[sdiy] Andromeda A6 hardware questions

[cheater cheater]

Actually, no, this topic is so derailed already I'll just start a new one.

D.

On Wed, Mar 3, 2010 at 14:35, cheater cheater <cheater00 at gmail.com> wrote:
> Neil,
> where can I read your notes on the W-30 disassembly?
>
> ____
>
> I don't think writing a real-time system is something that needs to be
> a challenge. The theory *and* practice have been there for decades. I
> can't see this being more difficult than, say, learning functional
> programming; I've done some reading up on hard realtime systems in the
> past, and it doesn't look like it bites.
>
> Linus had it easy because he collaborated with other people. That's
> the selling point here.
>
>
> Anyways,
> guys, seriously! I came here with some very specific questions about
> technical aspects of hardware at hand. This topic is not about
> religious issues so please leave that at the door step! If you don't
> know about the Andromeda hardware, the topic might not be for you, and
> I suggest starting a parallel topic about whether it's feasible or not
> to modify synth firmware - it might be difficult to believe but I have
> actually thought about this!
>
> I see ten VERY long emails and not one even near to answering any of
> the technical questions, but 99.5% of it is serious discouragement
> based on personal bad experiences - is this the spirit of SDIY?
>
> D.
>
> On Wed, Mar 3, 2010 at 14:16, Neil Johnson <neil.johnson97 at ntlworld.com> wrote:
>> Hi,
>>
>> cheater cheater wrote:
>>> Tom,
>>> as they say, every journey starts with the first step.
>>
>> 'They' also say "may contain nuts".
>>
>>> On the other hand, I think you are overestimating the complexity of
>>> the andromeda's digital part. It's just some menus, midi, patch
>>> storage, a simple sequencer, voice assignment, and modulation. That's
>>> not *that* much.
>>
>> Interesting ... on what basis do you make this judgement?  Many years of practical experience of designing embedded products?  Or something you read on Wikipedia?
>>
>> All the points Tom raises are very valid:
>>
>>> > I think it would. Designing a hardware synth of that complexity would be a
>>> > serious job in itself. Unpicking one that someone else has already done is
>>> > as much work, and then you've got to rewrite all the software, since you're
>>> > not going to get access to the original code. That adds up to twice as much
>>> > work by my reckoning.
>>
>> I would go further and say *more* than twice.  Lets just consider the code alone, excluding reverse-engineering the schematics.  Sure, you can get a ROM dump (or dumps if there are multiple micros).  Then decode that back into assembler (code, const, data, jump tables, etc).  Then pick out the structure (OS, library, application).  Then unpick the weird stuff (optimized code, etc).  Then try to work out all those little work-arounds that are needed due to bugs in libraries and hardware.
>>
>> Plus lots of other stuff that others may also suggest.
>>
>> THEN ... you need hardware with some way of injecting test programs into the micros (JTAG?  FLASH ROMs?  EPROMs?) to experiment with the actual hardware to try to reverse-engineer the datasheets of the custom chips ... and even if you did get hold of the datasheets you still need to check that they are correct.
>>
>> Now you're in a reasonable position to start on your new implementation (assuming by this stage you haven't got distracted by the next fun thing on the market).  Once you've got all the tools you'll need - assemblers, compilers, linkers, debuggers.  Then you're into bringing up the RTOS and writing device drivers.
>>
>> Oh, and which particular hardware build and software revision are you going to target?  Be sure to note any cuts and straps on the PCBs.
>>
>> Back to your own list:
>> = menus
>> = midi
>> = patch storage
>> = a simple sequencer
>> = voice assignment
>> = modulation
>>
>> There's a lot of realtime stuff there - easy to get wrong, very difficult to get right.
>>
>> By comparison Linus had it easy - writing a new, small kernel for a well-documented processor.
>>
>> Colin Fraser did some sterling work on the JX-10 assigner micro:
>>    http://www.colinfraser.com/jx10/jx.htm
>>
>> Noting the creation date of his notes spreadsheet and the date of the latest update he's spent ~10 years on this project. (Perhaps Colin could confirm this?)
>>
>> Another example, this time my own efforts with a Roland W-30:
>>    http://www.milton.arachsys.com/nj71/index.php?menu=2&submenu=4&subsubmenu=2
>>
>> Took about 3 years to pull that lot together to get to the point of rudimentary commented disassembly and notes on the system architecture.
>>
>> And Rainer has been at it since 1997:
>>    http://www.buchty.net/ensoniq/index.html
>>
>> But these aside, please, feel free to buy an A6, take it apart and start hacking - we'll all be watching with great interest!
>>
>> Cheers,
>> Neil
>> --
>> http://www.njohnson.co.uk
>>
>>
>