AVR Freaks

There are plenty of choices, it basically comes down to how much you want to pay. If you want something free, then go with Eagle. The limitations of the free version (boards size, two layers) probably don't bother you. The functionality is fine, there is plenty of forum support and one can get used to the somewhat antiquated user interface.

Every once in a while I try out a couple of others, but came always back to Eagle...

If you want to pay you'll have to name a budget for us to recommend specific packages at your price point.

Markus

http://www.bartels.de/bae/bae_en...

is AFAIK also for linux.

I'm also a BAE user (see daqq's posting above).

Things I love with BAE:

• well-thought structure of the layout parts, hierarchically structured; seems cumbersome at first but eventually turns out
  to be really flexible
  
• hierarchical schematics, schematics can consist of an infinite number of sheets
  
• well-thought schematics parts, good optical feedback about whether a connection to a part has been made or not (back in my
  Eagle days, I've been trapped by unconnected parts that looked
  as if they were connected several times)
  
• the size of the routing grid does not affect the performance, gridless routing is possible -- with today's multiple different
  part grid sizes (mil-based for traditional parts, 0.8/0.65/0.5
  mm for modern SMDs) this is really important
  
• a decent autorouter; I don't route the digital parts of my layouts manually anymore (only the analog parts); not only that
  the autorouter does the job pretty fast, it's also mentally much
  easier to re-arrange things later on, and just have the autorouter
  re-layout it again; might not look as aesthetically pleasant as
  a manual routing but it works pretty well
  
• certain things look complicated at first but turn out to be really useful over time; e.g. rather than touching a layout
  item with the mouse, and dragging it, you first select what kind
  of item you are going to handle; that way, you can clearly
  distinguish between moving a via, a complete part, an edge of
  a trace, a segment of copper, or parts of copper or documentation
  areas
  
• library elements that are used are always copied into the current project; the back-reference to the library is maintained (so you
  could e.g. update it in case you've really got a more recent version
  of the library) but you can always modify the project's copy of the
  element without affecting anything else but the current project; in
  the end, this model allows each project to also serve as a library
  for other projects so if you designed a certain schematics and layout
  part just in one project, you can later easily re-use it elsewhere
  
• much of the application is written in a (compiled, C-like) user language and can thus be modified by the user once you grasped
  the language (which takes years, admitted -- but it also allows the
  supporter to just send you an updated ULC implementing your wish)
  
• very good support even for the low-cost version (BAE Light)

What I don't like:

• the menues look chaotic at first, "historically grown"; they can be user-customized though
  
• the autorouter is a separate application rather than fully integrated into the layout editor, so you have to flip back and
  forth between both
  
• the color handling is quite old-stylish (16 colors which can be RGB-tuned, but you don't have more than those) but then again, the
  drawing speed resulting out of that is pretty good
  

Are you using the correct version of the Motif libraries? ISTR they
might perhaps be linked against a version that is not identical to
the one shipping with the most recent versions of the various Linux
distributions. ldd tells me it's looking for libXm.so.2, whereas
libXm.so.3 appears to be the current one.

If you've got troubles finding a copy of libXm.so.2, send me an email,
and I'll mail you that library. I'm not sure whether it's a LessTiff
or OpenMotif one, but both are free now so it shouldn't matter.

I still like Wintek's HiWire II, costing $900.00. See:
http://www.wintek.com/cadmain.html

I also have Wintek's SmartWork, costing $900.00, back when I bought it in around 1980. I think it cost about half that now. The issue with SmartWork is that the grid only resolves down to 0.05". The Auto-router was another $1,000.00, so I didn't buy it. But if you are strictly working with 0.05" & 0.1" pitch, this is by far the easiest PCB layout software that I've ever used - but it's only PCB layout and not schematic capture.

I have been using HiWire II for the past 20 years. HiWire II is both schematic capture and PCB layout software. There is also an auto-router at the cost of $900.00. I also did not purchase the auto-router for Hi Wire II - it's just too expensive for what it does.

