From the user standpoint, what I want (and this applies more to 
integrated circuits that discreets) is to be able to choose a 
representation which already has the key attributes defined.

For example, if I intend to use a OPA2677T in a design, I want to be 
able to choose the OPA2677T from a list, and have the correct pinout and 
footprint already attached.  Currently, the only way I know of to do 
this is to have a seperate symbol file (containing very similar symbols) 
for each of OPA2677[HTU].  A better way would be to have the capability 
to drop a standard op-amp symbol and later, as the design evolves, 
choose from a list/hierarchy the specific component to apply to that 
symbol.  The specific component would apply the correct pinout and 
footprint.

Mark Moss



Timmerman, LJH wrote:

> Hi all,
> 
> So if I understand the discussion/mail-threads correctly:
> 
> a) You place a function (let's say an AND gate or a FF) in the
> schematic/design under construction.
> 
> b) Then you make the decision what device you need (let's say CMOS or TTL).
> 
> c) And then you decide which package (power consumption, timings etc.) you
> are going to use (AS, ALS, HCT or whatever is available/neccesary) which
> will lead to a certain pinout.
> 
> 
> In this case you will need to add a lot of attributes for every gate in the
> schematic.
> 
> Or an attribute like "CMOS", "TTL" which explains it all (for gnetlist -g
> [spice, verilog, VHDL] ) for your analysis/synthesis software.
> 
> And an attribute like "AS", "ALS", "HCT" etc. for the above mentioned
> software.
> 
> And then you would need an attribute with the vendor specific information
> which explains what pinout to be used (for gnetlist -g [PCB, bom, bom2]) for
> the pcb design and the shopping list.
> 
> 
> The only devices (in gEDA) I know that are usable that way, are the discrete
> components (like resistor, capacitor, diode and transistor and such).
> 
> Well, all this would need a massive rewrite of symbols and program code.
> 
> 
> Is this part of the rewrite Ales was talking about the other week ?.
> 
> 
> Is this all neccesary ?, for most of the time, and for most of the users,
> gEDA is very well usable (see various responses made in the past).
> 
> And let's not forget the "80-20 rule" (80 % of the customers buy 20 % of the
> range of products), translate this in "80 % of the gEDA users use 20% of the
> possibilities of the gEDA software", and maybe 20 % of the users use 80 % of
> the possibilities of the software  ;-)
> 
> 
> Maybe we should have a poll of what we expect from the gEDA software ?.
> 
> So everybody can have his/her say on this.
> 
> 
> FWIW my EUR 0.01.
> 
> Kind regards,
> 
> Bert Timmerman.
> 
> 
> 
>> -----Original Message-----
>>From:         Mark Moss <[EMAIL PROTECTED]>@CORUS  
>>Sent: maandag 3 december 2001 0:42
>>To:   [EMAIL PROTECTED]
>>Subject:      Re: gEDA: Symbol Conventions
>>
>> 
>>
>>
>>Magnus Danielson wrote:
>>
>>
>>>From: Mark Moss <[EMAIL PROTECTED]>
>>>Subject: Re: gEDA: Symbol Conventions
>>>Date: Sun, 02 Dec 2001 13:38:50 -0500
>>>
>>>Mark,
>>>
>>>
>>>
>>>>If I understand you correctly, the better solution would be to use the
>>>>full part number as the symbol name, thus equating the symbol and
>>>>component.
>>>>
>>>>
>>>No.
>>>
>>>The better way would be to choose a component, or alternatively a
>>>function, and the actual symbol (which is just short for symbolic
>>>representation) is then an attribute to the component. You migth even
>>>select among several symbol representations of the same
>>>function/component.
>>>
>>
>>Agreed.  This would be the best solution.  However:
>>
>>
>>
>>>This division is certainly not something which gEDA is supporting today.
>>>
>>
>>Within gEDA's current capabilities, what is the best way to represent
>>similar components with differing pinouts?  I haven't heard of a good
>>way other than to have a different symbol for each version of the
>>component.  (For example 74ABT1234-DIP-1 and 74ABT1234-PCC).
>>
>>Which goes back to the first question:  How should symbols for the same
>>component in different packages be named?
>>
>>
>><Snipped...>
>>
>>
>>
>>>I claim that today's model is not good for either of these.
>>>
>>>Just go an look on the Xilinx Spartan II symbols I contributed. Each
>>>of them where larger than any previous contributed symbol in gEDA but
>>>the actual difference between them where relatively small, especially
>>>since they are nearly pin-compatible in the same package.
>>>
>>>These symbols also breaks a gEDA rule in that the powersupply lines
>>>must be displayed, since applying different powersetup for different
>>>parts of the chip is well documented and indeed an important feature
>>>of the chip, and the 8 banks have their separate IO-pad setup also.
>>>
>>
>>I never thought that this rule was all that good of an idea.  It
>>probably works well with many digital designs, but I can think of a
>>couple of analog designs I have done where two instances of the same
>>component use different power supply rails.
>>
>>
>>
>>>So, not only when doing obscure analog stuff but also using modern
>>>digital chips there are a number of unresolved issues with todays way
>>>of doing things in gEDA. It hurts to grasp them all, but the sooner we
>>>learn just how ugly it is, the sooner we migth fix it and ensure it is
>>>indeed a _VERY_ good tool to use. The longer we wait, the more it will
>>>hurt to fix it.
>>>
>>>Cheers,
>>>Magnus
>>>
>>>
>>>
>>>
>>


Reply via email to