CircuitPeople

Call it News or Call it a Blog, This is What's Happening at CircuitPeople

Twitter Updates

Follow me on Twitter

Engineering Software Companies going "To the Cloud"?

This morning I read something on Jim Quanci's blog Dances with Elephants. Jim is from Autodesk, one of the big names in design and engineering software and probably one of the largest vendors of desktop software around. His blog illuminates the acquisition "dance" from the buyer's perspective and is an interesting and recommend read for budding entrepenuers. This morning's article talks about how companies currently building desktop software should be approaching the cloud, and at one point asks "Still Have Doubts?" Well, yes. Doubts are healthy and rational, and I think it makes sense to consider some of the risks in going "to the cloud". I've been developing systems with substantial "cloud" components for about a decade (in fact it's what I left Autodesk to do), and I have never shaken the doubts.

First, from an economic point of view, it seems there is sublime shift in burden when moving desktop ultity (what the software does) to the cloud. The current model has the customer paying for computation (the price of their desktop PCs and internal servers) as well as the software. With cloud services the customer sees only the cost of the software (service), with the vendor accepting and bundling-in the cost of computation. The net outlay by the customer may be the same, but there is significant risk that the vendor will not be able to recover that cost. The common assumption being that the "multi-tenancy" of cloud offering allows the vendor to achieve inefficiencies by re-using and tuning compute capacity based on actual demand.  But that isn't automatic.  In practice the customer needs to be prepared to pay more *to the vendor* for the same utility plus covering the computation (at least a portion). This is unlikely to happen.

And that is my second concern. The unfortunate fact is that in the world of mobile and cloud software there is an expectation of paying less (or even nothing) for utility. The perception of value has been altered so substantially by the $.99 app and the "free to use" start-up[1] that a critical mass of customer may no longer be willing to support the shifted costs. For engineering applications where substantial computation is needed, this is a particularly precarious risk. Even more unfortunately, as marketplaces proliferate and desktop software delivery becomes more 'app store'-like (see Windows 8) the value perception of desktop software is likely to decline as well. If you want to know what it's like to compete in a a 500,000 or million strong app-store, find an ISV that's been there and ask. It isn't pretty. Competing with free, low-cost "simple" apps is the norm. Offering high-value products in the same space is extremely difficult. Hence the vicious circle creating more and more simple apps.

My third concern (I'll stop here) is the brand risk is offering free or low-cost services is substantial. Marketing probably isn't my greatest strength, but I strongly aware of the importance and value in maintaining a trusted premium brand. Once eroded, it is very rarely restored. While it may feel good to have thousands or millions of downloads of a free app, if the perception of value (or worse, quality) is low there is inescapable damage to the brand.  Any MBA program's library of case studies can provide decades of cautionary examples.

Yes, the cloud is awesome-sauce. It is a place where fortunes have an will continue to be made. However, to Jim's article's FUD point on risk, ISV's should go in with knowledge that their current business model will not work in the cloud. Re-think everything you do in that context and approach it for how it helps create opportunities for new ways to do things that are even more valuable than the old ways. Forget about the simple apps, and focus on *systems* with high utility that are only possible where you span the desktop, mobile and cloud access and infrastructure. In short, learn to love the web but keep the desktop.

That's what your kids want.

[1] Yes, I know I appear hypocritical here since circuitpeople.com is free. To make it worse, I've offered very low cost software in the mobile space since the 90's, too. On the other hand, it does perhaps allow me some authority to speak to the point.

We need a Packaging Convention for PCB fabrication that works

I've written a longer post on why I think the time has come for creating a convention-based packaging solution for PCB fabrication data. IPC has their solution on the table, and I think it is wrong-headed. So here is my proposal for open packaging conventions for gerber files

Also, if you are a PCB123 user and Arduino fan you might be interested this Arduino Shield Template for PCB123.

Where does CircuitPeople.com go from here?

I want to cover a couple ideas in this post, both on the theme of *you*. Creative, inventive people are the core of what made this site interesting to me, and what keeps me motivated to improve and promote it. If I could only have helped one person get something built faster, better or cheaper it would have been worthwhile. It's done much more than that and I really couldn't be more pleased about it. But more on that later.

