I kept meaning to write this for quite a while, and I kept not getting to it.  Ugh!

Anyway, a handful of people keep asking me, like, what needs to happen between now and when? And when am I going to do what? So I figured I'd better write out my checklist of what's between here and there. Basically, I'm splitting this up into four major milestones:

What needs to happen before crowdfunding ends?

• I'd like to get a final firmware update to a few early testers that has a representative UI that they can mess with and make some videos with.

What needs to happen before I order boards?

• Money needs to come in from Crowd Supply.
• There are a few changes left to X1P-002's schematic and layout that I need to make -- mostly, a few stability changes to the voltage regulator, and a few value changes to the overcurrent protection circuit.
• I need to validate using the GPIO on LAN9514 to reset RP2040.
• I need to quickly prove out in some early firmware that x1plusd can talk to RP2040, through a LAN9514, on a real printer. It doesn't have to support everything but it does have to work at least a little bit.
• I'd like engineering samples of the X1P-001-A03 cables to come back. That would be nice.

What happens between ordering boards and getting boards?

• Finalize SLS plastics -- do a sample order, and then order in volume.
• I need to fab out and check out X1P-003, the manufacturing test jig. That can be quick-turn since there are just a few of them. Design and print plastics for the test jig, and then machine a few dowels to hold the boards.
• Production samples of boards need to come back so I can check them out. Then, I can P-release boards.
• P-release cables (should be shorter lead time than boards).
• Final RP2040 firmware gets written then against the production-sample boards.
• The lead time on boxes is short, but I definitely need to do art for the boxes, because man, all those converging lead times creep up on you!
• Will I have time to make a trip to Shenzhen? That would be nice!

What happens when I get boards?

• Pizza party at my house! If you live in the Bay Area, and you want pizza and beer, send me some e-mail, and we'll have an assembly party. There's going to be a bunch of work testing and labeling boxes, and stuffing packing peanuts. And petting Peanut.
• Ship a bunch of large boxes over to Mouser.
• Hopefully that will coincide with X1Plus release 2.1. If not, hurry the hell up before people get boards.

Huh, wow, that list was a lot shorter than I thought. Is this actually going to work out? Stay tuned...

OK so I had planned to write a bit about, like, the manufacturing work still to go, and where we are, and what I'm hoping to have done before crowdfunding closes, and then between then and ordering boards, and then between then and boards coming back, and then between then and shipping to Mouser.  I still plan to write that post, but over the weekend, I put together something that I'm super stoked about and I hope you will be too.

I shot and edited this together, and it'll go out in a Crowd Supply update in the next day or so.  But I'm posting it here first because I think it looks insanely good and I can't wait to show it off :)

Since the Crowd Supply post is going to be a little broader in audience, I won't get to say much in terms of command line details there.  But for the Hackaday audience, for those curious, here's what I did:

• I hooked up my a6600 to the Expander, and set it to 1/4s exposure at ISO 1000, put a 55mm lens on the front, and stopped the lens down to f/9.  I focused right on the bed where the part was going to print.
• I then set an X1Plus Action on layer change to pause for a second, pull the shutter GPIO for a half a second, pause for a second again, and then release the G-code interpreter.  The pause was important because the card table I have in my filming space is kind of rickety, and I needed to give the vibrations time to settle down!
• After printing, I had 300 JPEGs or so, each 24MP.  I fed them to FFMPEG to get something I could bring into DaVinci Resolve.  I couldn't use my Mac's hardware codec for this, because it didn't seem to be able to deal with 6000x4000 images!  But x264 did just fine: ffmpeg -framerate 10 -start_number 1394 -i /Volumes/Untitled/DCIM/100MSDCF/DSC%05d.jpg -vf format=yuv420p -r 10 -crf 27 -y timelapse.mp4
• There was still a handful of shake from frame to frame, since the camera's image stabilization was working pretty hard to correct for the table wandering around.  I loaded the timelapse.mp4 into Resolve, and had Resolve do a perspective stabilization on it, which got it pretty solidly decent.
• Finally, I keyframed in Resolve set zoom and crop windows.  Since I had the time lapse set to 10fps, and I had a timeline frame rate of 30fps, I had Resolve interpolate linearly between frames.  Resolve also had optical flow that it could use for this, and it made even smoother looking video, but I felt like it was cheating for this particular case.

