The Grid, a factory-scale robot in Massachusetts that builds PCBs on demand
circuithub.com
circuithub.com
A domestic PCB manufacturing pipeline, has been a pipe dream for me an many other Hardware startup EEs for a long time. Until costs are more competitive, CN still gets the business. Hope you can bring costs for small batch prototype PCB manufacturing down further in the future, then you will have our business!
Does that include the cost of parts? One of the issues I had with sourcing prototype PCBA's in the US is parts are much more expensive (not 10x, but also not 1.3x).
I just had a board (PCB only, no PCB Assembly) made at JLCPCB for $62.91. That's 10 bare boards and a stencil, including shipping and tarrifs. This is a fairly simple 4 layer board, 50mm x 50mm.
I uploaded the same project to your site and I'm getting a quote for $620.26. That's after marking everything as DNP, so it's also just a bare board. No stencil I guess.
To be fair, the JLC board took two-ish weeks to get here. But even with the fastest JLC option (2-3 days) and fast shipping (2-4 days), I don't see how $620.26 is competitive. Am I misunderstanding?
Update: And holy moly. I just noticed the $620 quote is for "any origin", and if I want "USA only" it's $1,697.00!
Currently, all assembly is done in house (in the US) but bare PCBs typically are still made in CN. We used to produce everything overseas and just do the QA/test in the US but during the chip shortage I ended up setting up the production line to attempt to eliminate any possible shortages/tariffs.
It is a difficult struggle manufacturing commercial products in the US (doubly so with an openhardware project) since the final cost is the major blocker but I decided it was worth it for many reasons. Big issue though is knowing that even if I pay someone 2x the going rate for an employee it still will not be a living wage. Not sure the fix there. I see you are planning to lean heavily on automation/AI for this, will be watching to see how it all progresses.
There are still some major blockers in the US. Commercial/Industrial real estate being a big one. Costs are not going down...even if employee count is minimized.
I wish nothing but the best for the project.
Making boards for the Smoothieware/Smoothieboard project. [1]
Mostly small SMT stuff on there. 0402, BGA, etc.
Being mindful of wages is great, but eventually you have to accept that creating a job at a less than ideal wage is better than not creating any job at all. If the pay is not enough then you’ll get the signal through being unable to hire because workers are taking better paying jobs.
An unintuitive feature of the job market is that creating more low paying jobs actually puts upward pressure on wages. Ideally the low end of the job market should be saturated with job opportunities, which creates a situation where employers have to compete with each other to raise wages to find and retain employees, which means pay goes up for everyone.
So don’t let worries about not being able to pay as much as you’d like get in the way of paying anyone at all. The alternative (less jobs, more outsourcing) is worse in every way.
For the employee, maybe. For the employer, it depends on how much time/money you have to put into training and what turnover is. Paying non-living wages leads to high(er) turnover because people will leave for anything better.
Money is not unlimited though so “just pay everyone a living wage” isn’t a real solution. If you can’t afford to pay ideal “living wage” (which is an undefined target that keeps moving depending on who you ask) then your options are to either not create the job at all, or outsource the work.
So therefore, creating a job at any wage is better for your economy than not creating the job at all.
"Living wage" is an awful term because it intrinsically holds employers responsible for real estate costs, which are strongly correlated with higher pay (read: as employees get paid more, housing costs rise to match it).
$20 an hour is not awful pay when you can rent for $600/mo. So the actual fix is to just build more housing, and it will bring back the viability of all these disappearing jobs.
They sure have a lot of that capital.
"Living wage" in my area is $40/hr. That is just to live, not thrive or gain.
Rent prices have gone insane [1], which in turn have drove up COGS. Since COGS have gone up the real estate investors have raised their prices on rents/housing to make up for the increase. The feedback loop has gone on too long...I am concerned the only option is collapse and for a lot of investors to lose money. This will feedback into the system even more...it will not benefit either the renter or the investor. As they say..."the market can stay irrational longer than you can stay solvent".