User Voice

Before going into how the site's done, I want to start talking and thinking about what it *will* do. My vision for this site is to enable what I call "collaborative design review". That's a little too grand of a term because I'm hyper-focused on the manufacturing phase of PCB design. I want to make it much, much easier -- even the "default" -- to get peer review on designs before sending them to manufacture. Maybe those peers will be from the internet at large (for an open hardware project, for example) or perhaps they will be your coworkers, contractors or suppliers in a secure private review. Whatever the case, that kind of review should happen and receive the benefit of the best advice available to increase correctness, cost and reliability of your printed circuit boards.

Unfortunately, I'm not an EE. I tinker with electronics, and with PCB design. I know my way around tools like EAGLE, and helped build another PCB design tool (I'm a software guy, clearly). So I need *your help* and your suggestions. To help out with that, I've created a place to record them over at http://circuitpeople.uservoice.com. If you have suggestions, bugs, comments or any other feedback for us, and you don't mind it being public, please head over there and let us know. The UserVoice system lets you add suggestions, vote for ones you like, and find out if we start working on (and maybe even complete!) your suggestion. It's a great tool, and I look forward to seeing suggestions come it (I'm a little cynical, so I also half-expect to see few if any -- :) ).

Update on our Results

Now to look back a bit. At the time I created this site I wanted to give the community a way to create really good looking renderings of their gerber files without having to pay for or register some commercial software. I think the site does a great job of this, and apparently so did a *lot* of other people. Since its inception, more that 85K people have used the site and that number is growing faster than ever before (have the other gerber views gotten worse?). More than 60% are from outside the US, representing every continent (I can't actually confirm the Arctic -- anyone know what IP addresses are there?) and 159 countries. Hundreds of thousands of images have been created for you. I couldn't be more pleased. And, that level of positive response is why I'm committed to making the site better and better for you -- so share your suggestions!

And, never fear, my committment to leave this site ad-free and free-to-use is as strong as ever. My strongest "Thanks" to everyone for using circuitpeople.com!

Embed our Gerber Viewer on *your* Website, Blog, or whatnaught.

I would like to see a lot more PCB designs being shared around the interwebs, where they can inform and educate in a free and open way. Don't get me wrong, I all for keeping some designs close to the chest for commercial reasons. But there is a lot of room for sharing -- better sharing -- in this industry. So, the latest feature we're working on here allows you to embed the code for any set of gerbers you upload into other web pages. Here is an example (the FEZ Domino board):

For each of your uploaded packages you will find the embedding code down at the bottom of the ViewPackage page, along side the sharing by email option.

I would really like to see this integrated with PCB discussion forum software and websites. So talk to us if you have such a site and are interested in allowing your visitors to visualize gerber files.

Enjoy!

Is this the "death" of the PCB auto-router?

Probably not. Auto-routing is a key, sometimes critical, feature of PCB design tools. Routing itself is such a intrinsic part of the PCB design process that it is an assumption. A PCB *is* the routing, to a large extent. So much so that many still refer to them by the traditional name "printed wiring board". Wiring, as in wires, as in linear traces from point-A to point-B.

Well, here is a completely different take on how to get electrons moving from place to place, from pin to pin on your next PCB. The critical requirement is, of course, that they connect to the same peice of copper. The copper doesn't need to be linear, or have pads. Power planes (copper ours), for example, are often odd-shapes.

Marsette Vona, recently used the idea of Voroni diagrams to create "least effort" isolation paths for milling PCBs from copper clad board. His project, Visolate, takes Gerber files as input and generates CNC milling codes. Here is what that transformation looks like for the gerber files for the Ardupilot board:

Ardupilot PCB Gerber File Rendering Ardupilot PCB Gerber File Rendering

Original gerber on the left, and Visolate cutting paths (voroni diagram) on the right.  The result is not traditional looking but, erroneous handling of "stroked" copper pour aside, it is a functional PCB and, with the proper solder mask, solderable. The point here is that the effort to mill that PCB is far, far (like an order of magnitude) less effort than to isolate each of the traces as drawn in the gerber files. Less effort means less cost and less environmental impact (though that needs to be balanced with the greater consumption of copper, assuming the copper removed in traditional process is reclaimed and recycled).

While Visolate works from Gerber files, which means routing already happened. But it could work equally well from just the pin positions and rats nest (aka. net list). The advantage would be eliminating the need for routing entirely. From a practical standpoint, that would be hard to generalize and make work for complex designs. But, most designs aren't complex. For the home prototyper, or PCB fabricator looking for a less expensive way to serve the hobby market, this seems like a nice place to start.