It's hard to find something to replace HiWire II. Though, I have recently purchased the Eagle Professional version and am currently trying to adapt my way of thinking to it. I have completed three small projects using Eagle. It'll take a while before I fully make the transition - but that is where I think I'm headed.

I have the evaluation copy of Dip-Trace, as well. I might play with it and see how it performs. I hear it cost about $150.00 for the full (read, legal) version.

I have tried Mental-Automation, or what ever they call themselves, at the cost of $250.00. The software is not too bad to use, except I had to make my own libraries. What killed my interest in this software was that, the libraries seem to need to live on the computer they were created. The libraries that I made won't work on any other machine.

And then I spent about $300.00 spent on Proteus Lite. I couldn't even get a PCB laid out with it. The frustration level was that I couldn't get the traces to behave when I moved a component around on the PCB. The whole thing turned into a mess.

I've also tried several other "Freebie " Schematic Capture & PCB lay-out packages - without success. The frustration level was just too high!

I think that the phrase "You get what you pay for! " definitely applies to schematic capture & PCB layout software - at least from my meager experience.

I think I'll stick with HiWire II and SmartWork, and just transition into Eagle!!! HiWire II and Eagle are both feature rich and provide a means to output to my mill - SmartWork does not. They both (HiWire II & Eagle) allow inverse printing. I.E. printing negatives. And, I can do mechanical etching, if I choose with both HiWire II and Eagle. They both will create standard Gerber files, allowing for professional board manufacturing, if I want. And, they are all bought and paid for.

So, by today's hobbyist standards, I've spent a fair amount of money (at least $3,500.00) on schematic capture & PCB lay-out software packages over the years. and, I've tried many other schematic capture & PCB layout packages that I can't even remember the name of. To say the least, it's been a fairly expensive learning curve over the years.

Unless something comes along that really revolutionizes the schematic capture/PCB lay-out process, it's HiWre II and Eagle for this tinkerer!!!

I do a lot of small R&D boards. About twenty a year. Originally I used Protel 98 but the company would not payout for the jump to Protel DXP.
So now I use Eagle and for simple R&D (SMD etc) it is quite acceptable.
I would not go back to Protel in a fit, the fact I can't convert old Protel files to Eagle is a pain.
I use Eagle for hobby work as well and it is a bit disjointed but for the price who can complain.

I'd love to have a go at some of the full on systems but sadly don't have the opertunity or cash.

It's all down to need and economics. But I do spend too much time getting the layout just right, that is a real time waster. But I regard it as similar to a game of logic and I hate links (zero Ohm resistors) I think they are cheating. my work mates make a lot of jokes about my need to remove them.

> ... the fact I can't convert old Protel files to Eagle is a pain.

But that complaint would go to Eagle, wouldn't it?

I recently re-engineered a Protel layout by reading in the Gerber files
generated from it. Except that BAE had some issues with “negative”
parts in the Gerbers, it's been a fairly quick job to let it generate
its own layout from just the Gerbers.

Of course, that doesn't give you the schematics back. But I don't think
there's any two EDA tools that could fully convert the projects of one
tool into their own one.

There should be a standard format for devices. A small xml-file to go with the data sheet which describes the physical and pin properties of a chip or device. The chip manufacturers would supply this file and the PCB/Schematics programs would allow importing it into their libraries.

Unfortunately I'm dreaming...

Markus

That would be gEDA/PCB.

Just don't underestimate the time, energy, and algorithmical knowledge
it takes to create a good autorouter. Everything else is relatively
simple compared to that part.

I did consider purchasing PADS and HyperLynx, but found I could not justify the price for the amount of PCB's I am designing. I found the signal integrity modeling and automated termination recommendations of HyperLynx to be very cool.

Thanks for the feedback, Carl -- the dongle is hassle, parallel-port based even more so. I didn't realize HiWire II wasn't keeping up with technology changes. Yes, I think Eagle (with all its warts) hits a certain sweet spot, especially with their freee version to get people started using it.