I run a very small company that is largely bootstrapped. I fix all my own machines, do my own setup, move my own equipment. Personally, I work on every step from design through packaging (janitor>CEO). There is not enough money to pay a living wage for others...so I end up doing everything myself. 70hr weeks are not uncommon. (I also cook my own meals, fix my own car and bake my own bread...otherwise I would not make it)
If you have not been a renter for the last 10 years you may not have noticed. Building more is not helping since the only things being built are done so they can set them to the new "normal" price. It is to the point where they consider an "affordable home" as $400k (in my area they literally named this development as "Miracle on Britton"). Meanwhile the average renter makes <$40k/year.
I have not been outsourcing more due to this. Just working harder myself...but even so I barely get by. 5 years ago I gave up the hope of ever buying in this area [2], currently I am just working to stay here long enough to take care of family. If I did not have that I would be long gone.
Doing what I can to bring back manufacturing to the area, but I am a small fish working mostly in Opensource Hardware. If the project starts making money again very likely it will be cloned to the point where there will be no way to make profit (already happened once).
[1] https://www.zillow.com/rental-manager/market-trends/coeur-d-... [2] https://www.zillow.com/home-values/136166/coeur-dalene-id/
The fix is you will just have to buy American and pay higher prices for higher "living wages". There's no way around it besides taxes or tariffs on foreign goods to equalize the wage/price difference (which does have some down sides).
I buy American and western-made products as much as possible, and I pay higher prices (assuming high quality) for those higher wages. If you want low cost, you can't by definition support "living wages".
There's no "oh if only the CEO makes less" way around it either. Labor is simply more expensive in the west and will remain so and you can vote with your wallet, as I do, or you can pay people less than living wages.
And how many people will be willing to pay a premium when the locally made products are worse quality than foreign products? It's happening right now with Chinese consumer goods (e.g. robotic vacuum cleaners), EVs, etc. Eventually there's a high possibility you won't be able to buy anything in the product category you want.
No clue, but there are millions. Just depends on the product.
Here are two examples. One made in France, the other in America:
[1] https://www.debuyer-usa.com
One of those pans is made in France (western) the other is made here in the US. The quality of the US pans are a bit better which is part of why they command the higher prices. But you can also buy a pan from Wal-Mart for 1/10 or 1/5th the cost. But that's up to you. You can support higher wages or you don't have to if you don't want to.
> And how many people will be willing to pay a premium when the locally made products are worse quality than foreign products?
Why are you assuming the American and locally made products are worse than foreign products? Sometimes they are, sometimes they aren't.
> Eventually there's a high possibility you won't be able to buy anything in the product category you want.
Well for a lot of these we can just do without, but those are the kinds of things you have to figure out if you want "living wages". Wages are higher in the west so we will automate things to keep some prices low/lower and we'll just pay higher costs to support the living wages. Or we won't and then we won't have living wages. You can do things like try to raise the minimum wage or w/e but then you're just making American/western products less competitive with the cheaper foreign products. Then you'll have to implement tariffs and agree with Trump there in order to offset the wage differences.
China already figured this out, mind you. Their wages have been exploding as the country has developed, and due to their horrible one-child policy they are facing a giant population cliff. They have been pushing very hard to switch their factories from cheap labor to automation, and those dirt-cheap Chinese EVs our Western politicians are working so hard to block in order to keep the horribly outdated Western car manufacturers alive are the direct result of that.
Curious, why are surface mount USB-C connectors not able to be done as part of the surface mount components (treated as a through hole I think)? Too fiddly?
The app feels very vibe-coded, and gives impression it's not well tested and polished for a new user/project.
I appreciate PCBs aren't the usual the consumer-level type of user, but I felt confused by a lot compared to JLCPCB. Hope that helps.
Also, one question to pick your team’s brain: What’s your team’s AI PCB workflow and design tips to work with your production line? Any specific tips for ESP32-related integrations?
"Never put any friction between your customer and giving you money"
Dude, absolutely get rid of the requirement to make an account to get a quote. Understandable(ish) to need one to check-out, but hell "Check-out as guest" was probably one of the most profitable innovations in e-commerce ever.
