I wanted to introduce our startup SnapEDA to the HN community. We recently completed Y Combinator, and have been quiet about the platform while we've been improving it. With that said, we’d love to get feedback from the HN community!
Our goal is to build a canonical library for making circuit boards: one trusted, centralized place to get digital models. These digital models include PCB footprints, schematic symbols, and 3D models. The library exports to a growing set of popular EDA tools: EAGLE, Altium, KiCad, Cadence OrCad/Allegro (Beta), & Mentor PADS (Beta).
The library is free because we believe in making this data widely accessible to enable innovation. The purpose of this new feature, InstaPart, is to give designers an option to "skip the queue" and get a part quickly if it doesn't yet exist in the free library. Once that part is made, it is then made available for the entire community to download for free. Growing the library is a top area of focus, so we hope to eventually render the InstaPart feature obsolete and just have everything available natively. :-)
In terms of standards, all new libraries are being made to IPC, and we also source models by partnering with component manufacturers. To ensure quality, we have an automated verification checker on each part page that provides a pass/fail result on common manufacturing issues that we plan to expand with additional checks.
I've had to use a few parts that claimed to follow a standard footprint, but where subtly off. Another part I've used if you read the datasheet carefully you ended up with a completely different footprint than what first glance would suggest which is my biggest problem with upverters part concierge as they helpfully "fix" my footprint to the incorrect one.
Will we be able to comment on the footprints or leave reviews on how well that footprint worked?
1. You can upvote or downvote the part
2. You can leave a comment on the part
3. We have automated algorithms that uncover common defects in parts on the Validations tab. This is a patent-pending technology we developed that we plan to expand further to include more checks.
Shiiiiiiiiiiit, this is awesome. I spend quite a bit of time trying to find components, and you never know whether the footprint is correct or whether you're in for an unpleasant surprise.
Thanks for this! I'm especially loving the fact that you do KiCAD exports.
Oh wow, I hadn't thought of this, but it's so obvious!
I personally haven't dealt with hardware since college, but if you pull this off, you will be the go to source for every single student working on a project or lab that's wondering why the part they just bought is not working according the datasheet. It would have saved me many hours, and that's just as a student, can't imagine for those that depend on those datasheets professionally.
Wish you the best of luck, this sounds very promising.
"Upvote" should mean more than "I liked it". It's too easy to get likes. Something that says "I used this footprint to make a board, and the part soldered into place OK" would be helpful.
I hope this is constructive, but that chat invitation on every freaking page drove me off the site. By time I'd logged in and searched for two parts, the incessant chirping and popups of Clippy were too much.
Very constructive, thanks. We recently installed the app (Intercom) and it has really helped provide a channel for feedback which we have then used to prioritize improvements. I'll talk to my team about reducing the number of pages the automated pop up shows up on though.
One of the ways I've seen it done, and done it at previous companies is to not present it to every user on page view. Pop it up for only 10%, even if you have a button/link/menu to bring it up for everyone. That way you can solicit feedback while still gaining a lot of the benefits of having live interaction. It'll also still give you some insight to where the chat is most useful, but it'll take longer to gather useful data on what pages it gets used the most.
I'm cool with a cookie and an [X] or a Close button on warnings / notifications / advertisements / popups / music controllers so you know that I have seen it, and do not want it displayed to me again.
I think a callout the first couple of times is fine (still not my preference, but I think this is a cultural thing - the US style "how can I help you" in retail rubs me up the wrong way too).
But I think after that it'd be very much preferable to taper off - especially the sounds. As another commenter said, know your audience - your icon is obvious and descriptive, so we don't need to be told 20 times what it does.
> and verified using proprietary patent-pending verification technology
Is "proprietary patent-pending technology" really necessary? It's the kind of phrasing seen on uninteresting product trying to pretend they're interesting (ex: "Our Proprietary Patent-Pending Technology Makes Our Product So Great This Single Mom Can't Even Talk About It. Discover Now For $99.99")
Here's our motivation for that wording: We know how important verification is and we spent a ton of time building an automated verification system (still in Beta) that provides transparency into common issues with CAD models. It can uncover and display defects, such as the centroid not being at 0,0 which will cause issues with pick-and-place machines, eventually pin mapping issues as well. We were trying to find a way to succinctly communicate this focus & investment into verification technology which differentiates us from any other library in the market. Apparently that was not communicated well and we'll re-word to make this more clear. :-)
Lookw great! I used many footprints from SnapEDA before.