Ordering PCBs without ever generating Gerber Files? Now, that would be intersting!

P.S.: My great sympathies to Marsette and his struggles with the "undocumented details" of the Gerber file format as expressed in the acknowledgements. If you're working on Gerber File parsing and rendering solutions, please give us a shout.  We're here @circuitpeople to help.

Fun with Words

This is a word cloud (Wordle, more specifically) created from the text of this blog. It's a fun bit of visualization, and confirmation that I'm staying true to the topic:

Wordle: CircuitPeople.com Blog

Now can feel better about diverging?

The Joy of Aperture Files (or, why RS-274D should die)

Gerber files provide endless "fun" for PCB fabricators, assemblers and the like. Saavy CircuitPeople.com users may have found some of your Gerber RS-274D files failed to display on our site. Not many people use the older "D" files (less than 10% of the uploads here, overall) but I still feel it is important to fix problems when I find them. Here is an example of one recent oddity in an aperture file:
Photo-Plotter Apertures Report
==============================
Position  Width  Hgt/ID  Shape    Qty
========  =====  ======  =====    ===
   11       10       0    LINE
   12       12       0    LINE
   
   ...
   
   81       45       0    RND      40
   98      130      30    ANNL      4
   107      210       0    RND      32

See the problem? Aperture 98 caused the aperture file parser to crash. A basic assumption I made with these old ".rep" files is that they only define apertures from the "CROP" set (Circle, Rectangle, Obround and Polygon). I was wrong.

I can't be sure what the ANNL line stands for, though it's probably an "annular ring" shape with outer size 130 mils and inner hole size 30 mils. The circle shape in the Gerber format is perfectly capable of handling that, so I implemented the fix for this easily and have it up on the site already.

Normally I wouldn't bother writing about this kind of little bug but this one just perfectly illustrates the problem with "Gerber" (I use quotes around the term because these aperture files really have nothing to do with the Gerber format proper). Aperture files were invented to compensate for the lack of such information in RS-274D. Many, many little variations exist from different CAD systems on the content of these files depending on which CAD or CAM system created them. Supporting Gerber RS-274D is something we're pretty proud of because of problems like this, and we're glad to do it. But it sure would be nice if we could all just move on over to Gerber RS-274X with it's superior (though still flawed) information content.

And in that spirit we've introduced another new capability to the email auto-responder: conversion of Gerber RS-274D and aperture files to standalone Gerber RS-274X files. This service is free and no-strings-attached, so give it a try by sending the old files to gerber.images@circuitpeople.com.  Our robot will convert the "D" files to "X" and respond with an attached Zip file containing the new files, the old files and images of each layer.

Finally, we appreciate all the goodwill and positive feedback (as well as the constructive criticism) we get from our users.  Thank you!

Bad News and Good News

This weekend CircuitPeople.com was moved from one hosting provider to another. The site was relocated from a dedicated peice of "iron" in a rack in a datacenter to a virtual server hosted in Amazon's Elastic Computing Cloud. Unfortunately the move didn't go without a hitch, and for a short while (a week or so) some things may not be quite "right" -- just let us know if you see something odd.

The good news about moving to EC2 that we will be able to dramatically improve performance.  Over the past two years, the number of gerber files rendered by the site has grown dramatically. Our original design was simple, and served very well.  However, we can see a time in the near future when things will start to break and are working to rebuild the "inner" workings to allow for continued growth.  Along the way we're planning on adding some useful new capabilities that will make the site even more valuable to PCB engineers and designers. We've added a twitter feed to the top of this page to make following our progress easier (if you're into that).

PCB Panel Math and a Pre-Manufacturing Checklist

It always makes sense to get the most for your money, but perhaps a little more so recently. To help out, we've posted a new article that goes into how your PCB design fits onto panels and why you care.  It's a short-and-to-the-point look at an often overlooked cost optimization for custom PCB manufacturing.  At the bottom of the article you'll find a visual calculator showing just how much panel utilization (and cost) can vary from one board size to the next.  Use it to see if you're getting the most for your money.

Moving from design to fabrication (ordering boards) for the first time can seem daunting. It really isn't rocket science, and so we've put together a basic checklist of things you need to provide to your manufacturer of choice. It isn't a complete list, but we're hoping that some kind folks will help us fill in the blanks until we have somewhat of a tutorial for those new to the PCB design ranks. If you feel like helping, drop us an email.