Oh, they (Wintek) do offer a Windows extended version of the HiWire II schematic capture and PCB layout software. But they want $1,600.00 or $1,700.00 for it. And an additions $16 or $17 hundred dollars for the auto-router.

I'm not really interested in another DOS patch!

Yes, yes! I've heard that Eagle is a kluge of DOS conglomerations too! But, at least there are hundreds, if not thousands of individuals or companies using Eagle - I'm not alone anymore. I at least have compatibility with part of the world. And most important, I have compatibility most of the AVRFreaks community.

Pulsonix is intended for professional users. They also have Easy-PC, a less-complex package for hobbyists and students:

http://www.numberone.com

It's much easier to use than Eagle and costs about the same. I used it for over 15 years until Pulsonix came along.

Leon

If you download the demo you can try importing designs for yourself:

http://www.pulsonix.com

I've done it once or twice with OrCAD schematics, it worked OK. If you want to try Pulsonix properly, they will give you a full 30 day license.

Leon

Tom Hausherr and the group at PCBLibraries.com have built a very nice and very complete library set, but it's quite expensive. And I don't think that this is the way to go.

Given the practically universal connectivity these days, the manufacturers should all be using a standard extensible XML format to describe their parts, much like the analog guys to with spice models. They should be available at a standard location within their company parts website, so that any EDA package can reach them.

The XML files would contain a complete description of the part. Everything from the schematic symbol(s), spice models if applicable, footprint(s), 3D models, variations, the works. To avoid data duplication, a lot of this could be references to other chunks of data - I'm thinking footprints mainly here, but can be anything.

This way you don't need a large library of parts locally that may or may not be accurate. Ever been burned by a stock library footprint that was wrong ? Or wrong schematic symbol connectivity ?

Browse to the vendor/manufacturer website, click click click on the parts you need (AVRs for example), or their libraries of passives or common parts. Those XML files come down and the EDA software translates the well-known and standard XML schema to whatever internal mechanism it wants. Internal reference links are resolved and the refs are downloaded as well. It would contain non-default-download references too, like datasheets and Design Notes - those could be pulled down at will.

This puts the onus of accuracy on the manufacturer, and you can bet they will be sure that those files are correct and accurate ... Much more incentive there than the EDA vendor. Dated updates can be posted on their site, and the software can check periodically - "ah, ATMega64 has been updated, download and update ?"

It really wouldn't be hard ... A basic XML schema to cover the basics would be the start - schematic symbols, connectivity, footprints. The hard part is getting the manufacturers on board, but once an Analog or Linear does, the others would come along I expect. Features can be added easily - it's XML - the EDA software just ignores any elements it doesn't know/care about.

They used to have prices on their web site, I'll ask them about it.

IIRC, it starts at $2,000 for 1000 pins and four layers, and goes up to $15,000 for an unlimited version with all the options.

Leon

The Pulsonix one is very good, it's actually Electra. Things like ADCs should be routed manually.

Leon

bphillips wrote:

> Even the autorouter is not that important. I have been designing
> PCBs for over 25 years and have never used the autoroute feature of
> any board program. Since 99.9% of my boards involve analog circuits,
> I've just never felt comfortable letting an autorouter handle the
> design.

Well, it's not important whether they *involve* analog circuits, but
rather whether there's a substantial amount of circuitry *besides* the
analog part.

For plain analog designs, there's probably still nothing that would
beat a human. However, the ability to just layout the analog part of
your design manually within a day or two, and then tell the autorouter
to run the other 50 % (the digital part) within 5 minutes is
something I wouldn't want to miss anymore. The big thing here (from
my experience) isn't even the ratio between another day of manual
layout vs. 5 minutes of autorouting, but rather the ability to
eventually go ahead and change the entire digital part after noticing
the typical 5-to-12 mistake, and then simply have the autorouter
re-run the connections within another 5 minutes. When doing manual
routing, you'd for sure do *everything* to avoid the step of
re-layouting after recognizing the mistake.