In many cases I prefer to draw the schematic symbol myself though. I like organizing the pins in a way it fits nicely in the schematic. It takes a bit more time to start but it makes the schematic cleaner and easier to maintain.
That's awesome to hear! Would love to chat at some point and learn more about your preferences with symbols. We'd definitely like to see if we could incorporate more user preferences there.
We plan to work with component manufacturers to make sure their parts are well-represented on SnapEDA. We also have Premium versions for users who need more functionality. You can learn more at https://www.snapeda.com/pricing.
"You can sign up for a 10-day free trial of SnapEDA premium. During the trial period, you can cancel any time in your Account Settings. After the trial ends, you will be charged $297 ($99/month, billed quarterly)."
This stinks of dark patterns and maximizing the amount you can hit people who forget to cancel for before they notice. It's rather unlikely that someone would actually want to pay quarterly at the higher monthly rate.
If you don't want to sign up for a subscription, you can purchase individual parts for the $29. Also, our free version is really powerful so if you're not in a rush, you can stick with that one. :-)
That doesn't answer the dark pattern concern at all and you're just dodging the issue. When people hear trial they think "free time to make my decision" not "10 days for $300" which is what this will be for anyone who forgets to cancel (hint: any EE on a deadline but I guess you know that). This is just predatory.
Isn't that standard practice? For example, after the Netflix free 1-month trial, you immediately get charged for the upcoming month. So isn't it up to the customer to make sure he/she doesn't break the terms of the trial i.e. try it for "too long"?
being standard practice doesn't make it any less of a dark pattern. it's just that tons of companies have decided it's fine to cheat users for an extra revenue stream
Exactly how is it cheating? As long as the service clearly states upfront that your card will be charged at the end of your trial, it's the user's decision whether or not to proceed.
I'd argue that it would add some amount of inconvenience to the user, and lower conversion rates as a result, if the user has to pay to continue using the service.
put it this way - if you asked users up-front "when your trial expires, should we convert you to a paying customer and automatically start billing you?", what fraction of them would say "yes"? i'd guess it's pretty low, and most companies know it. the fact that they're deliberately taking advantage of people's tendency to forget to cancel counts as cheating in my book.
Thanks for the feedback. It might make sense to remove the free trial then because the main feature in it is just parts requests which people can now test with this new InstaPart feature.
At $DAYJOB we run a SaaS with a free trial, and immediately start billing for the following month after the trial expires.
However, we don't ask for credit card details until the first payment, so, there's no risk that a user would be charged accidentally. It also reduces onboarding friction as a credit card isn't required for signup. If a user doesn't want to continue after the trial, we advise them to disregard the invoice.
I would suggest you to continue with this.
There is another reason that is done - evaluation is done by engineers and purchase decisions needs finance. What you want to do is move the process upfront where the approvals are in place by the time the trial ends.
This is why this is pretty standard practice - keep sending reminders every week and also the day before you are going to charge.
I (an engineer) have gotten subscriptions for things that would never have gone through if I had to go through and file a report on why this should be bought.
Also, not making a credit card mandatory for sign up will get you non-serious customers. You will need to have infrastructure costs to support a large number of these.
If you have other killer features in the trial then just provide discounts for parts (say $19 during the trial) and remove the dark pattern instead of removing trials all together. I actually do need this service now and $29 per part is a drop in the bucket, especially since a lot of parts I need have nonstandard footprints. I can check and correct a footprint against a datasheet much faster than I can draw it so there's little risk to my deadline but the trial dark pattern is very off putting even if $300 is nothing compared to revenue. I'd love a service like this but it won't save me more than 2-4 days of work during a 2-3 month long contract so soft concerns like this can easily keep me from paying for a product.