I think this came out super duper rad, but I also sort of feel like this is just scratching the surface of what's possible.  I imagine that people who actually know things about video editing are going to have an absolute field day showing off their parts with these time lapses, and when I finally get a chance to breathe, I kind of want to try to put together a turntable to spin a camera around a part as it prints.

Anyway, here's the part where I'm supposed to, like, do marketing, and stuff.  Have you remembered to place your order yet?  How about for the shutter release module?  Do it now!

Whoah!  I'm on the plane on the way home from Supercon, and man, I am exhausted.  This weekend, I talked about X1Plus as a whole at the conference, and more interestingly to you, I (and my amazing Crowd Supply project manager!) got X1Plus Expander launched, and you can give me money now!  (Also, we changed the name from "X1Plus Expansion Board" to "X1Plus Expander").

So, ok, first thing first: you should definitely give me money and buy this thing!  Here's the Crowd Supply page, if you forgot.  It's $125 for Expander, $25 for a nice pre-printed SLS case, and a few bucks more for varying add-in modules.  I'll write more about the economics of it... much later, but I promise that I will.

But also you might be interested in watching the video that I alluded to.  I'm really proud of how it came out!  Here's 9 hours of filming and editing for a 90 second product video:

All this socialization really makes me more ready than I have ever been before to sit down and write a UI for the Expander configuration.  Which I guess is a good thing, because we need one of those soon...

I've got a checklist of other things to take care of in the near future, basically, in categories of "before crowdfunding completes", and "before units arrive on my desk".  I'll write that up soon!  But for now I'm going to go home and sleep, and hope that my half-ass job of masking saved me at least a little bit from the nearly-certain con crud.

Are you at Supercon this weekend?

If so, come say hi.  Below, an X1Plus Andon Board hacked up as a SAO for the Supercon 8 badge:

So we've been sending some Expanders out to folks who have been involved with X1Plus since the very beginning.  It's validating to hear that they're exciting enough for someone else to bother to write a plugin for an I2C module they have lying around!  WolfwithSword, who was responsible for a lot of the MQTT Home Assistant integration work for Bambu printers in general, just submitted this patch to support the PMSA003I air quality sensor.  The data WwS has gotten out of it is very cool so far, showing very reasonable levels when printing with PETG:

And some truly unpleasant results with ASA:

Very cool.  What other sensors do you think are worth adding support for?  How would you render these data on screen?

One of the things I had promised [1] to talk about is not just the fun technical bits of this whole project, but also the boring logistical bits.  This is a good opportunity to talk about where this is, and what's between here and launching the campaign and taking your money!  A user, as far as I can tell, joined the Bambu Lab forums explicitly to say they wanted my thing, and then posted nowhere else other than to complain that it's taking too long, which definitely feels like quite a compliment to me, and some solid market validation.  I gave a short overview there, but here's the longer answer.

But before I give a longer answer, well, ... I guess I'm going to have to ask another question.

Another question: What do I actually need to launch a product?

Roughly, there are three categories of certainty that I need to have before I can actually start taking money.  What I need is:

• I have to be sure I can make the thing I'm promising to make, and I have to know how much it's going to cost me, and I have to have a good idea of who's going to do what.  There are some baseline things here that you'd expect that just take some time: for instance, I need to get some quotes from contract manufacturers (done that), make sure that the contract manufacturers are prepared to actually build it for me (mostly done that), run the numbers on what my margins will look like in "goes great" and "goes bad" conditions (done that; I promise I'll write about that, but only when everything is all said and done).
  
  There are things in here that surprised me, and took longer than I expected.  For instance, I had expected that every CM in China would be able to get me the same shipping rates to the US; the CM that I really wanted to work with, though, seemed only to be able to get much worse rates from DHL than I had expected.  They were quoting me like $400 to send a box of X1Plus Expansion Boards to me, whereas the other CMs were quoting more like $150.  Then, when I got a quote directly from DHL, they were saying it'd be more like $1,000!  This clearly would not do.  I spent like a week of worrying about this trying to learn about Shenzhen freight forwarders (answer: they exist but you have to go to Huaqiangbei and talk to them yourself), and then finding some US import brokers who at least had a corporate rate with DHL.  Then, after trying to find the best of bad options, the DHL corporate account guy finally called me back and said that he was behind on his e-mails because he had been in a car accident, and that he could get me completely reasonable rates with my shipping volume, and that was the end of that.
  
