Re: gEDA-user: Is there a *official* JFET symbol in gEDA?

[John Doty <jpd@xxxxxxxxx>]

On Jun 26, 2009, at 5:39 PM, Anthony Shanks wrote:

> On Fri, Jun 26, 2009 at 3:43 PM, John Doty<jpd@xxxxxxxxx> wrote:
>>
>> On Jun 26, 2009, at 11:51 AM, Anthony Shanks wrote:
>>
>>> I see, I didn't know whats how spice-sdb works, never really used  
>>> that
>>> backend, just the spice one.
>>>
>>> What do you mean by overloading? Nearly every symbol shipped with
>>> gschem
>>
>> Those are just the starting points for a proper set of project symbol
>> libraries.
>>
>>> has a device attribute (caps, resistors, etc.), so it's safe to
>>> use the device attribute unless people are manually deleting it,
>>
>> Of course people are changing it to fit their project's flow.
>>
>>> thus
>>> freeing up refdes (this is how every other schematic/netlisting  
>>> flow I
>>> have ever used works in the professional field). In what instance
>>> would people be setting the device to a part number?
>>
>> Oh, for example, when the device= attribute identifies the part for
>> the bill of materials. Many library symbols have the part number as
>> the device= attribute (bav70lt1-1.sym for example).
>
> Ok, it looks like the bav70lt1-1 is a diode (from datasheet, didnt
> know what it was offhand). If you are putting this device in your
> schematic you are either simulating it as an ideal or using a subckt
> model provided by the vendor. How would you make that distinction in
> your schematic?

With the model= or model-name= attribute.

> In my mind logically how I designed my netlister to
> work, if you use device=DIODE it will netlist as a ideal diode, if you
> use device=SUBCKT, it will netlist as a subckt, with value being the
> name of the part (in this case, bav70lt1). Only 3 attributes are
> required, the flow is logical, and the part name is retained.

That's incompatible with the spice-sdb back end of gnetlist.

>
>>
>>> spNet is designed
>>> for the purpose of spice simulations,
>>
>> Have you tried the spice-sdb back end to gnetlist? It has its quirks,
>> but it's eminently usable. I've designed a 6000 transistor mixed-
>> signal chip using it. Stuart is really nice about fixing problems
>> with it. At the very least, you should understand the existing tools
>> before making something incompatible.
>>
>
> gnetlist regardless of the backend used does not expand subckts
> hierarchically

So what? There's no difficulty in making hierarchical SPICE netlists  
using gnetlist and a makefile. Typically I'm doing other  
manipulations of the SPICE netlists (such as appending files defining  
stimuli from independent sources), so make is the tool of choice.

> and its missing a few other features that industry
> netlisters have and I'm trying to bring this to the gEDA community.
> I'm trying to make this netlister 100% compatible as a drop in to the
> current set of gEDA symbols thus the existence of this thread.

I don't believe that is doable. The gEDA symbol library is an  
extremely mixed bag of things created at different times to support  
different flows. In my opinion, most of the difficulty with spice-sdb  
resulted from Stuart's attempt to make spice modeling using printed  
circuit symbols work transparently.

>
>>> so if people are using discrete
>>> analog components and subckts this shouldn't be an issue. If people
>>> are using spice models of the components they are using this is  
>>> taken
>>> care of with the value field.
>>
>> You should be using the model-name= attribute for that. That's the
>> standard. Don't overload value=
>
> Even gnetlist uses the value= attribute for this.

No it doesn't. gnetlist doesn't do anything at all with value=.  
Specific gnetlist back ends sometimes do specific things with it.  
spice-sdb can, in certain circumstances, use it for the model, but  
generally you should use value= for the value and model= or model- 
name= for the model. See the detailed explanation at http:// 
www.brorson.com/gEDA/SPICE/x150.html.

> Why would I change
> it?

Indeed. But if you don't want to change the convention, you'd better  
understand how things currently work.

> For example, if you netlist a schematic that has hierarchy below a
> cell it will use the value= attribute for the subckt name.

But spice-sdb has used model-name= for this purpose for five years or  
so.

John Doty              Noqsi Aerospace, Ltd.
http://www.noqsi.com/
jpd@xxxxxxxxx




_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user