From [1]: "A Dark Pattern is a user interface that has been carefully crafted to trick users into doing things, such as buying insurance with their purchase or signing up for recurring bills."
Yes there is trickry. A "Free Trial" does not mean 10 days for $300 which is what it essentially is whenever someone forgets to cancel. This is the same technique slimy online businesses (i.e. credit reporting agencies) use all the time along with making it hard to cancel. They even made it quarterly so that they can make three months off of you instead of one.
If you think I'm being obnoxious you should perhaps reflect on your own behavior. Others in this thread clearly agree that this is a dark pattern.
I think they are wrong just like I think you are. That's no argument.
You're defaming one of the most beautiful services that has ever come out of YC. A service that will help the life of so many hardware hackers and professionals. The pricing model for InstaPart is the most noble I've ever seen and you're complaining about them being predators and dark pattern practioneers?
So you're designing a little board in KiCad and you want to add an LPC1768 to it to do some processing. You need to get a schematic representation and a physical footprint of the part into your CAD software. You could spend an hour carefully reading the datasheet and recreating all of the pins and landing pads, but this is the 21st century! You look at the manufacturer's website, SparkFun's part library, google a bit, and you can't find premade files for the part. So you go on InstaPart and pay $29 and someone at SnapEDA does the work for you and then releases the files for free. Now they're listed on the SnapEDA database for anyone to use, and you know that they're going to be correct.
Just adding a bit to what the others have said, you could almost think of the "parts" in a CAD program as being akin to the templates in a header file for a C library, i.e., they provide a symbolic representation plus an interface definition that are joined together.
In the case of electronic CAD, a "part" is what the part looks like in the schematic diagram, typically a box with numbered and labeled connections, and a representation of its physical dimensions and the location of its connections. Somebody has made sure that the physical details are correct, and that pin 23 on the schematic is pin 23 on the physical board.
Creating these parts is kind of a form of costly drudge work for designers, and errors in those parts can be hard to correct when your prototype doesn't work for some mysterious reason.
Imagine you wanted to build a spaceship out of lego but first needed to mould each block.
If you wanted to build a circuit board, you would need a digital model of each component, and these are often made from scratch because designers can't find one readily available. Likely someone at a nearby company, or even a nearby cubicle, has already made that model. So this wasted effort hinders innovation and is a huge productivity drain.
SnapEDA makes a huge database of these models available that hardware designers can drag-and-drop into their designs.
Is there any way to get support? I've been having problems trying to login or do anything since yesterday, but your support (or contact) email is bouncing...
Ah sorry we switched our email servers yesterday afternoon. Info@snapeda.com is back up and running, or you can also contact us through the chatbot in the lower right corner of the website.
still bouncing... I'll try the chatbot. You might want to remove the email from your page. It's very frustrating to try to contact a company, but all the info on their website points to a non-working email :(
We haven't tested it yet but if Proteus supports P-CAD then you can try downloading an Altium format which we export currently to P-CAD file format. Let us know how it goes.
I'm a professional hardware engineer, and doing this kind of work is something I do frequently and begrudgingly. Having this kind of service available will be a huge help. Some thoughts:
- Who writes the style guide? How do you make aesthetic decisions?
- Will you support multiple symbol styles? Would it be possible to upload stylesheets or specially annotated schematics and then regenerate already-extant parts in that style?
- Is there any intention of making the file writers open-source? Altium, in particular, has a stupid and annoying file format and it would be a gift to the community to be able to write good PcbLib and SchLib files. It would make it easy for me to write a linting and style-casting tool.
- Is there any chance of bringing down the latency, possibly with the application of more money? My rule of thumb is that it's worth spending about a hundred dollars to save myself an hour. A typical smallish library part takes me about five or ten minutes if I have to do both schematic and footprint, so waiting a day isn't really attractive at any price. But if I could throw money at you to get a result turbospeed, that'd be worthwhile.
Re: Style - As a starting point we follow IPC-7351B for footprints and our own internal standard as outlined here: http://snapeda.com/standards. Over time we'd like to allow for more customization to account for different styles and preferences.
