Eagle is horrible software

[liquibyte]

It sounds more as though it's the choice to use Linux that's giving you problems.

It wouldn't be if Microsoft hadn't illegally dominated the PC market from day one.  More and more people are using Linux and more and more people will be as the support for XP has dropped away.  "If" they ever fix Kicad to the point of being usable, I expect that it may just replace Eagle as the defacto hobbyists software to use for PCB design.

I've been using Linux since '96 exclusively as my OS and telling me that I have to put up with software problems because I'm a Linux user is a shills argument.  Paying $400 for an operating system is why I use Linux, paying $10000 for a software package to design hobby boards is a non starter in my book, hell, paying $100 is too.  There are only two software categories that lack in the open source realm and you'll be especially surprised at what they are.  CAD and simulation.  Want to use a horribly designed CAD program?  Try BRL-CAD, the most overly complicated and unintuitive thing I've ever tried to use.  I wish people would stop bashing Linux because windows is not better, not by any stretch of the imagination.  Eventually all our problems will be solved because that's what Linux users are, problem solvers.

[IanJ]

"If" they ever fix Kicad to the point of being usable, I expect that it may just replace Eagle as the defacto hobbyists software to use for PCB design.

I've been using Linux since '96 exclusively as my OS and telling me that I have to put up with software problems because I'm a Linux user is a shills argument.  Paying $400 for an operating system is why I use Linux, paying $10000 for a software package to design hobby boards is a non starter in my book, hell, paying $100 is too.  There are only two software categories that lack in the open source realm and you'll be especially surprised at what they are.  CAD and simulation.  Want to use a horribly designed CAD program?  Try BRL-CAD, the most overly complicated and unintuitive thing I've ever tried to use.  I wish people would stop bashing Linux because windows is not better, not by any stretch of the imagination.  Eventually all our problems will be solved because that's what Linux users are, problem solvers.

"if they fix Kicad".........the same can be said for Eagle!

A good friend of mine went down the Linux route many, many years ago as I stayed with Windows........and we've had many, many discussion on Linux's evolution. The net result is that both OS's have their good points..........but the bit that wins for me is that it's much more adopted and in terms of apps if you want the best of them then you need to hand over some money to MS. Anyone that goes Linux will compromise on that part.
Yes, I've tried CAD on Linux (I use AutoCad on Win7 daily) and yes it sucked BIG time.
I'd also hazard a guess to say Linux also sucks at Pro audio/music mixing/production apps......I could be wrong here though.....just a guess.

Ian.

[liquibyte]

"if they fix Kicad".........the same can be said for Eagle!

The only thing I truly don't like about Eagle and can't reasonably work around is library and part management.  They seriously need to fix that or I'll never consider a license.  If they did, I might, maybe.

I'd also hazard a guess to say Linux also sucks at Pro audio/music mixing/production apps......I could be wrong here though.....just a guess.

Depends on if you want someone to hold your hand in point and click world or if you're comfortable with the command line.  I'd hazard that there are musicians out there that quite successfully use Linux in post production.  I've never recorded before but occasionally consider trying but my barrier to entry is the hardware itself.  I've successfully done video and can say, at least in an editing context, that Linux runs circles around windows.  I've only messed with the sound apps a little so can't really say for sure how good they are to someone in the pro world.  I've seen the complaints though. 

Realistically, the problem isn't the OS, it's the devs.  If people would start porting or writing for posix then we'd have the apps and this conversation wouldn't have to be had.  The kernel itself has come a long way since I started using it and can be used by someone that knows nothing about "under the hood".  For those of us that do, windows is just not an option even if there's more software out there in Redmond land.

[BloodyCactus]

2D CAD is pretty good, DraftSight runs on linux and its an AutoCAD clone. Renoise is a fantastic DAW. There is no good NLE video program I find (something like sony vegas etc)

Back to Schematics and PCBs...

Right now I'm going to try DipTrace under Wine and VirtualBox and see how it goes.

[T3sl4co1l]

As a linux user, I have no other choice but...

gEDA?

If your answer is anything but, you must be a terrible *nix user

[BloodyCactus]

As a linux user, I have no other choice but...

gEDA?

If your answer is anything but, you must be a terrible *nix user

gEDA is a steaming pile of garbage. I'd rather bang my head against kicad than use the gEDA suite of tools.

[Monkeh]

They say driptrace can run under wine, but I dont trust that kind of crap emulation.

Wine Is Not an Emulator.