What a Season

This has been a difficult Fall.  An economic fall.  A fall from power.  It's difficult to grasp the reality of so many trillions of dollars in wealth evaporating over the course of a few short weeks.  Was it real money or not?  It appears painfully real for many, many people.  I cringe to think of retirement accounts dropped in half, meaning more years of work for people who have long since earned a rest.  Ultimately, I worry that the incentive to fix our broken system evaporated when 700 billion dollars were allocated to "solve" it.  Thankfully I don't hold the power to change it, and like 300 million others I will make the best of it.

From the Mailbag: Why Gerber Files? Why Online?

Not that we get a lot of email with questions (EE's are a self-sufficient bunch), but it's been asked often enough I thought we should answer it for everyone here on the blog. Unfortunately I went a little overboard and ended-up with a few pages worth of text on the background of the gerber format. So, please take a second to read What is a Gerber File and let me know how I did.  The bottom line:  Gerbers are important because they form the de-facto communication line between PCB designers and PCB fabricators. Providing this rendering service online was simply a way to make it available to the biggest audience -- Linux, Windows and MacOS are supported simply because we're on the web and attend to browser compatability.

On another note, we were really happy to be featured on the Make Blog today -- and equally glad our server and software handled the load without blinking. It's verification that our software components are robust and reliable even in high-traffic environments.

Getting Make'd is a Good Thing

Sure, I've heard of the slashdot effect, but I've been Make'd. Welcome to all the visitors from Make Magazine's Blog today. I hope you find our little service useful!

External Apertures

There are two type of gerber files in the world: X and D. It's very easy to tell the difference between the types, and here is how to do it:

  1. Open the file in notepad, wordpad, MS Word or any other text editor (even Internet Explorer or Firefox)
  2. If you see a lot of percent signs, it's probably Gerber 274-X
  3. If you see no percent signs, it's probably Gerber 274-D

But what difference does it make? To PCB fabrication folks it makes a huge difference. The stuff between the percent signs in a Gerber 274-X file defines the apertures, polarity and other critical information about where the copper needs to be on your circuit board. A Gerber 274-D file doesn't include that information, so you need to supply at least one other file to give the manufacturer what they need to understand your board. The aperture file usually has a file name that ends in ".apr" or ".rep", but the format of the file will vary depending on which CAD system was used to create it. There might be one aperture file per gerber file, or one aperture file for all gerbers -- either case will work -- but they must be supplied to make a complete set of information (plus drill data, outlines, etc.)

Here at CircuitPeople we need the aperture files, too. And as of today we added a message to the ViewPackage page that tells the story:

Apertures not included in Gerber 274-D files

We can't render a Gerber 274-D file by itself. That means if your CAD system is stil outputting Gerber 274-D, you will need to put the Gerbers and apertures into one zip file and upload them as one package. We simply can't render the gerber without the apertures. My other recommendation for those still outputting Gerber 274-D files is "stop." If at all possible, change the settings on your CAD system to include the apertures in the gerber files.

Upgrades and Enhancements

We're happy to announce some upgrades to the CircuitPeople site today:

Better Gerber Rendering: We continue to refine our rendering technology based on issues identified by users.  If you've experienced problems viewing files, it might be time to try those troublesome files once more.

This Blog: Since you're reading this, the existence of the blog should be self-evident (I'm read, therefore I am). We'll be posting tidbits of information that we find interesting, and hopefully you will find useful.  One of the topics we'll be addressing soon is who we are, and why we're building this website. 

Accounts and File Storage (Beta): Up until now your use of www.circuitpeople.com has been completely annonymous. And, if you'd like, it can stay that way. However, we also now allow you to create an account and have your uploaded design packages saved on our servers. Sign-up for an account on the registration page.

Spread the Word: We aren't advertising CircuitPeople.com, so we count on old-fashioned "word of mouth" to reach new customers.  If someone you know might find our site interesting, please consider letting them know by signing-up for an account and using the Spread the Word form to send them a note.

Sharing: Finally, with an account you also get access to the "Sharing" ability of the site. When logged-in, you'll find a place to share your design at the bottom of the ViewPackage page. Just type in their email address and click "Share". The person getting the email will be able to simply click on a link and view the images from your design. We plan on ehnacing this sharing capability substantially as time goes by.

Thank you for using CircuitPeople.com!

