I posted last month about our startup SnapEDA which makes free symbols and footprints available in multiple design formats when we released InstaPart: https://news.ycombinator.com/item?id=12571200.
We just built new plugins so that you can search SnapEDA right inside your design tool and thought I'd post again to share it (hope that's ok with everyone). These new plugins are for Altium and Eagle. The plugins allow you to search models without having to leave your design environment.
This is a beta, so would love any feedback you might have. One thing on our roadmap is to add cloud-syncing of the libraries and user authentication right from the plugin. Would love to hear your thoughts on other features to add next!
I want to personally thank you! This I feel will quickly become a very useful tool for me when I design projects for Burning Man. The libraries for Eagle can be very very lacking.
I feel like the KiCad support is lacking. As the community EDA of choice, support for it is quite important.
What I would really like to so is a repository of 3D models. Footprints are a dime a dozen and faster to churn out than 3D models. Also, most EDA suites support creating footprints/symbols but few/none support 3D modeling.
Also I've had SnapEDA provide me with footprints of similar parts that in actuality were not similar enough. A fair chunk of time was wasted on that.
Hi Robert, thanks for this feedback. We'd definitely like to add a KiCad plugin next. About 10% of our users use KiCad so it's definitely a community we want to focus on.
Over the past few months, we've made a lot of changes to how parts are linked. In addition to no longer showing related parts (this is where a lot of confusion was), we have also added notifications when a part hasn't yet been validated. As a startup, our team is constantly improving this too!
In terms of 3D models, you can toggle to the 3D model tab to see an example of the 3D models we're starting to incorporate. Example here: https://www.snapeda.com/parts/ATMEGA328P-AU/Atmel/view-part/ (toggle to the 3D view). These are currently a premium feature, but our goal is to make these also free at some point too.
Since you guys are doing things with 3D models, I would love to throw out my 2c. My team (flair.co) sorta splits electrical and mechanical design along people lines (even though we are comfortable with both sides). I don't end up doing much board layout but in terms of vetting a relay or some Caps for fit, being able to drop them into my mechanical design would be great. My flow right now is, go to the internet (if in the states) or sometimes the market (if in shenzhen), measure the parts with calipers or look at the data sheet, build a model in Fusion360 and then import that part into my design. If you made a plugin for Fusion360 and perhaps a few of the other popular mech e packages(solidworks, onshape, etc.) that would be fantastic!
Why do you want 3D models? What is the use case? Are you doing a lot of really space-constrained boards?
KiCad has spent a lot of effort on making it possible to view 3D models of boards. Upverter has recently spent a whole bunch of time on making this available (time that could be better spent on a lot of other things). And I don't get why. Outside of a small contingent of space-constrained parts, I just don't see what the use of 3D parts is.
I've personally spotted many mistakes in the 3d view, mainly spacing ("oh, I thought this would fit but it doesn't look like it would"). It's not my first priority, but it would be nice to have.
An awful lot of boards have to fit into a small gap in a machine somewhere, or parts are being stacked (maybe it's under the brick power supply), and you need to be sure everything fits.
we make devices (flair.co) and regularly ran into making our boards and our mech-e components aligned/fit etc. we used eagle and altium and fusion and especially without 3d parts it was a lot of extra lifting on our part to model everything and check for fit/alignment etc. it is fairly contrained in terms of size
Sigh. I don't need yet another plugin for my generic parts--that you're going to try to monetize on me eventually if you achieve lock in.
Let me be blunt. I don't need extensions or features to my PCB tools. Certainly not "Yet Another Database Frontend(tm) You Can Rent".
I my tools to fscking WORK.
I need a tool that runs, fast, on Windows, Linux, and even OS X.
I need bug fixes for the features that they actually have.
I need a tool that lets me script things in a modern programming language.
I need an autorouter that doesn't suck. This will never happen because our algorithms don't like to work with PCB limitations of small numbers of layers and obstacles that penetrate multiple layers.
ALL the PCB software tools are buggy piles of garbage because the amount of money in that industry is now fixed. Charitably: everybody is trying to extract rent instead of innovate. Uncharitably: nobody actually in development actually knows how the board designers use their tools anymore.
If you want to go chase a field, go chase Cadence, Mentor and Synopsys up into the real EDA market where algorithms matter. Creating $10000-level tools that can do actual VLSI design to take advantage of the fact that transistors are stupidly cheap in volume would kick off the custom silicon revolution everybody has been promising.
Actually our parts library is free and will always be free for designers. As an EE, I'm committed to this. We launched with this vision in October '13 and haven't wavered. Tens of thousands of hardware designers already use it, so if we wanted to switch to that approach, we would have done it by now. We do have Premium features for things that we can't give away for free (ex: on-demand requests for things like complicated connectors) but otherwise we have not wavered.
Please read the Show HN guidelines: https://news.ycombinator.com/showhn.html. They say: "Be respectful. Anyone sharing work is making a contribution, however modest."
