PIC Errata sheets

[echase]

I was quite surprised to look at the Errata sheets for a couple of PICs, one that has been on the market for years, to find that it is basically saying things like “Analogue input AN3 is inaccurate so don’t use it” or “ Before turning on feature x make sure you set register y.” When you develop your files for each processor do you take note of these Errata and write in any workarounds to avoid these problems?

If not I have to understand the inner workings of the PIC and C Code a bit more than I had expected in order to get reliable software. Fortunately the Errata are more common in the more complex features like SPI interfaces that I don’t yet use.

[JohnCrow]

Hi echase

You have raised a good point here.
I would imagine as you as, the erratas mostly effect the more advanced functions in devices.

But it could be a good idea to check the microchip errata sheets, if you have an obscure problem with a flowchart.(After fully checking everything else)

Data SHEET is a term that always makes me smile, something like the 16F877A data sheet is in the order of 400 pages. PDFs make it a lot easier keeping this data.
I worked for a firm in the 90's where the component engineer had about 3 shoulder height bookcases full of databooks, you could get all that on 1 cd-rom now

[echase]

My impression of Microchip is that they rush out impressive product after product to the market with little QA going on, so there are numerous things that don’t work very well. Although they do release bug fixes, e.g. new versions of the silicon, they don't clear all the problems and it looks as if they have no intention of ever clearing all of them. E.g. one newish 18F chip I am looking at has 20 Errata items. Only 3 have been cleared after 3 revisions in the silicon. Most are quite minor, like not performing to spec at low temperatures, but some are quite major like a bit on an SPI interface not being read with no workaround suggested .

Also I just bought one of their development kits. The supplier CD has no installer on it so does not install the supplied files into the right place on one’s computer, necessitating a call to Microchip to find out what to do.

I have no equivalent experience of other chip suppliers to see if Microchip is just the same as any other.

This is all nothing much to do with Flowcode but it does make it harder to get a working software out of Flowcode. How often do people come across a case of an Errata issue causing non working code?

[Benj]

Hello,

For my two pence worth I have made lots of Microchip related boards over the years using wierd and wonderful chips, peripherals etc and never really had a problem that could not be solved one way or the other. Microchip do seem quite good at getting the key problems fixed quickly, maybe because as you say they are making so many devices.

I have also experienced very down in the detail problems with Atmel devices where a INT pin and a UART interrupt were conflicting each other so it's not just Microchip that sometimes drop the ball on the silicone design.

This is also another reason why simulation is never going to replace having the thing there working in front of you.