Embedded Systems Conference 2008

ESC is a fantastic place to mingle with designers and engineers from all corners of the EDA and geographic world.  The theme of the keynote was bit retrospective (this was the 20th show) but also raised the question "What is an embedded system?"  A great question, but a quick look around the room had me thinking "Who's an embedded engineer?"  Over the course of three days I had the great pleasure of chatting with people from software, PCB, FPGA, ASIC, millitary and toy backgrounds (not to mention many folks designing PCs and associated components).  This show attracts a really diverse and exciting crowd -- if you get a chance to attend next year's show, go!

And now back to our regularly scheduled programming...

What is a Gerber file, and why does it matter?

Let's cut to the chase: In a practical sense, Gerber files only matter because they matter. It's kind of like Paris Hilton being famous for being famous. Once upon a time, or maybe twice, there was a little company making machines to automatically expose photographic film to light in a controlled manner. This was a great boon to PCB designers because until this new kind of machine appeared they had been drawing their circuits by hand (the lucky designers had the help of rolls of various widths of black tape.) These "photoplotters" needed a control language, so the people building them took from the popular machine control language of the time, RS274, and modified it to suit their needs.  The result was the informal specification we in the PCB world calls "Gerber RS-274D".  It's evolved a little in the "Gerber RS-274X" variation, but the heart of the format remains.

Fast-forward to today.  Modern PCB manufacturing hardware rasterizes (i.e. make pictures from) the input files to expose the film using precision lasers to draw.  These machines can (in theory) do their work given any sort of appropriately scaled image file.  So why throw gerber files at them?  Because we always have.  In reality, Encapsulated Postscript would make a better format to send to these machines.  But until every part of the PCB fabricator's workflow systems can support EPS, we're stuck with Gerber.  It kind of sucks, because nearly identical machines are used in the printing industry and sold for a fraction of the price of the gerber-enabled machines for the PCB industry.

So, Gerber is basically a 20-year-old machine control language that is being used as an image format; and that explains a lot about its many quirks.

Step and Repeat and Pre-Panelization

One common use of gerber file step and repeat commands (G36 and G37) is to create multiple instances of a board design on a single sheet, probably with scoring or perforation for later separation of the individual boards. Why? Usually to speed the assembly and testing processes. Loading and unloading boards into stencil printers, pick and place machines, solder machines, probe testers, etc. can consume a considerable amount of time. By putting multiple boards through those same processes on a single panel, a lot of time can be saved overall. This is a good thing.

On the down side, pre-panelizing your boards when going to a high-mix fabrication shop can increase their manufacturing costs.  How? Well, for small quantities the fabricator will want to mix your boards on panels with other people's boards to achieve the highest possible utilization of the available area. Fabricators like to work with uniform size panels because it allows them to minimize loading and fixturing costs. When "mixing and matching" boards onto panels, it's much easier to fill all the space with small boards. Large boards can leave awkward spaces that are difficult to fill (imagine a board that fills 100% of the panel height, but only 51% of the width).

Oddly, it can be cheaper to pre-panelize when ordering via some fabricators. Nonetheless, If you're going to pre-panelize small quantity orders, please keep in mind that it may lead to greater waste. Generally, by contacting your circuit board fabrication house you can negotiate the same price for individual boards as pre-panelized PCBs (it's probably better for them anyway). Something to consider.

And, if you do pre-panelize please make sure your entire data set is correct. Here at CircuitPeople we see various combinations of gerbers for copper, silk, mask and stencils that are out of sync. It's very common for the drill (Excellon) and mechanical files to be out of sync. For those of you creating completely accurate sets of files for your manufacturers -- kudos!

One use I'd like to see for G36/G37 step and repeat in gerber files? BGA land patterns and fanouts. With high-pin-count BGAs, it can save a huge amount of processing and overhead to define the pads and via constructs once, and then pattern them 1000 times.  Here's a sample:

1024-Pin BGA Landing Pad

Not bad for less than 20 gerber commands, eh?

APEX 2008 and EDA Mashups?

I'm just back from APEX, and what a great show it was this year -- I felt a real sense of change and optimism on the exposition floor. Some long-awaited technologies are arriving in production-ready form, promising to make it easier and faster to innovate in the electronics space. By APEX 2008 I expect lead-times for PCB fabrication and assembly will be down by a day or more. That's really, really cool.