A longer list of accounts is not worth more than a longer list of orders.
What's preventing a competitor from creating an account?
It's industry standard to accept gerbers and BOMs. Now that I've reviewed their legal agreements, using them is a hard no right now.
Now there's a login before I can even figure out if it's a quote or a cart.
Byeeee.
I won't complain about the price, I understand it's going to be more expensive than China. But it's a tough one for me. I have a budget that I have to work with, and while I have some flexibility 3X prices (your site priced $9k compared to JLC's $2k, even after tariffs and shipping) make the project nonviable for me. I hope you can continue to improve pricing over time as you scale.
Also your parts selection interface needs a lot of work. I have a very simple board and I had to go through clunky search and select for parts, even when the digikey part number was in the description.
Nothing in the legal terms explicitly prevents your employees from using my designs for other purposes. If you leak my IP and I find out 13 months later, your terms prevent me from seeking damages because you only cover 12 months from manufacturing.
I like that you have an NDA that I can opt in to, but you should be offering the NDA by default when I upload the files. I should not have to go looking for it in your documentation.
Edit: I realize that there is also limited protection and a lot of risk using a Chinese manufacturer. But requiring source files makes it much, much, easier to look at my design, recognize what it's for, and reproduce it, or at least copy the innovative parts of my design.
I have some level of confidence that JLC is going to protect my IP. They are big and if someone's IP leaked it could be devastating to their business. Their legal agreements are a lot more detailed and favorable than what's currently on circuithub.
So even if nobody ever sees my schematic or gerber files, they could still clone a lot of products if they just have someone spend some time with a multimeter on a device they buy.
And why should I have to sign an NDA? You aren't giving me any confidential data, I am giving it to you.
> And why should I have to sign an NDA?
I think you misunderstood?
---
For additional legal protection of your intellectual property please sign our NDA.
---
To make it worse, you can't even read what their NDA says without signing up with your email address on a third party site.
Isn't capitalism neat‽
Sounds like their actual customers don't agree with you at all...
---
In order to receive a quote on CircuitHub, you will need to upload the original design files from your EDA tool, or an export that carries the same data. This allows us to provide instant quotes and automate the manufacturing process to lower lead times and cost. Currently, our platform works with Altium, KiCad and Fusion 360 files, and with ODB++ and IPC-2581 exports from any design tool, including Cadence Allegro and OrCAD.
We do not accept Gerber files on their own because they do not contain all of the design and assembly data required by our automated process. We do use your Gerber files when you upload them together with an ODB++ or IPC-2581 export, or with Fusion 360's EAGLE exports, which describe the parts.
---
Other vendors don't require this. They just need the gerbers, BOM, and CPL files. That's way different than source files or ODB++/IPC-2581 exports, which is basically the same thing as source.
Good riddance to the compatibly hellhole of 30 different flavors of Gerber/Gerberx/NCdrill
EDIT: haven't even mentioned everyone's different goofball CSV BOMs
Are you interested in partnerships? I'm running pcbjam.com (web port of KiCad with collab, sharing, git, AI etc.) and I'm looking for a partner to do what EasyEDA is to JLC. Where can I find you?
This is one of the benefits of being a company. You use the tax ID instead of your social security number.
Most of my international customers who order bulk require my EIN (and theirs) to be on every invoice.
Nothing new here... I suspect if there is any change it is due to the de minimus being removed. There was major fraud happening in this area...it is a shame it is gone though since I was hoping to be able to order small things without major costs.
Yep, I'm sure that's what's causing this. Prior to that you didn't need to worry if it was under like $700, IIRC.
My feedback: until the moment I click the "order" button, do not require me to log in. If "Checkout As Guest" can possibly be sufficient, at least for the "Grid Proto" tier, then do that.
You have a digital payment instrument for the order regardless of login. That should be sufficient for whatever legal obligations exist, real or imagined, so have the courage to make that policy despite the pushback.
Zimbabwean dollars I hope because 1 million usd PER BOARD?