Re: Symbol styles - that's really interesting and is an awesome idea. Will think about how we can do this. Definitely have thought about storing user preferences. For example, some people like NC pins on symbols while others like them hidden.
Re: Open Source - Haven't thought about this side yet but will consider it. We have made open our API and maybe we can expand that to include the exporters.
Re: Faster turn - Anything available on the site already is obviously free and instantaneous. I think it's worth exploring with my team if we can do a faster turnaround on InstaPart. We definitely have ideas on how to do this. Our part creation is a mix of automated and manual work. The longest part of the process is verification. But again, hopefully eventually this stuff just all exists readily available without requiring a special request.
I second the speed requirement. The most I've ever spent is an hour on making a schematic symbol and two hours for a footprint, both when I was very inexperienced with Altium and using a ~500 pin part. Now that I'm comfortable with Altium's IPC templates and batch editing features (you can pull up a copy pastable spreadsheet that allows you to quickly edit all shapes on a footprint without the UI), even the most complex parts take under a half hour. $29 is nothing for 20 min of work but depending on your workflow, waiting a day is unacceptable. I can rarely predict the kind of part swapping that will happen before PCB layout so footprints are the first step that has to happen in the first few days after schematic capture. Since the footprints are a blocking concern for part placement, I don't have much patience especially on a deadline since I can just draw all the parts myself while waiting for the first footprint to come in.
The other problem is that the vast majority of professional engineers check the footprints meticulously before fabrication even if they made the footprint for a previously successful project. It only takes a day or two and must be done regardless so it removes a lot of the added value InstaPart can provide.
Although most of this doesn't apply to the more hobbyist market, they generally aren't willing to pay $29 for anything let alone a single footprint.
I'm surprised this is a retail business. Doing this in conjunction with Digi-Key, Mouser, or Hamilton/Avnet would seem to make more sense. They already distribute the data sheets for the parts they sell. Also, it would be far cheaper to make footprints if you did them in bulk, working your way though long catalogs of similar parts.
Maybe the retail thing is to get some credibility so they can deal with the major distributors.
We'd love to work with distributors to help proliferate this content. And yes internally we prioritize development of content in bulk. But sometimes there's a model that is needed that is outside of our development efforts. This is where InstaPart fits in. It's an option to 'skip the queue' and get any model quickly if it's outside of the models we are prioritizing internally.
That is the process, yes, but it's not really that lengthy. You can usually just add the first pin and then copy/paste the rest in the row really quickly.
Yeah, for the most part library stuff is quick. Some tools let you pattern place the pins and plop them all down at once. High pin count parts like FPGAs and connectors with weird footprints are the big time sinks.
Weird footprints are really the only time consuming part with modern EDA software. Tools like Altium have templates, footprint generators, and features for bulk editing symbols and footprints so all you do is create a quick Excel spreadsheet with pin number and formulas for x and y (especially easy for FPGAs and processors with pins in a grid). You copy paste that spreadsheet into Altium and you're done. It takes me less than 10 minutes to manually draw a footprint for a 500+ pin processor this way.
Yeah, when I was doing library stuff I used Mentor Expedition which has all kinds of stuff for speeding up footprint creation. Symbols can be time consuming for the various XXGAs because some of our customers have fairly... elaborate standards for visual appearance, pin locations, etc.
Bookmarked this site. The database seems super useful.
I have mixed feelings about the custom footprint service -- I've worked with ~900-pin BGA SoC-type parts, and paying $30 to do that would be a no brainer, but paying the same for a 8-pin LDO would be a tougher sell -- maybe scale prices with part complexity?
The tougher sell to me is trust / verification of the InstaPart models before the community can vote them up or down. For most teams that I've been on, the most time-consuming part isn't really the pinout generation, but rather the checking of large parts (often 2 engineers checking pin-by-pin to ensure that footprint matches data sheet). I'd be much more comfortable using it if you outlined how pinouts are verified before sending them out to the customers.
I hope this works out! Making footprints is a huge PITA, and I'd love to be rid of it.