While new hardware is cool (and absolutely critical to a thriving industry) perhaps just as interesting is a new sense of "openness" being whispered. And I mean "openness" in a variety of admirable forms: openness as in honesty; openness as in the willingness to share; openness as in inclusion; and openness as in collaboration.

One beachhead of this "new openness" is Open Source software thriving in the embedded systems space, and appearing in products big and small. Another beachhead appears to be forming among some EDA "content providers": EAGLE is going to support XML for version 6; Sunstone Circuits is talking about "barrier free" collaboration and has opened their proprietary CAD system with a open-source SDK; RoHS compliance information, part availability and pricing, component libraries -- they're all out there for sharing.

So what is this building toward? IMHO, we're very close to seeing an explosive event around "EDA Mashups". The web introduced the concept of a mashup when Google's map technology hit the scene. Other so-called "Web 2.0" sites built on the concept to create networks of interrelated applets (aka. gadgets). If you are interested in this kind of innovation through integration and need a little more background, you can get a good head start reading about Mashups at Wikipedia.

If (when) shared EDA content is made available as web services I will incorporate it at CircuitPeople.  Maybe someone will start incorporating CircuitPeople's viewing and rendering into their site.  Where will it lead? I have no idea, but it sure is going to be interesting...

APERTURE.LST, Etc.

I just wanted to put a note to everyone that although you may not see substantial changes to the site, constant work is going on behind the scene to make our Gerber file rendering the best it can be. Every day since Feb 29th we have improved something about the site. Some of the changes have been small (like supporting APERTURE.LST files, which are slightly different than most other aperture files) but there are larger changes in the works as well.

Some Stats from GA

Just an interesting note from the Google Analytics data for the past seven days:

  • We've processed more than 1,000 packages of gerber files. Wow!
  • Just shy of 150 people used the service.
  • Those requests originated from 35 countries.
  • And 22 languages.
  • Surprisingly, almost 50% of requests to the site were made from Firefox -- much higher than "the conventional numbers". 

Part of the reason for the large number of package uploads is that some folks are uploading individual gerber files (as many as a dozen) instead of zipping them together and uploading them all at once.  There is a problem with sending Gerber 274-D files one-by-one -- we can't render the gerber without the apertures.  Gerber 274-X files stand by themselves because of the embedded aperture information, but if you're going to upload the "old style" Gerber 274-D files, please put them into a zip file (e.g.: choose "Send To" and them "Compressed Folder" in the Windows Explorer context menu for the folder containing the files) with the corresponding aperture files.

We have some other work to do based on the above observations: We must start working on Portuguese and Spanish translations of the site;  And, our Firefox compatibility must stay at a level matching or exceeding Internet Explorer.

Finally, we're still waiting for some gerber files using knockouts.  Frankly the gerber format specification just doesn't go into enough detail on this feature and we'd like to have some real-world examples to guide our implementation.

Where'd We Go?

For some reason Google has dropped our site from their index.  This is interesting because the automated report from them indicated our site is "404" -- can't be found.  Obviously a bug on their part.  The most frustrating thing about this situation is that it's nearly impossible to contact anyone at Google about it.  My little enterprise is just one of several million on the web, but I want to make sure it's as easy as possible to find for the people that need it.

Google Analytics

As of today, we're starting to track usage of the site with the infamous Google Analytics.  Why? It looks like this is going to be a fairly popular service, and I'd like to understand some basics about who is using the site.  GA doesn't capture information about individuals, it only reports bulk statistics.  Two questions I hope it will answer quickly are:

  1. Should we invest in translating the site to other languages?
  2. How many browser types do we need to test compatibility with?
  3. How high is the resolution of most people's monitors?

Sure we could make assumptions about these things, or use the "commonly accepted guidelines", but we'd rather get the data straight from the source -- the people using our site. 

Incremental Measurements

That didn't take long.  Within three days of opening the site we received the first set of Gerber files that were encoded with incremental units (something we didn't support until today).  Fixed.  That leaves two remaining holes for us, one of which is Knockouts -- we don't support them at all.  If you have some gerber files that use KO parameters, please send them our way.

Welcome to CP

Seriously, welcome.  I'm really glad you're here reading this, and I'll be just ecstatic if you upload a few gerber files and provide CP with a little feedback about the results. 

Copyright © 2008-2014 CircuitPeople LLC, All rights reserved.  CircuitPeople™, Got Gerbers?™ and TrueGerber™ are trademarks of CircuitPeople LLC.