[time-nuts] Follow-up question re: microcontroller families

Bob Camp lists at rtty.us
Sun May 26 21:07:27 EDT 2013


Hi

Ok, *but*

Power?
Mounting / circuit board?
LED's?
Resistors?
Bypass caps?
Control inputs?

I believe I first learned how fast wire turns into "something else" when there's a short while JFK was president. That lesson in doing things quick and dirty has stuck with me. If a gizmo is going to keep running for quite a while you need to make it right. The original start to this thread was something that should work for years. The cost of wiring it up (PCB etc)  will dwarf the cost of the CPU.

Even with something simple (actually especially with something simple) all that stuff is going to add up. Even more so if you want the result to be reasonably rugged. Can you scramble wire it to a set of D cells and steal the LED's from a kid's toy? - sure. Does that make the rest of it free? - not so much. Doubly so when the kid comes looking for the toy :). 

Free parts can be a slippery thing. First simple test - can a total stranger (who keeps a neat house) get one for free as well? I can probably get all sorts of things "for free" if I use the work phone to make the call. That does not pass the test….. Second simple test - can I make 100 of them for a lower cost? Cost should go down as volume goes up….

If the objective is a time related gizmo (this being TimeNuts) simply making the LED blink can't be the the end of it. It's *got* to display the ADEV of the lunar orbit in morse code or some other highly useful thing. If it doesn't do something like that it's (possibly) off topic for the list. Turning this from a simple $ XX LED blinker to something much more complex probably adds less than 10% to the total cost. 

Optimizing one aspect of a design without considering the rest of it is rarely a good idea. Assuming your time is worthless generally leads to projects that don't complete. For that matter McDonalds will happily pay you minimum wage for your time. My apologies to those who enjoy hours of soldering mag wire in 3D constructs, most of us don't look at that as a learning experience the 100th time around. Most of the really fun stuff is done quickly, the rest of it - McDonalds is "fun" as well….Your time *always* has a cash value. 

That said - the 74LS74 isn't all that good at driving LED's - not a lot of current drive on those outputs. Same is true of the 74HC74. The 74AC74 is the one you want, 52 cents one up DIP at Digikey.

Bob

On May 26, 2013, at 8:17 PM, Hal Murray <hmurray at megapathdsl.net> wrote:

> 
> bownes at gmail.com said:
>> For making a blinking LEDs, it is hard to beat a 74LS74. However, a PIC, is
>> probably less expensive! :) 
> 
> Thanks.  :)
> 
> I was going to send a wise-ass comment, but then I checked some numbers.
> 
> Digikey, one-off DIP pricing:
>  $0.62 SN74LS74
>  $0.55 PIC10F200
>  $0.33  SN74HC74
> 
> So technically you are correct, but only because you are picked an ancient 
> technology.
> 
> Besides, the '74 needs a clock while the PIC has an internal clock calibrated 
> to 1%.
> 
> I suspect what's going on is that the '74 is pad limited(*) so you are paying 
> per-pad rather than per gate.  The PIC only has 8 pins, so if it's close to 
> pad limited it will be cheaper.
> 
> The PIC10F200 has 3 output pins so it can blink 3 LEDs independently while 
> the '74 only has 2 FFs.
> 
> High volume (whole tape, 2-3K) prices are $0.34, 0.22, and 0.10
> 
> ------
> 
> *) If you aren't familiar with pad-limited, it's a cool idea.  Consider a 
> chip that has N pins.  Each pin needs a pad for the bond wire.  Arrange those 
> in a rectangle around the perimeter of your chip.  That leaves a hole in the 
> middle.  Put your logic in there.  If it doesn't fit, push the pads out until 
> there is room.  That makes your chip bigger and more expensive.  If it fits 
> with room leftover, you can add more logic for free.  That's why low cost 
> watches have so many features.
> 
> 
> -- 
> These are my opinions.  I hate spam.
> 
> 
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.



More information about the time-nuts mailing list