Note that if you use a 40-pin chip where a 20-pin one (or smaller!) would work, you're wasting circuit board space, and you will pay for that, and so will your customers if it's a commercial product.

The PIC16F84 has been obsolete for years, but Microchip keeps making it because people just won't stop buying it, even if it does less and costs more than other chips. You can't blame them for grabbing easy money.

[notbad]

Thank you guys for your advice.

[Ttelmah]

For 8pin, look at the 12F1501. Amazing little chip. Has ADC, PWM, NCO, CWG, comparator, DAC & CLC. It can replace pretty much any other 8pin device (means you can keep 'number of devices' low). I tend to use these as 'glue logic' in quite a few places. Cheaper than a PLD, for small things.

Tend to jump to 28pin then.

The 18F26K22, or 25J50. Latter if I want USB.

Recently on 40pin, have been tending to use the 18F46K50, or move up to some of the 24F chips.

Key thing before making any decision, is to think what peripherals you are _likely_ to use, and then look at the errata. Many of the newer 3.3v devices, actually seem to have less problems than the 5v devices from perhaps 5 years ago. There are rather a lot of devices that seem to have been made in a hurry a few years ago, that have some really nasty errata....
Provided you get the A4 revision of the 26K22, it has very few errata that are likely to affect you.

Annoyingly the Microchip site 'used' to lead you straight to a page for each chip with the errata for that chip listed. Now you seem to have to go odd routes to get to these. Easiest I find, is to put chip number into the main search entry (not data sheet), then the very top of the page given, has 'device page', which is the old page with everything on it. Select 'documentation & software', and you get data sheets, errata, and programming specifications, all listed.

Best Wishes

[asmallri]

You did mention that you considered PCD to be a standard tool for you. In this case, I have to admit to being spoilt as I consider the PIC18 to be legacy, and PIC16 to be prehistoric. I now use a PIC24/dsPIC33 for most general purpose projects - Better system architecture. Although for specific customer projects I will look at alternative architectures, I am more likely to use a PIC32 and a PIC18F.

I have not used a PIC16 for more nearly a decade.
_________________
Regards, Andrew

http://www.brushelectronics.com/software
Home of Ethernet, SD card and Encrypted Serial Bootloaders for PICs!!

[gpsmikey]

Yeah, but you drive on the wrong side of the road there ... oh wait, that is US that drive on the wrong side :-)

So which of the 24F ones do you like for a general purpose one ? I realize there are lots of different chips out there, but if you pick a good general one, it is amazing how many projects can be accomplished with it. I had liked the 18F26K22, but looking at the 24F, something like the 24FJ32GA102 looks very handy although it does require the next level up compiler.

mikey
_________________
mikey
-- you can't have too many gadgets or too much disk space !
old engineering saying: 1+1 = 3 for sufficiently large values of 1 or small values of 3

[temtronic]

One thing often overlooked in finding the 'right' chip is the overall time spent.We all _know_ we'll need 2-3 more pins, RAM ,etc. as the project takes shape.Then there's the 'new chip syndrome' where a feature or function that did work with another chip suddenly doesn't.Or migrating to a new series / compiler also has a 'learning curve'.
Yes, I know a 40 pinner takes up more space however I'm not concerned about 20-30 cents of PCB, or spending a buck more for the 'bigger' device. Consider the cost when you lose a week of R&D at $100/hr ! Parts are cheap compared to us humans.Something like 'penny wise.pound foolish' comes into play here.Getting a very reliable product to the clients in a reasonable time should be #1 priority.

oh yeah...by the time you've decided on the 'right' chip...Microchip will have announced 3 'new and improved' versions that are now the 'right' chip.


jay

[jeremiah]