Agreed on the verification aspect. And what happens when an error makes it through and your customer has to scrap $xK worth of boards?
I work at a $largeco with dedicated footprints people and a long DFM pipeline, and footprint errors still get through from time to time. When there's tens of thousands of dollars riding on a board turn, you're still going to check it yourself.
Yes, that's a very good point. Most of our users are professionals like you mentioned so I think this definitely merits design changes. Also, regarding iPhone, we aren't yet compatible on mobile (most of our users visit from their desktops where they download the files), but it should at least still be usable, so thanks again for letting us know. :-)
I think your footprints should be shown with dimensions, because before I drop hundreds or thousands of dollars on a run of PCBs, I'm going to have to check your footprints to make sure they are correct, and that could actually take me longer than drawing them from scratch.
As a designer, this looks like a great convenience since making a symbol and footprint usually takes a half hour on average. The 3D feature is the most useful part as a proper model takes much longer but at $79 dollars it could be expensive for people like me who design boards with many ICs and and unique components.
With that said, in a very time constrained project with a lot of new components which we have no symbols for, if I could select and buy everything I need in a packaged deal, that would be appealing.
Good to know. You can actually download unlimited 3D models with the Premium version ($99/month). It is custom 3D models that cost $79 for 24-hour turnaround, and $39 for 5-day turnaround. (Although symbols & footprints available on the site are free, 3D models are currently Premium only.)
This is actually quite useful, as an occasional user of a variety of layout packages, once in a while run I into that one rare part which isn't in a library, and end up wasting an hour finding the specs/measuring the part and figuring out how the component editor works.
There's the whole section at the top about availability and average price, but no link to go buy it? In addition to the InstaPart revenue, are you also going to make money with affiliate sales to parts sites (a la Octopart)?
If you go to the Pricing tab you'll see links to buy the parts through popular North American and European distributors. This data is supplied via Octopart.
Ah, I completely didn't see that tab... Then perhaps some usability feedback; maybe create links of the already orange text in the upper left hand box (the "Average Price") and have it auto-choose the Pricing tab below and scroll the page there? Make it super easy to find :)
Is this where people should go to make a production level PCB's for products that are prototyped on something like a raspberry pi? Or is it serving a different market?
Yes, exactly. It's for people designing custom PCBs. That includes people who have something prototyped on a Raspberry Pi/Arduino, etc. and want to migrate to a custom board that's ready for production (ie. to get cost, form factor down).
Interesting. Is there a guide for someone who's not a PCB designer to get his raspi project as a custom board ?
I want to make a raspi remote controller that controls IR devices around my house via a web app. This would be a perfect, but I have no idea how to get it into a PCB design.
Why? Making schematic symbols and PCB footprints is not time consuming, and at least you'll have the proper paste mask. Altium has an IPC footprint generator wizard, and 3Dcontentcentral has lots of user contributed parts in STEP format.
I was just making some yesterday. It's time consuming if you use KiCad, which has a draw program with a UI from the 1990s. I could have drawn a footprint in Autodesk Inventor's sketch mode in a tenth of the time, and without having to use a calculator to compute distances from the data sheet drawing.
(CAD program UIs are really hard to do well. The big companies have it figured out at last, but the technology hasn't filtered down to open source yet.)
The pre-canned libraries will never have the typical customization needed for PCB.
The paste mask aperatures need sizing depending on what stencil thickness you use.
Need vias under the QFN belly pad for RF or heat sinking purposes. Do those need tenting on the back side? They also may need to avoid the paste aperatures?
How about solder mask relief? Copper defined pads or mask defined pads?
Are the part centroids on the correct layer? Also the keep out regions, and 3D model?
What I'm getting at is the footprints are more than just copper pads; that's the easy part. They will still require lots of customization, so you are better of just doing them yourself and building up your own library.
Now having correct schematic symbol pinouts for that 1024 ball BGA. That is useful.
PCB library vendors have come and gone over the 20+ years I have been an EE.
Also, as someone else mentioned, the schematic symbols are defined by if you have a logical (digital) or physical (RF) schematic. For example, pin spacing should be such that bypass caps can be squeezed between. Even data sheets have multiple symbols for the same part in the same data sheet; Linear Technology is a good example of this.