My understanding is that Shenzhen dominance so far is from 3 pillars - low minimum order size, low per-part price, and fast response. They can do that because they have lot of highly trained labor that work mad hours. So to compete you would need to leverage a lot of automation... which would be a lot less flexible and very rocky at the start. Would love to get a correction from someone more familiar with the situation.
The main latency for this is typically sourcing. The partners I've worked with will get PCBA done within a day or two once they have parts.
The claims here about a breakthrough in setup time also don't really fit with my experience. What is this, putting a new stencil in, adding different components to a hopper? I've toured PCBA factories and it's not like their lines are typically down for set-up. I think this takes like 30min?
Also, it doesn't look like they substantiate their 2x speed claims? They "benchmarked" this, but with what suppliers? Their claims would make sense if they found random suppliers, not exactly specialized in rapid prototype style PCBA, so they're not comparing like to like.
Don't get me wrong, I really want this to happen too, but the desire for this to happen seems to be making the community gullible. I'd love the founders or someone with circuithub to link to real benchmark evidence and explain why their changeover procedure can produce several days of difference.
I've run hundreds of low volume orders on conventional lines, it's death by a thousand cuts, tough to describe exactly why until you do it.
You should. You’re the underdog.
FOUP = thing that holds the PCB.
AMHS = thing that moves the PCB.
Reticle Pod = The thing that holds the component(s) that can add value to the PCB.
You can create a much more capable factory if you trade a fixed, serial line for a scheduling and logistics problem. Assuming your logistics and robots are capable enough, this can provide enormous uplift in high-mix, low-volume manufacturing.
For example, running a test lot (up to 25 wafers) through a modern semiconductor plant is almost a trivial affair. It is like a leaf floating down a river if the scheduling systems are well engineered. I suspect CircuitHub is attempting to leverage similar dynamics.
However I have my reservations because China is incredibly cheap, incredibly fast, and incredibly skilled. Our domestic made boards tend to be lower quality with more mistakes than Chinese made ones (while also costing up to 50x more).
I'll run some designs through this later, and I really hope it is broadly viable (not just unique niches that fit their tooling)
I remember documentaries from the 90s/00s showing Chinese factories with people hand soldering stuff onto PCBs, but most components used in electronics seems way too small for that.
I mean VLSI (chip design) DFM is automated and that's a much harder problem.
I think the availability of cheap humans in China kinda nerfed the drive to automate this (for comparison you can't hand-DFM a chip even if every human on earth worked on it), but it still needs to happen.
JLCPCB is so cheap because everything is standardized and automated. Their tooling analyzes your input, gives a big fat warning if it spots something it knows it can't manufacture, and renders their processing output for final approval by you. Chinese labor might be cheaper than Western labor, but it is not even remotely cheap enough to babysit every hobbyist's one-off board.
They also assume you know what you are doing. You order a board with a part rotated 180 degrees? Well, it physically fits, so they will make it like that. The final board not working as you intended is a you-problem. You put a via inside a pad and it wicks away all the solder? Not their problem. You forgot to re-render the area fills and short ground to power? Not their problem.
The integration of lcsc is incredible.. let's see how they beat this in practice
Some components have large amounts of mass and need to be manually placed because the PNP doesn't have the vertical space for pickup and movement to the pad or do not have feeders which can successfully hold onto the part.
And some times it's just faster to have the human place the part because they're faster than the machine due to physical limitations (like needing to move slowly so as to not drop the component or throw it at the end of travel.) Many board shops will have the PNP places the ICs/Caps/Resistors/diodes and then the placement of connectors (especially zierich) or inductors gets done manually. That is how it worked where I worked. This also provided an opportunity mid-process to do quality checks instead of just letting things go from the PNP straight into the oven.
To take a particular example, they claim to have 11065 SKUs of microcontrollers in stock https://jlcpcb.com/parts/2nd/Embedded%20Processors%20&%20Con... of which 6 are "promotional extended": STM8S105K6T6C (1213 in stock), STM32F103C8T6 (231496 in stock), STM32L151C8T6 (287 in stock), ATMEGA328P-AU (40476 in stock), STM32F030C8T6 (73009 in stock), and STM8S003F3P6TR (152762 in stock). 100 units of this last one will cost you US$30.47.