Your comment is a mixture of dismissiveness and helpful (though perhaps tangential) information. It would be much better if you took out the dismissiveness.
Sorry, you are correct. My post was far too dismissive. He accidentally tripped over a hot button of mine and it was not his fault. Mea culpa.
As a power user and as an executive manager, I am extremely frustrated to see yet another company creating a "part database" (we have more than enough of these, thanks, and they don't actually solve a problem that a user actually has) rather than tackling ANY of the actual technical problems in the space.
I understand your frustration having used EDA software, and worked in software development on these tools. However, we set out to build an interoperable content repository rather than yet another EDA tool.
First, with regards to not needing another parts database: when I was leading the marketing of an EDA tool, designers would often tell me that they didn't need yet another new feature in the tools - they needed content so they could actually be productive with the tools (rather than spending hours making the content before they could even start designing).
I would hear this over and over when I went to technical events, tradeshows, through our feedback systems, etc. If you search HackerNews, you'll also see many people citing the same need (top comment here): https://news.ycombinator.com/item?id=8329373
Existing parts solutions in the space are of dubious quality (lack of standards, quality issues) and are often only solving bits and pieces. Designers would tell me they didn't trust them at all, so they would end up making their own models from scratch. This is one of the reasons large companies hire whole teams and dedicated librarians to create CAD models.
We thought what if we could provide this data more systematically, in a way that brought standardization, interoperability, & transparency to design content, that we could help the entire industry, especially small to mid-size companies that don't have dedicated librarians.
This content today is symbols, footprints, and 3D models, but we plan to expand to other forms of content, such as SPICE models and eventually even larger interoperable reference designs built on vetted, trusted data.
> Please don't assume the OP is male or not solving technical challenges. She's a YC founder recently profiled in a technical magazine (PCB Design - see: http://www.iconnect007.com/index.php/article/100587/).
1) Press coverage does not automatically imply either vision or competency. See: Theranos.
2) Do not assume that I do not have the bonafides to critique properly just because I am willing to have my rudeness pointed out. I've probably been fighting with EDA tools longer than she has been alive. And I ran a group which had a budget of more than $70 million for EDA tools.
I assure you I have very strong opinions about EDA tools that are backed up by very hard won experience.
And one of the many of those opinions is that not once did any engineer with experience ever say "Gee, we really need somebody else to manage our parts or footprints." TO A PERSON they all said "We'll do it. Anybody else will screw it up."
So, quite certainly, I have enough vision of the field to deem it "not a technical challenge". You can feel free to disagree, but you're going to have to bring some experience or data to convince me.
And, in case you think I'm being unfair, I can assure you that I ripped several layers of upper managers at Altium a new one for buying Octopart instead of fixing their tool.
I agree, actually - we have every solved problem, solved 3 or 4 times (to varying degrees of quality), with zero interoperability and no vision. It's fun, yeah, but it actually makes sense to pay for a 10k tool if it does what OrCad, Altium or Mentor can do.
I figured the fastest way to solve it was to:
1) Get ridiculously rich
2) Donate 20 Million dollars to a foundation to build open source CAD tools of professional quality.
I think your InstaPart feature looks cool. Lots of big companies have CAD teams that just make component libraries for designers. It is nice to get a bit of that on the small side of the industry finally.
That being said, limiting it to 5/month is very constraining. Can you accrue InstaParts? I think most designers typically need more than that per project, but not necessarily each month.
Sure. The way it works is if a part is already in our library it's free and you can download unlimited models. If not you can get it in 24h with InstaPart, and then that model is included for everyone in the free library.
You are correct that we have heard from customers who subscribe that they want more than 5 (right now they don't accrue because it's an add on to the other features). We are considering making these special requests available via a credit system. Would something like that work? Approximately how many do you use per project and what is the complexity of the designs?
23 comments
[ 2.3 ms ] story [ 62.3 ms ] threadI posted last month about our startup SnapEDA which makes free symbols and footprints available in multiple design formats when we released InstaPart: https://news.ycombinator.com/item?id=12571200.
We just built new plugins so that you can search SnapEDA right inside your design tool and thought I'd post again to share it (hope that's ok with everyone). These new plugins are for Altium and Eagle. The plugins allow you to search models without having to leave your design environment.
This is a beta, so would love any feedback you might have. One thing on our roadmap is to add cloud-syncing of the libraries and user authentication right from the plugin. Would love to hear your thoughts on other features to add next!
What I would really like to so is a repository of 3D models. Footprints are a dime a dozen and faster to churn out than 3D models. Also, most EDA suites support creating footprints/symbols but few/none support 3D modeling.
Also I've had SnapEDA provide me with footprints of similar parts that in actuality were not similar enough. A fair chunk of time was wasted on that.
