ASIC will be replaced by microcontroller....how?

[cupoftea]

Hi,
Suppose you have a 48 pin ASIC on a PCB which is programmed in C. But now its functions
can be done with a modern 48 pin microcontroller which is cheaper.
So you want to replace the ASIC with the microcontroller.
(the whole actual circuit and its functions will remain the same though.)
The software engineer obviously wants to just use the same code again in the microcontroller
that was used in the ASIC...but obviously
the pins of the microcontroller are different than the ones of the ASIC.

So you must do a Pin-mapping diagram to show how the software guy should manipulate
the details so that the ASIC code can be re-used in the microcontroller.
I believe there is a standard procedure which takes care of this?

Do you know what details are required to facilitate the software engineer in doing this?
ie, what is the minimum set of details that is needed?

Obviously an excel list needs to be made which should list all the pins of the ASIC and show what the NET NAME attached to each pin is.

Then you obviously need an excel list of pin numbers of the microcontroller, and next to each pin number you need the NET NAME that will be connected
to that pin.

(obviously the net names will be the same because the circuit is to remain the same, just that the ASIC will be swapped for the micro.)

So what is the minimum official procedure for doing this....? And what are the minimum details required.

 

[betwixt]

You won't ever replace all ASICs with equivalent microcontrollers, one does all its work at logic speed, the other at software speed so it depends on the timing needs of the final application.

Typically, somewhere is a C program there will be a table of "#define" directives that map a physical pin number to a name. Done properly, it will be in a header (.h or .inc) file but it can be in the main program if preferred. I have been programming for nearly 60 years and never used excel for pin lists, in fact I don't think I've used excel at all for over 20 years. The information is simply a short text file with lines like "#define spi_clk 25" so a spreadsheet is overkill and besides it makes it harder to add comments if wanted.

I am not aware of any program that converts an ASIC design to executable software, except in cases where the ASIC design may originally have been written in C.

Brian.

 

[dick_freebird]

Likely a fair bit of close-in part and value changes, supplies, signal level, etc. ASICs aren't made for no reason and may hide "special stuff" with a pile of digital attached, and the spec you may (or may not) get often is more of an acceptance test doc than enlightening about the guts

How much BOM cost savings times how many units to recover the development cost of figuring it all out to the point of blessed "major product change" qual and customer acceptance?

If this is a piece part availability/ life cycle thing, easier most likely to get the mask art and design licensed (EOL may make a deal) and shop a new home for it as-was.

 

[cupoftea]

The information is simply a short text file with lines like "#define spi_clk 25" so a spreadsheet is overkill

Thanks, and i believe also needed is another column saying what pin on the micro now corresponds to the pin 25 on the ASIC?
Also , another column for what the function will be called in the microcontroller...since often it will be a slightly different name?

You won't ever replace all ASICs with equivalent microcontrollers, one does all its work at logic speed, the other at software speed

Thanks, the ASIC had a 32.768kHz crystal attached, so does that pertain to the speed of its execution of commands?

Also, do you think its likely that the ASIC will draw a lot less Quiescent current than the microcontroller here?......kind of general "off the top of head" likely-hood guess on current consumption comparison?

 

[danadakk]

Thanks, the ASIC had a 32.768kHz crystal attached, so does that pertain to the speed of its execution of commands?

No, typically ASIC has a bunch of PLLs generating internal clocks, "usually" at much higher speeds.

Also, do you think its likely that the ASIC will draw a lot less Quiescent current than the microcontroller here?......kind of general "off the top of head" likely-hood guess on current consumption comparison?

Very application and function specific, so not a "right" answer here.

If you have to replace an ASIC there are processors out there that have not only loads
of standard peripherals and Analog capabilities but also have ability to create special
HW functions using schematic capture and/or Verilog, eg. they have a fabric that can
be used. And low power thru peripheral/resources power management.

 

[cupoftea]

they have a fabric that can
be used. And low power thru peripheral/resources power management.

Thanks, and since this ASIC was designed 20 years ago, its very likely that the modern microcontroller will be more efficient regarding current consumption?

 

[danadakk]

Thanks, and since this ASIC was designed 20 years ago, its very likely that the modern microcontroller will be more efficient regarding current consumption?

In general yes, but again

Very application and function specific, so not a "right" answer here.

 

[betwixt]

It is possible 32.768KHz is multiplied up in a PLL but more likely it is used because its 2^15 so can be divided down to exactly 1Hz for timing purposes. It is a good compromise between the amount of logic needed to divide 15 times and the physical size of crystal. 32.768KHz crystals are also mass produced for digital watches and clocks so they are very inexpensive.

I think you need to give us more information about what this ASIC does and how time critical are the signals it produces.

Brian.