[BloodyCactus]

They say driptrace can run under wine, but I dont trust that kind of crap emulation.

Wine Is Not an Emulator.

  thats right, it just approximates. close enough is good enough for wine.

Seems stable so far under wine1.7. I need to go through some tutorials and see how it handles.

[Refrigerator]

What about ExpressPCB ?

[zapta]

Eagle is a horrible software but it's also the best I found for my needs. I first tried diptrace but it was unreliable (e.g. I lost changes when moving from library to the schematic) and had poor library (not even 0.1" headers).  Kicad web site was down for about two weeks (!) and latter when I tried to install it, all I could find for the 'supported' Max OSX was a script to built the binary myself.

Eagle on the other hand worked out of the box, has a rich library, is very stable and is supported by vendors like OshPark and OshStencils.  It's not perfect but works. I was even able to export outlines for laser cutting and the auto router gives me very good results (after routing manually the critical networks).

[BloodyCactus]

I first tried diptrace but it was unreliable (e.g. I lost changes when moving from library to the schematic) and had poor library (not even 0.1" headers).

ha! I  just ran into this testing diptrace. the issue is, the parts exist but no schematic symbol exist. so you have to make the schematic, and just attach the existing footprint.

crazy they have parts that have physical footprint but no schematic drawing!

[tom66]

Adding further to this.

About 2 weeks ago, the class received their boards back.

Here's a favourite pick of a number of errors EAGLE introduced.

One guy had an array of buttons hooked up to his MCU. He drew a button with a pulldown resistor and copied it four times for four buttons.

The small trace between the pulldown and the switch had the same NAME property... and so all his switches were shorted together.

There is NO VISUAL INDICATION in the schematic of this. He did not check the PCB thoroughly, but given this was a first-ever PCB project... how much can you expect?

About 1/10th the class had problems in part caused by EAGLE and its unintuitive interface. As such no actual bugs in the program but unclear user interfaces and odd ways of doing stuff.

[Tac Eht Xilef]

One guy had an array of buttons hooked up to his MCU. He drew a button with a pulldown resistor and copied it four times for four buttons.

The small trace between the pulldown and the switch had the same NAME property... and so all his switches were shorted together.

There is NO VISUAL INDICATION in the schematic of this

AFAIK (though it's been a long time since I've used it), Altium won't give you a visual indication or warning of duplicate net names either. Net labels give an indication - but you can turn them on in Eagle too, and it's still dependent on you noticing the duplicate labels.

He did not check the PCB thoroughly, but given this was a first-ever PCB project... how much can you expect?

There's a good lesson there, and it's not restricted to Eagle: always check the PCB layout - because that's what's getting made, not the schematic.

[kizzap]

He did not check the PCB thoroughly, but given this was a first-ever PCB project... how much can you expect?

There's a good lesson there, and it's not restricted to Eagle: always check the PCB layout - because that's what's getting made, not the schematic.

Hell, it is a good lesson to everyone. If you are sending out gerbers to be manufactured. CHECK THEM. I once got burnt by CAM350 changing the size of some vias on me.

[zapta]

Here's a favourite pick of a number of errors EAGLE introduced.
One guy had an array of buttons hooked up to his MCU. He drew a button with a pulldown resistor and copied it four times for four buttons.
The small trace between the pulldown and the switch had the same NAME property... and so all his switches were shorted together.

That's why I verify manually the net list before starting the layout (print the net list, go one by one, mark on the schematic and make sure all is correct). With waiting time of ~8 days for the PCBs to arrive from manufacturing, I want to reduce the chance of having to redo them.

[Mechatrommer]

some lesson learnt is that 2 persons is better than 1 person. hence a good DRC in SW + a guy manually checking it is better than a guy checking it alone. a good DRC will save alot of time, and time is more expensive than gold.

[Kjelt]

Adding further to this.
One guy had an array of buttons hooked up to his MCU. He drew a button with a pulldown resistor and copied it four times for four buttons.
The small trace between the pulldown and the switch had the same NAME property... and so all his switches were shorted together. 

Well I just tried to reproduce this using v6.4:
- I grouped: a net from a powersupply to a resistor, the resistor, a universal net (named N$2) from the resistor to a led, the led,  and a gnd net from the led to the gnd.
- I copied the group using the copy button and CTRl-right mouse clik
- and the result was that the universal net between the new resistor and the new led is named uniquely to the next open possible value in my case N$8. So NO PROBLEM here

So when does this go wrong? Probably if you (re)name the specific net with some personal or reserved value other then the eagle generic netname.
In that case it might go wrong because eagle assumes that the user specifically WANTS this net to be the same why did he otherwise rename it from the eagle generic netname in the first place?

