We've written a lot about copper, silicon and thermal vias. This one is about the other input to the design. 662 of you wrote us paragraphs about what DockFrame should be, and we let them change the product. So here's the whole list: what you asked for, how many of you asked, and what happened to each one. Including the three we turned down.

If you've read the multimeter post, the power supply post, or the BreadBoard Card post, you know the DockFrame pitch: a modular USB-C dock built around the Framework Expansion Card format, hosting standard Framework cards and our own Tool Cards. The sign-up form on hwlab.io/dockframe had one open question on it, no checkboxes, just a text box. This is what came back.

No personal data appears anywhere below. Every number is a count of people, and the quotes are unattributed.

The numbers

What you want to do with it

Before the feature requests, the use cases. Every response was tagged with each theme it touched, so these overlap:

The top two bars set the priorities. 260 of you described a dock first and an instrument second. That matters more than it sounds: DockFrame has to work as a plain USB-C dock on the day it arrives, before anyone slots in a Tool Card, because that's the job most people are hiring it for. A lab instrument wearing a dock costume would have failed this audience.

And 203 of you led with the Framework cards you already own:

That's the single most repeated idea in the whole dataset, phrased a hundred different ways, and it's why a proprietary card format was never on the table. Anything we invented would have been better shaped for our own cards and worthless for the drawer of cards you already paid for.

Every request, and what happened to it

This is the whole list, sorted by how many people asked. Use the filters to see only what shipped, only what we punted, or only what we turned down.

Tool Card requests and hub feature requests were counted separately in the raw data and merged here, taking the larger count where a request appeared on both lists.

Tool Cards: the part that gets people excited

The top five card requests are the four cards in the box plus one that Framework already sells. We'd love to claim product vision here. It was a spreadsheet, sorted by count, built top to bottom.

The bars below the fold are the interesting ones for what comes next, and they came from people who knew exactly what they were asking for:

The two requests that changed a card

Most feedback confirms a decision. Two requests reversed one, and those are worth naming because this is the part people rarely show.

The power supply was going to have fixed rails

Our own published roadmap listed the Power Supply Card as "adjustable DC: 3.3 V / 5 V / 9 V / 12 V." Four buttons, four voltages. Cheaper, simpler, smaller. Then this arrived:

Plus a long tail of notes from people who needed a voltage that wasn't one of our four. The card that shipped is a continuously adjustable 0 to 20 V, up to 5 A supply on an I2C-programmable buck-boost, set from your browser or a script, with input and output current sensing. Any voltage in the range, not four of them. It cost us board area, a much harder thermal problem, and one of the three re-architectures that card went through. It was the right call and we wouldn't have made it on our own.

The multimeter's isolation went up because you pushed

The early plan treated isolation as a nice-to-have. Enough of you described measuring things attached to mains, or to a car, or to a bench supply capable of hurting a laptop, that it became a hard requirement. The card that shipped has 5 kV of isolation between the USB side and the measurement side, which is a real cost in parts, board area, and layout difficulty on a 60 x 30 mm card.

Feature requests for the hub itself

Display output was the number one hub request and it's also the hardest thing on the list. We said last year that DisplayPort Alt Mode was "genuinely hard," and that was accurate. The hub carries DP Alt Mode on the downstream multifunction ports with a 4K 60 Hz target, alongside USB 3.2 at up to 10 Gbps. Target is doing real work in that sentence. It's a design goal, not a number we've measured on a bench, and the PoC will tell us where it actually lands.

Ethernet and storage got the same answer, and it's the best answer this architecture can give: you don't need us to make that card. Framework already sells both, they fit the slots, and 34 people asking us for Ethernet is 34 people we can point at a card that exists today. That's the entire argument for using someone else's open form factor instead of inventing one.

What you said it should cost

Fifty-four people volunteered a price without being asked, which almost never happens and was more useful than a survey would have been:

Straight answer: we're more expensive than that, and by a lot in places. Single cards start at $169, the Framework User Kit (Hub + dual USB-C card) is $269, and the Full Kit with the hub and all four cards is $399.

Here's where that money goes, with the numbers we work from.

We're ordering 500 of something. Not 500,000.

This is most of the answer. Volume is the single biggest lever on the price of any electronic thing, and we have almost none of it.

Our own quotes show the per-unit manufacturing cost falling roughly 12 % every time the batch size doubles. Run that curve out from 500 units to the tens of thousands a consumer dock is built in, and the same bill of materials, the same board, the same enclosure costs less than half of what we pay. Nothing about the design changes. Only the quantity on the purchase order.

Passives and connectors are the obvious version of this. The less obvious version is everything around them: a factory schedules a 500-piece run between two big jobs, so you pay for the setup, the stencils, the line time, and the operator's attention as if you were a rounding error, because you are. Injection molding is worse. A mold costs the same whether you shoot 500 parts or 500,000.