Over the past few months, we've made a lot of changes to how parts are linked. In addition to no longer showing related parts (this is where a lot of confusion was), we have also added notifications when a part hasn't yet been validated. As a startup, our team is constantly improving this too!
In terms of 3D models, you can toggle to the 3D model tab to see an example of the 3D models we're starting to incorporate. Example here: https://www.snapeda.com/parts/ATMEGA328P-AU/Atmel/view-part/ (toggle to the 3D view). These are currently a premium feature, but our goal is to make these also free at some point too.
Actually, most footprint maker plugins now happily generate 3D models for you. They're almost always good enough.
Why do you want 3D models? What is the use case? Are you doing a lot of really space-constrained boards?
KiCad has spent a lot of effort on making it possible to view 3D models of boards. Upverter has recently spent a whole bunch of time on making this available (time that could be better spent on a lot of other things). And I don't get why. Outside of a small contingent of space-constrained parts, I just don't see what the use of 3D parts is.
It is especially nice with the ray traced 3D view of KiCad.
Let me be blunt. I don't need extensions or features to my PCB tools. Certainly not "Yet Another Database Frontend(tm) You Can Rent".
I my tools to fscking WORK.
I need a tool that runs, fast, on Windows, Linux, and even OS X.
I need bug fixes for the features that they actually have.
I need a tool that lets me script things in a modern programming language.
I need an autorouter that doesn't suck. This will never happen because our algorithms don't like to work with PCB limitations of small numbers of layers and obstacles that penetrate multiple layers.
ALL the PCB software tools are buggy piles of garbage because the amount of money in that industry is now fixed. Charitably: everybody is trying to extract rent instead of innovate. Uncharitably: nobody actually in development actually knows how the board designers use their tools anymore.
If you want to go chase a field, go chase Cadence, Mentor and Synopsys up into the real EDA market where algorithms matter. Creating $10000-level tools that can do actual VLSI design to take advantage of the fact that transistors are stupidly cheap in volume would kick off the custom silicon revolution everybody has been promising.
Your comment is a mixture of dismissiveness and helpful (though perhaps tangential) information. It would be much better if you took out the dismissiveness.
As a power user and as an executive manager, I am extremely frustrated to see yet another company creating a "part database" (we have more than enough of these, thanks, and they don't actually solve a problem that a user actually has) rather than tackling ANY of the actual technical problems in the space.
However, your point is well taken. My apologies.
First, with regards to not needing another parts database: when I was leading the marketing of an EDA tool, designers would often tell me that they didn't need yet another new feature in the tools - they needed content so they could actually be productive with the tools (rather than spending hours making the content before they could even start designing).
I would hear this over and over when I went to technical events, tradeshows, through our feedback systems, etc. If you search HackerNews, you'll also see many people citing the same need (top comment here): https://news.ycombinator.com/item?id=8329373
Existing parts solutions in the space are of dubious quality (lack of standards, quality issues) and are often only solving bits and pieces. Designers would tell me they didn't trust them at all, so they would end up making their own models from scratch. This is one of the reasons large companies hire whole teams and dedicated librarians to create CAD models.
We thought what if we could provide this data more systematically, in a way that brought standardization, interoperability, & transparency to design content, that we could help the entire industry, especially small to mid-size companies that don't have dedicated librarians.
This content today is symbols, footprints, and 3D models, but we plan to expand to other forms of content, such as SPICE models and eventually even larger interoperable reference designs built on vetted, trusted data.
1) Press coverage does not automatically imply either vision or competency. See: Theranos.
2) Do not assume that I do not have the bonafides to critique properly just because I am willing to have my rudeness pointed out. I've probably been fighting with EDA tools longer than she has been alive. And I ran a group which had a budget of more than $70 million for EDA tools.
I assure you I have very strong opinions about EDA tools that are backed up by very hard won experience.
And one of the many of those opinions is that not once did any engineer with experience ever say "Gee, we really need somebody else to manage our parts or footprints." TO A PERSON they all said "We'll do it. Anybody else will screw it up."
So, quite certainly, I have enough vision of the field to deem it "not a technical challenge". You can feel free to disagree, but you're going to have to bring some experience or data to convince me.
And, in case you think I'm being unfair, I can assure you that I ripped several layers of upper managers at Altium a new one for buying Octopart instead of fixing their tool.
I figured the fastest way to solve it was to: 1) Get ridiculously rich 2) Donate 20 Million dollars to a foundation to build open source CAD tools of professional quality.
I think your InstaPart feature looks cool. Lots of big companies have CAD teams that just make component libraries for designers. It is nice to get a bit of that on the small side of the industry finally.
That being said, limiting it to 5/month is very constraining. Can you accrue InstaParts? I think most designers typically need more than that per project, but not necessarily each month.
You are correct that we have heard from customers who subscribe that they want more than 5 (right now they don't accrue because it's an add on to the other features). We are considering making these special requests available via a credit system. Would something like that work? Approximately how many do you use per project and what is the complexity of the designs?