  Manufacturing is death by a million cuts.  The more of my ducks I get in a row beforehand, the more schedule margin I have later to mitigate chaos later.
• I have to actually have enough creative collateral to put on a launch page.  You can basically put anything on a prelaunch page because nobody is giving you money -- they're just giving you their e-mail address, and who cares about that?  But when it comes time to get people to give you dolla billz, you have to prove that you actually plan to put in enough effort to give them their thing, and the best way to prove that is by showing that you've put in a decent amount of effort already on getting it ready for launch.  It's just a respect thing for your audience!
  
  So, 2,500 words later, my poor Crowd Supply project manager has to deal with editing my overly-lengthy writing.  Every good Crowd Supply page has a comparison table; this one needed one too.  Every good crowdfunding page of any kind needs to be frank about the risks of the project, and then pump the creator up to show that they know what they're doing and that you should trust them.  A decent crowdfunding page has text and pictures; a great one has a video, so I guess I need to print myself a steadicam grip for my a6600.
  
  Even if I were just throwing this over the wall onto a Tindie page and walking away whistling, even if it's not mandatory to turn on a "buy" button there, this sort of stuff would be important.  It would still be a lot of time to write this for a Tindie page (but I wouldn't have an editor).  It would still be a bunch of time with a camera (but I wouldn't have someone to fix my poor lighting decisions).  No matter where you launch, you need launch collateral.
• I have to have people who are ready to buy it.  The good news is: this campaign has been a surprisingly large organic success so far!  I have a lot of people's e-mail addresses!  Like, way more than I had thought!  Like 700 of y'all seem to be excited about this thing!  I have no idea how many of you are going to convert to buying boards, though -- and once I have a buy button available, I need to have the outside world ready to continue to drive traffic to that buy button.  In the 3D printing world, that roughly means getting YouTube creators on board with your stuff, because it's the future, where all information comes from videos.  (But it's always nice to have, say, a friendly blog post!  Hey friends, write a blog post about X1Plus Expander Board!)
  
  In order to seed creators with units that they can start playing with and talking about and getting people excited about, though, that means that I need firmware that's useful by people more than just... me.  I mean, I can get my printer to trigger my mirrorless, and, like, that's definitely rad, but unless there's GUI support for it, or at least documentation, I'm just not so sure that anyone else is going to get to do that.
  
  If I want creators to start making content that intercepts the middle of a campaign, then roughly, that means that I need them to be seeded with boards and firmware ideally before the campaign launches -- even if they're not production boards or production-quality firmware, they have to at least be able to play with it!

Woof, that was a lot.  Ok.  But now that you have the context for it, that's the beginning of an answer to: what's taking so long?  So now I get to tell you my secret plans, or, at least, what I think is left between here and launching the campaign.  Luckily, the list is actually pretty finite!

• I need to finish a round of edits on the text for the campaign page.  I spent all of yesterday writing a mess, and the first draft is basically done; now, it needs to be copy-edited, and then churned into the Crowd Supply CMS.  Thank your local technical writer, folks.  You have no idea how hard their life is when they have to deal with engineers like me.
• I have a video to shoot.  That's probably a next weekend job.  I have a script for it all laid out, with all the shots I need; I need to move my X1 up to my brightly lit office and take a bunch of video, then load it into Resolve.  How hard could it be?  I did it once before, after all.
• I have to go do the math on wholesale pricing.  I'm pretty sure I understand what campaign pricing is going to be at this point, but I need to figure out what pricing is going to be like for resellers.  While I'm at it, since Mouser is Crowd Supply's primary distributor, I need to send them some documents to convince them that I'm a real business.
• I need to write enough of a GUI for X1Plus itself so that you can at least turn on and off the LEDs, and wire up some GPIOs, and make people wish that they had a better configuration framework.  Probably some written documentation wouldn't be a bad idea, either.
• I need to get my slides together for Supercon, where I am talking about X1Plus and this board!  (Do you have tickets yet?)

So when does this all have to be done by?  Oh, well, say, how about ... Supercon?  Yeah, that sounds like a good deadline.  It would be nice to be at Supercon and have people be able to click 'give Joshua money', rather than just 'subscribe an e-mail address'.

We'll see if it happens.  I think it's not an overly ambitious goal.  There's a fair bit to do but I think it's doable.

[1] Ok, I promised Heiko.  The rest of you have to suffer through my promise to him.