Where do I find the corresponding information on your web site?
Through hole can be automated, but it doesn't really make sense until you have pretty good volume. So that leaves it still mostly to hand.
Then there are also all the unique customizations, like non-standard mounting specs or running jumper wires.
Some components, particularly transformers, might also be constructed by hand for custom/complex structures. That's component assembly rather than board assembly, but it's a consistent source of manual labor involvement in the supply chain.
Design for manufacturability is one of the important goals being considered in any PCB design. High-volume or low-cost boards will optimize for automated assembly and find ways to substitute or eliminate components with higher assembly costs. Extremely high-volume boards like in smartphones can justify custom tooling if it helps automate assembly.
Manufacturing at small to medium scales is so heavily determined by small blockers/latencies. Missing some component, or need a modified PCB? In the US or Europe you'd be happy to find a replacement within a week, or more likely have to wait a month for it to come from China. In Shenzhen you can probably walk 2km and pick it up yourself in under an hour.
Still, looks super cool and I don't think the price difference is very high for a company doing prototypes.
Buried vias are /super/ expensive from any board house- and very uncommon in most prototyping designs as a result.
Obviously a bit different if you're doing boards with super dense BGA layout chips, but even then you can usually avoid it.
The video doesn't show how that happens. The robotic systems seem to be moving boards around, not parts. Yet that's the big manual part.
I don't think it was meant to be misleading because the small font style was used throughout, and the bottom row wasn't really designed as part of the table, but it sure looks like it at first glance.
I'm not expecting it to be cheaper than China, but in the same general range, and being able to at small quantity/fast turnaround is important. Overall: I'm skeptical of the cost and time claims in the article. I am thinking "What assumptions" are here. Do they mean some part of the process is faster than that same part from China, etc. So I'm curious about what the real-world order-to-arrival time is; what the final CC charge is, and then compare that with China. They advertise faster turnaround: Is that really true than 10 days order to arrival we are used to?
Seems to me that Shenzen still wins at least for the non-USA world.
Plus if you can get someone on the phone there can be value in that.
And if a sales engineer comes out and sees your operation you might get more than a free lunch…
Meanwhile as a hobbyist I'd be glad for the discount "we will get around to it sometime in the next two weeks discount"
In Shenzhen you can have Some Guy in a truck doing a parts pickup round around town a couple of times a day. In the US you're waiting for overnight Fedex from Digikey. And sure, there's a lot you can stock in advance, but virtually every board will have this one weird part you don't stock yet.
They are different political stunts though. If your supply chain is vulnerable to politics then you want options so that any one stunt cannot hurt you.
Also, while there are political stunts, my current reading is the US stunts are more temporary and the next president won't continue them. While China's are likely longer term - there is a reasonably possibly China will attack Taiwan and Europe will join with the US and defending. Only time will tell, but that is my read on the situation.
I am not sure you understand the damage this administration has done.
It's a little sad that Trump is such a divisive train-wreck spectacle that many haven't noticed that Clinton, Obama and Biden were achieving many of the same outcomes as Bush/Trump, just more smoothly and with more appealing rationalizations.
The best thing about Trump is that he's such an obvious self-interested kleptocrat and he doesn't even care to hide it (in fact, he seems to enjoy flaunting it). But is it really that much better to be sold out by a more likeable persona who seems like part of our tribe and has the skills to mask the long con more deftly? I guess maybe at least then we won't notice as much or feel as bad about the same outcome?
https://www.ctvnews.ca/vancouver/article/canadian-military-e...
And here in Europe we're busy divesting from both. The priority for now is getting rid of the US simply because of the erratic President but if you read between the lines, a lot of the plans also aim for getting rid of the dependence on China.
I'd imagine that just because they've decided it's good enough to go public with it, doesn't mean it's done getting better. Or that someone else can't do it better, now that it's been shown to work.