So if you really want to blame I would suggest differently:
1) the teacher that should supervise the process with students that do this for the first time and do not know better then to screw up , but hey that is called learning, if you never make mistakes what can you learn?
2) the person between the chair and the keyboard.

[zapta]

...
So if you really want to blame I would suggest differently:
1) the teacher that should supervise the process with students that do this for the first time and do not know better then to screw up , but hey that is called learning, if you never make mistakes what can you learn?
2) the person between the chair and the keyboard.

The problem is with the tool. It allows two schematics with different net list to look the same on the screen.

(I am an Eagle user so no hate here).

[Wilksey]

I just did something similar with 6.5 and I custom named the net and when I copied the group it named the net the same but with '1' on the end, and they were not shorted together.

I suspect they may have improved this, what version did your students use?

I don't think there is an excuse for education to use older versions as the lite version is free and students would surely not need more than the board limitation imposed on the lite version?

[Kjelt]

The problem is with the tool. It allows two schematics with different net list to look the same on the screen.
(I am an Eagle user so no hate here).

Unless you label the nets. But how does another tool for instance Altium show this then?

[IanJ]

Guys,

In the release notes for 6.5 (UPDATE_en.txt).

Ian.

Code: [Select]

  - COPY of net or bus wires: Creation of a new segment for the wire copy
    if it's endpoints are apart from the source segment.

[silent]

I have been using Eagle for my projects. Yes, it has its disadvantages, like net naming problems, no differential routing in versions before 6 and so on, but there's also some good sides.
For example the text commands and ULP (it's very convenient to type the command from keyboard instead picking the option from menus) and it's very stable (at least in my case). It didn't crashed (or malfunctioned) even once during last few years.

It's nice for learning (intuitive, IMO) and small things, but for the larger project... yeah, time to learn something new, like Altium. ;-)

[oPossum]

One guy had an array of buttons hooked up to his MCU. He drew a button with a pulldown resistor and copied it four times for four buttons.

The small trace between the pulldown and the switch had the same NAME property... and so all his switches were shorted together.

There is NO VISUAL INDICATION in the schematic of this. He did not check the PCB thoroughly, but given this was a first-ever PCB project... how much can you expect?

There is a visual indication in the schematic. Use the show command (eye button on toolbar) to show everything connected to a net. It will be very clear. Also the airwire in the PCB is clear. This was an error due to inexperience with PCB design, not a problem with Eagle.

About 1/10th the class had problems in part caused by EAGLE and its unintuitive interface. As such no actual bugs in the program but unclear user interfaces and odd ways of doing stuff.

How may would have had problems cause by KiCad?

[Monkeh]

How may would have had problems cause by KiCad?

Surely it'd be easier to count the number of people KiCad didn't crash on?

[larry42]

A small user report here.

I needed to quickly get a PCB made for testing something at the hospital lab. I downloaded DipTrace and within 3 work days I had added a 80pin Samtec QSE connector and a 96pin DIN 41612 connector to the library, done the schematic (~400 pins), routed the 200x80mm 4 layer board by hand, created the Gerbers and sent of to production.

I had to refer to the DipTrace manual / google search about 3-4 times during the process. The DipTrace UI is just really intuitive. I've used Mentor BoardStation and Zuken CR-5000 before, which are industrial grade packages, so I'm fairly au fait with the PCB design process. 

In the meantime - Eagle, which I have been experimenting with for a couple of weeks, requires me to look things up for nearly every operation. I believe Eagle is more powerful than DipTrace, but the Eagle UI is just obtuse and many of the translations don't conform to any of the vocabulary used in the PCB industry (again, I'm looking at this from Mentor / Zuken). Eagle is hard to use, because it's poorly designed and dripping with legacy.

The Eagle autorouter is better than the Diptrace one, but the only reason I use the autorouter in Eagle is because the manual routing UI is just awful (can't move traces whilst maintaining 90, 45deg angles, etc). By contrast the manual routing in DipTrace is pretty good. Not as good as CR-5000 with real DRC and push-aside routing, but much better than Eagle.

I'm seriously considering buying DipTrace license instead of Eagle, especially as Eagle require the Auto-router module to get online DRC during manual routing. I don't want to pay for the autorouter which I'll never use anyway.