PIC10 Support ...

[ELCouz]

Just wondering if anybody here tried them (not mentioned a lot on this forum) with the CCS compiler.

They look restricted... but powerful for simple task.

Any experience with them?
_________________
Regards,
Laurent

-----------
Here's my first visual theme for the CCS C Compiler. Enjoy!

[Ttelmah]

They have been mentioned here on some occasions.
They really are only useful for the smallest of tasks, due mainly to the lack of pins/RAM on the smaller ones. Once you go up to the 8pin versions, you might as well use one of the PIC12's. Some of the newer ones of these have significant peripheral abilities (12F1840 for example), and so rather than carrying a lot of different chips, chips like these can make good 'glue'.
The only time the PIC10 then gets used is to save the last cent, where something can be done in the small device, and large quantities are involved.

[temtronic]

I've never really considered any of the GluPICs. Though they might have their uses, IF you consider the cost of R&D, inventory, PCB production, etc. versus using your 'regular' PIC, I seriously doubt you'll save any money and probably cost you more. Though I have used 8 pinners, I generally use an 18 or 40 pin devices. YES, overkill but save me time($$$) ! and already have a huge library of known, working code.
R&D is a real cost and spending a day learning the 'new' PIC is a lot of cash. Everyone knows you'll need more RAM, a wee bit more FLASH and another I/O pin.

just food for thought

Jay

[ELCouz]

By glue chip you mean replace discrete components (trans/gates/binary counter etc) with a tiny pic?

Makes sense could save money on components but may have issues with reliability in the long run vs discrete parts...

Pic are rock solid tho.

Never heard of a misterious random bitflip (see cosmic rays) that make the pic crash...
_________________
Regards,
Laurent

-----------
Here's my first visual theme for the CCS C Compiler. Enjoy!

[Ttelmah]

But (conversely) the PIC may have advantages over discrete parts.
Look (for instance) at the old monostable multivibrator. At low frequencies, dead easy to simulate in a little PIC like this. If you want (say) better than 5% timing accuracy, for the mono, you'd probably have to use polystyrene capacitors. The PIC, 'out of the box' beats this, then look at the price of a mono+poly, versus the PIC....

[temtronic]

Hmm.. Mr. T brings up a great use for 'GluPICs' ,especially in timers. I can see where they'd be cost effective in 'altering' the interface between the 'new' sensor and old controller,where it's cheaper/easier than redoing the controller. Mind you I'm a bit of a 'dinosaur in design', liking stuff I used a 1/4 century ago. Hard to believe they still make 74123s and 555s these days! Still I can't see ANY SMD devices which really takes the fun out of R&D for me.Sigh...guess it's now a young guys game...

Jay

[Ttelmah]

I just use a really big magnifier....

The really big problem with SMD, is the lack of serviceability. If you supply kit that ends up in the middle of a jungle somewhere, with conventional parts a small box of components, and soldering iron allows systems to be maintained. On SMD stuff, you end up having to have spare boards for everything. Not good.

However funnily you have hit exactly 'where' I've used several of the PIC12's, controlling and timing interfaces, between old and new kit. RS232 to 485, with buffer control in the board for example. For jobs like this, the 'mini PIC' is a perfect extra tool.

[ELCouz]

Thanks guys!

Never thought of that...to use pics as bridge between old parts discontinued and new parts.

Quite handy those tiny pics!
_________________
Regards,
Laurent

-----------
Here's my first visual theme for the CCS C Compiler. Enjoy!

[temtronic]

re:...I just use a really big magnifier....

yeah, but HOW do you keep your hands steady enough to solder them itty bitty things!!

ever since I turned 60,the body's not what it used to be....too bad vacuum tubes are so dear now...at least them things are 'man sized'.


Jay

[Ttelmah]

Now I have used SMD's for years, and a conventional soldering iron doesn't go near them. Hot air is the answer. You put solder paste on the pads, one drop of a latex adhesive where the part is to go, fit it with the magnifier, then run a hot air nozzle down the legs. Manual soldering is basically impossible, especially when looking at parts with leg pitches like 0.4mm. Once you have learnt the technique, the idea of using an iron 'disappears'.
You can always use a pantograph manipulator for small movements.
It's a different technique.

'Real valves' are big, bolted seals, sometimes around 10,000v eht. The funny thing is that some are still used. Things like RF heating, and radar.
Do you get that 'warm glow' thinking about them?...

My neighbour for many years, owns and runs the 'Chelmer valve company'. He has now moved to live on their new site, but a still relatively thriving business supplying valves.

Somewhere in my garage, I have my father's old coherer. That was how radio should be done!. Microelectronics, pah!.

It's terrifying where we are. I remember my first computer. Just under 30000 transistors all hand soldered. Couldn't do what a small PIC can do now, and the electricity bill was crippling. On some of this sort of electronics, you really felt you 'knew' where the electrons were going. None of this 'uncertainty principle' rubbish.

Now, back to 'on subject'.

The key thing about the small PIC's, is that they are so cheap, and can encourage splitting up of designs. It's foolish to waste a week trying to pack everything into one chip, if you can simplify massively by having some pre/post processing, and allow the main chip to just be handling the main work-flow. Things like Manchester encoders/decoders are things I have done recently in a pair of PIC12's. I prefer the ones that have things like I2C, since it makes it much easier to then communicate to the main system. I used the 12F1840. Very low errata list. Important. Makes working with a chip much easier. Has lots of peripherals (UART, MSSP - the master on this in SPI mode is one of the few errata that matters, CPS, PWM, DSM, etc. etc.. Enough ROM to do the encoding and decoding. Fast internal clock. The two chips are pre-programmed for the encoding/decoding jobs at the factory.

For example, I suspect I could write an I2C UART module in one of these in perhaps an hours work using CCS, with programmable baud rates, and a few characters of buffering, and the price would be less than most of the off-the-shelf I2C UART's available. Add one or two pins for an address, and most of the "I haven't got enough UART" problems could be fixed.
It's an 'attitude' thing. It seems so much easier to do jobs like this externally.

However when you start instead working on a design that is going to be used in large quantities, saving every part matters, and it is worth spending a week of coding to get rid of a $0.5 chip. However then the other argument may appear, if a part like this can do five or six jobs on one board, the commonality when ordering may allow cost to be saved.

In a very real sense, this is part of the power of CCS. Because it uses the same language for multiple different chip families (PIC10, 12, 16, 18, 24 & 33), you don't waste time having to learn a new language, when mixing chips from multiple families. Current project has 2*PIC12, 1*PIC24 1*PIC18. The user interface, timing etc., is all done by the PIC18 (powerful enough for the job). The front end signal sampling & timings are handled by the PIC12's, while the main maths (quite a complex 2 dimensional 'surface' equation to linearise the result from three sensors), uses the DsPIC. Also had the advantage in this case, that the signals from the actual sensors, could be returned in digital form, from the tiny potted 'head', rather than trying to bring analog signals any distance.

So for me, the smaller PIC's are always considered as part of the overall design.

Happy New Year everybody.

[ELCouz]