OK, try to think 10k rather than 1M, gets a bit more realistic picture.

I'd buy professional tools if I had to make 2 units of a significant size. If that, and other projects, was not enough to carry the load, I would decline. This, of course, apply to "other" processors, I am fully tooled for the '51.

I would gladly make a small, say 1k, app with "no" tools, but the bigger the project, the more important the tools become.

I have seen more junk made with the excuse "we could not afford the tools" then for any other reason.

mainly to Jan: I was pulled, kicking and screaming, to C about 12 years ago and still go against the stream and code assembly when needed. I strongly recommend, you get a book and start adding C to your warchest, even if you may be restricted to SDCC. After a period of grumbling and complaining "I can do that better in assembler" you will start learning to like it. If your volumes are not huge saving a few hours by using C can easily pay for the extra $2 you may need for sufficient memory. In my young and inexperienced days I attacked everything to achieve optimum, but have since learned a

very important engineering formula:

 [the cost of engineering hours saved]/[no of units made] = ["free" additional component spending].

The corrolary: if [cost of engineering hours saved] > [cost of tool] buy tool.

Can you imagine the attitude towards engineering hours when 600.000 units are made. I got enormous praise for saving $0.87 by a memory trick (still pure C code). On the other hand, in a case where the anticipated production was 10, it was "do whatever to cut engineering cost, do not worry about parts".

Erik

PS the above does NOT state that a few can not make a workable unit without a tool budget. Just that too many try.