It's not a prototype. The Grid has been running for customers for five years. It already builds tens of millions of dollars of electronics a year. Across CircuitHub we've shipped more than 2.2 million boards, placed more than 150 million parts and served more than 26,000 engineers.
Sadly neither really compete on price with jlcpcb and despite a hemisphere of delivery advantage jlcpcb can be quicker! I once got in a discussion with a Project Manager about whether the design was worth keeping some level of traceability for ~$50... he just wanted it cheaper.
Love that it's fully self-serve and not "please call sales", and will give this another shot with new board designs when I still have total flexibility in part selection. But for existing projects I'm probably just going to re-order from Shenzhen.
Except the components are mostly coming from China anyways, so the same disruptions are potentially in play.
Still, cool to see domestic options.
If we were serious about domestic Manufacturing we'd both be doing massive automation and open borders with Latin America. You need the massive inflow of temporarily cheaper labor.
Automation picks up up as immigrants run out.
Do you see these costs coming down further with mature automation or process improvements?
Though, of course, if this has been in operation for five years already, that cannot have been the spark for this. But it might be what it needs to become fully viable.
Whether the guy responsible for said horror show actually thought this far of course remains questionable.
80% of the US population lives east of the 100th meridian, which bisects Kansas.
It's probably not best to optimize for the lowest maximum ship time in the lower 48, but something more weighted for population. Lowest average transit time. I would have guessed something like Nashville, TN. Claude is telling me Nashville is up there but Dayton, OH is the best. 14% <= 1 day, 46% reached in <= 2 days and 74% reached in <= 3 days. 2.83 average.
Compared to Kansas City, KS at 5%, 25%, 61%. 3.11 average.
This fab is in South Deerfield, MA which is 15%, 26%, 52%, 3.32 avg.
I guess population is also technically not the best measure. It would be better to measure the potential customers. Which is interesting because those are going to be skewed towards the coasts to some degree.
Maybe your Kansas suggestion is more correct than I thought. I'd still wager its more like the midwest than the plains though.
While logistics and delivery (both inbound parts and outbound product) are a key factor, time and cost don't map linearly to distance. Rather than being exactly central, the optimal solve is often somewhat geographically central but leverages proximity to transport corridors, ports and other factors such as weather variability.
KCI isn't a central air hub and winter weather can disrupt flow. Being further south near an East-West hub like DFW can be better (or Houston if an Atlantic port is valuable). Or if bidirectional delivery is everything, base near a Super-Hub like Memphis (Fedex) or UPS WorldPort (Louisville, KY) and go all-in on reaching lower rate tiers with one preferred domestic delivery partner.
For high-automation physical plants, scale can tend to chunk into increments beyond which 'one double-size plant' isn't meaningfully cheaper than two 'single-size' plants. Then an optimal build-out end game can be East and West plants that are each one air-hop from all major metros on that coast and a one-day truck/rail hop from 50% of customers on that coast. If it's a rapid-turn tech product, Reno can work well - half-day road trip from bay area for outbound and a Pacific port for inbound (Stockton). I don't know the quick-turn PCB assembly biz but maybe basing across the street from Digi-Key's biggest distribution center is a 'checkmate' move that beats all other advantages?
Logistics strategy is fun!
Is the robot really working with individual boards and not panels? How do you deal with balancing copper area, and varying etch times?
People often focus on the assembly side as “hard” but quick-turn manufacturing of the PCB is vicious.
Can't wait to see how quickly I can get boards next time I need a batch! The adage 'hardware is hard' is getting less true every day
A supply chain that also comes from Shenzhen... last time I checked, North America doesn't make chips or other components either.
That's never been true (at least since the first chips or components). A lot of my Shenzhen orders have to wait on international airfreight on US-made chips because the ones I spec'd (basically the only ones that would do the job) are US-manufactured and warehoused.
Recommend you look at a world map of Intel fabs. Or Micron Fabs.
Basically anything aerospace and defense related will care more about speed and compliance than cost.
You spent 5 years developing this fab and somehow ignored regular industry standards the entire time? That does not instill confidence in your services, at least from my standpoint.
Needing to sign up for a quote? Who are you selling this data to?