The march to having something we're ready to productize continue.  X1P-002-C01 (that's the third revision of the mainboard, which I talked about in the previous update), with the new buck converters and the RP2040, came back on my desk last week, and it seems to basically have worked on the first go.  It looks like there will be a few changes to passives -- it turns out that these Silergy regulators are really finicky -- but everything seems to work fine other than that.  Below, a "half-build" of an X1P-002-C01 (no LAN9514, yes all the new power supplies, yes RP2040) happily blinks an Andon Board:

One of my tricks of the trade for making sure that I actually make it through all the things that I needed to check out on a board is that I do a checkout spreadsheet, with all the tests I care about.  That way, when I think of more tests, there's an obvious place to write them down so I don't forget, and I can also take notes on a debug if things aren't going the way I wanted.  Here's the final tally for the X1P-002-C01 checkout:

There's been kind of exciting software work recently too, but that's food for another post.  Maybe later this weekend...

Well, that was a fun adventure.  Over the last week or so, Dan and I were putting together some changes to prepare for the next rev of the board, and one of the things I had on my list to check out was to replace the FT2232 on my board with a FT4232, and make sure that everything was OK.  The problem that I was trying to solve was that we had two of these lovely FTDI high-speed I/O engines on the '2232, but the other two ports on the top shared USB interfaces, and couldn't really be commanded independently -- and I really wanted four homogeneous ports on the thing.

I depopped the '2232 on a test board that I had, and soldered down a '4232.  Clearly I had done a poor job of soldering it, but while I was flipping through the datasheet for indications as to why the '4232 wouldn't work in the '2232's place, I came across the following distressing table:

Hey, what gives?  The FT4232, despite having a 4 in the name, doesn't have four homogeneous channels either!

Well, I'm glad we figured that out now, rather than after having manufactured a bunch of these.  That would be really annoying software to write.  But truth be told, I had been getting pretty frustrated with the FTDI's limitations already -- most notably, the fact that it drives some of its I/Os push-pull on startup until initialized.  So, after plenty of trying to avoid it, I did what I probably should have done from the beginning: I kicked out the accursed FTDI, and designed in an RP2040 ("only moderately less buggy, but, hey, at least it's cheap!").

I am only vaguely looking forward to writing some firmware for that thing to implement a cheesy pair of USB endpoints that can permute some GPIOs -- and do some high-speed serial things -- on demand... and very much not looking forward to rewriting all the FTDI code that lives in x1plusd is going to be even more annoying.  But it's going to be easier than making the FTDI work long term.

A cut-down version of X1P-002-C01 with the RP2040 and some higher efficiency switchers, and better overcurrent protection, is out to fab... I guess we'll see whether I got it right in a week or two!  If so, this should be pretty close to the last rev before I'm ready to go to production.  Fingers crossed.

One of the goals of this project has been to support plugins to the rectangular GPIO headers on the top.  Early prototypes of the X1Plus Expansion Board have FT2232H, but sooner or later I'll be changing to FT4232H to make all four ports homogeneous.  We have three plugin boards that we're working on right now:

• X1P-004: a WS2812B-style LED strip level shifter board
• X1P-005: a compact light stack, beeper, and button board
• X1P-006: a camera shutter release trigger board

X1P-006-A01 (the first rev) came back a few weeks ago, and we identified some changes needed: new 2.5mm connectors that had *all* the pins connected (d'oh!), pinout changes for safe initialization state on the FTDI, etcetera.  To validate the changes before we fab out X1P-006-B01, I hacked up an X1P-006-A01 to match the new specification, and bit the bullet and upgraded my ancient Sony NEX-C3 to a fancy new-to-me used a6600.  Of course, the first thing I did with an $800 camera was to connect it to a piece of experimental electronics dangling off of a 3D printer!

(Peanut is the official First Layer Inspector of the X1Plus project.)

I spent a long while writing some software to glue G-code actions to GPIO outputs on the FTDI -- like, a week -- and took two time lapses: one with the printer's built-in chamber monitoring camera, and one with my a6600, using the X1Plus Expansion Board.  You can compare them for yourself, if you like:

• The printer's built in time lapse looks about like you expect.  Grainy, boring.
• The X1Plus timelapse with my a6600 really shows off what you can get with some decent glass and a different angle -- and this is just my first attempt at it!

I'm pretty excited about how this came out so far.  This is the sort of validating experiment that tells me that this might be worth doing!