The one-time bills land on 500 units, not 50,000

CE and UKCA certification with the EMC testing behind it's a five-figure invoice that arrives once, before we can legally ship anything into Europe or the UK. Spread across a 500-unit first run, that alone is about $24 on every unit. At 10,000 units it would be a dollar and change. Same paperwork, same lab, same test chamber.

The injection mold for the case is the other bill of this kind. Tooling has to be cut and paid for before a single part exists, and it costs the same whether we shoot 500 parts or 500,000. It's one of the things this campaign funds. The prototypes on our bench came out of our own pockets.

The price covers the versions that didn't work, too

Here's the part that never shows up on a spec sheet.

The Power Supply Card went through three significant re-architectures before the schematic settled. Not tweaks. It started as a fork of the multimeter card, lost everything that wasn't a power supply, then grew a PD sink, a wide-input buck, and a programmable buck-boost. The first pre-fab review flagged six blocking issues. One of them was that the buck-boost's thermal pad had zero thermal vias, which is the kind of mistake that turns into a hot, unreliable card and is unrecoverable once boards are back from fab. Fixing it took three vias on the pad and about 107 more stitched across the stage.

The BreadBoard Card is on its fifth schematic revision, and it's the simple card. The multimeter runs to twelve hierarchical sheets because isolated measurement doesn't fit on one page. Every one of those spins means new boards, a new stencil, another few hundred dollars of parts, and two to three weeks of waiting to find out whether the fix worked.

Nobody pays for the revisions that didn't work. The price of the version that does has to cover them.

Two people, most of a year

There are two of us. Not a hardware team with a layout department and a test engineer. The schematics, the routing, the DRC fights, the firmware, the browser apps, the enclosure CAD, the documentation, the bring-up, and the debugging are all the same two people.

That time isn't in the price at anything like a market rate. Billed the way a consultancy would bill it, these kits would cost a good deal more.

Tariffs move under our feet

The boards are made and assembled in China. As of right now, goods coming from China into the US carry a 50 % tariff, and that number has changed more than once while we've been designing this thing.

A crowdfunding price is fixed on day one and can't be changed halfway through, even though the tariff can. So the price has to carry a buffer for a number we don't control and can't predict. If tariffs drop before we ship, we don't get to hand the difference back mid-campaign. It goes where the rest of the margin goes: into the next revision.

A quarter of what you pay never reaches the workbench

Between payment processing, the platform's crowdfunding fee, and a flat per-item fee, close to a quarter of the price is gone before we buy a single component. Depending on how we ship, it's more than that.

We chose that deliberately. It buys fulfillment out of a real warehouse, support staff who answer email when something goes wrong, and free worldwide shipping with no customs bill at your door. Two people running all of that from a spare room is how small hardware projects turn into six months of apologies.

What we did about it

The arithmetic is what it is, but the lineup is shaped around that chart.

The $269 entry kit exists because of the cluster in the under-$100 columns: hub plus Mini Hub Card, no instruments, for the people who described a dock and nothing else. Every card is sold on its own, so nobody has to buy an oscilloscope's worth of hardware to get the one card they wrote in about. Shipping is free worldwide and included, rather than a surprise at checkout. There are no accessory upsells, no subscription, no app store, and no cloud account. And everything is open source when the campaign ends, so the thing you buy stays useful, and fixable, after we move on to the next board.

Where things stand

One thing we need from you

A logistics note, because it decides whether you hear about launch day at all.

Crowd Supply runs the campaign, and their subscriber list is what sends the launch notification. Ours does not. We could take the addresses on our list and push them into the signup form on the campaign page, and every one of you would get that email. We aren't going to. You gave your address to us, not to a third party, and a project whose whole argument is that you should own your hardware doesn't get to be casual about your inbox. Same reason the Tool Card apps run in your browser and never call home.

So there's one click we can't do for you.

And thank you, properly. 662 written responses is an absurd amount of free engineering input for a project that was a landing page and some renders at the time. The four cards in the box are in the box, in that order, because that's the order you ranked them. The power supply is continuously adjustable because one of you pointed out that knobs were the wrong interface. Keep doing that. We do read all of it.

DockFrame is a modular USB-C hub for Framework laptop users, makers, and engineers. Open source hardware, open process. Follow us on GitHub, join the Discord, or tell us what you want to see.

Support Us on Crowd Supply

We're on Crowd Supply, and we need your help to bring DockFrame to more people. Your backing makes it possible to keep prices low and quality high, so everyone can enjoy the power of an open-source modular USB Hub ecosystem.

Image

Check out the campaign and get involved: DockFrame on Crowd Supply