Anyone doing PCB professionally will make their own symbols and footprints. A hobbyist using free layout software won't pay $29 for a footprint.
Seems like an interesting service. I'm kind of confused by the site though.
On the landing page it says, "Get any schematic symbol and PCB footprint delivered in 24 hours. Just $29." Then I did a search for a part and clicked request and, after signing up for an account, it said that to get it in 24 hours I need to pay $79 and $29 was for 5 days service. I also somehow ended up on a page at one point that said that you could request any part for free. So which is it really?
I also found the social network aspect of the site off putting particularly since there was no mention of that and then after I signed up for an account to give the request a part service a try I see my user name plastered on the sites front page in a feed of recently signed up users and deactivating my account doesn't remove that.
I think you must have been on the 3D model tab, which cost more than the 2D. (The 2D models are $29 for InstaPart.) And if the parts are already available they're free. We'll figure out how to make this more clear... Thanks for the feedback
As for the social media aspect, good to hear that as well. I talk with my team and see if we can add an option to suppress that, or re-design that page.
Anyone who cant spend a few minutes in their EDA tool of choice to create a symbol or footprint and instead chooses to pay $29 here should not have a job (or a hobby) making electronics.
"Individual CAD symbols and footprints on SnapEDA are licensed with a Creative Commons Attribution-ShareAlike 4.0 International License (CC BY-SA) with the Design Exception 1.0. This license allows designers to use CAD files downloaded from SnapEDA for commercial and personal use free of charge, and make derivative works. If CAD files, or derivatives thereof, are shared publicly, attribution must be given to the source (SnapEDA and the original author, as applicable), as outlined by the requirements of the license.
Use of the SnapEDA website is governed by the Web Site Terms and Conditions of Use.
In short, you cannot use the SnapEDA website for the purpose of sharing the content on the Internet or within an EDA tool, but otherwise you are free to use the files for designs."
That's too restrictive for something I'm paying for. Particularly at 29USD a footprint.
I'd love it if there was an open option here. Particularly if it integrated with Kicad well and I could automatically upload my footprints.
This terms of use was originally written for our free library. Basically the way the license works is that as soon as you use a model in your designs (Design Exception), there is no license; it becomes your IP. You can re-post the models online in small quantities so long as you provide attribution to the source (SnapEDA). For InstaPart, we haven't re-evaluated the license, but we're open to it if you have better ideas. All our InstaPart users are professionals who are simply using the models in their designs (ie. not re-posting to the Internet), and therefore the attribution is not an issue. The models immediately become part of their IP. Basically the motivation behind our license was to be as flexible as possible without encouraging people to scrape and re-post vast quantities of the library without attribution.
We get Fritzing requests occasionally so would like to include it eventually, but don't have a timeline for it right now. Basically we're prioritizing formats based on demand so if there's enough demand from Fritzing users, we'll speed it up.
I wanted to introduce our startup SnapEDA to the HN community. We recently completed Y Combinator, and have been quiet about the platform while we've been improving it. With that said, we’d love to get feedback from the HN community!
Our goal is to build a canonical library for making circuit boards: one trusted, centralized place to get digital models. These digital models include PCB footprints, schematic symbols, and 3D models. The library exports to a growing set of popular EDA tools: EAGLE, Altium, KiCad, Cadence OrCad/Allegro (Beta), & Mentor PADS (Beta).
The library is free because we believe in making this data widely accessible to enable innovation. The purpose of this new feature, InstaPart, is to give designers an option to "skip the queue" and get a part quickly if it doesn't yet exist in the free library. Once that part is made, it is then made available for the entire community to download for free. Growing the library is a top area of focus, so we hope to eventually render the InstaPart feature obsolete and just have everything available natively. :-)
In terms of standards, all new libraries are being made to IPC, and we also source models by partnering with component manufacturers. To ensure quality, we have an automated verification checker on each part page that provides a pass/fail result on common manufacturing issues that we plan to expand with additional checks.
Thanks HN!
Natasha