Hi everyone – we’ve had 3 days of production with a pace of 24 finished, boxed, ready to ship units per day! We should be able to fill the first 200 units over the course of the next 2 – 3 weeks. With a few exceptions, we will be shipping MultispeQs in the order they were received.
Thank you all for your immense patience as we have worked through this long process.
On a personal note, this is my last week as part of the PhotosynQ project. My hope is that the project continues to strive for the values of open software, hardware, and data in science, and to create tools which are broadly useful to the both the scientific and non-scientific communities. I know that these ideals are sometimes hard to hit, but if not us then who, and if not now then when?
Thanks to everyone that I’ve interacted with over the years for your patience, enthusiasm, support, and collaboration. In the future I’ll be on google plus +GregAustic and twitter @GregAustic. Hope to see you in my next project!
Hi everyone – it’s been about 3 weeks working here at Saline Lectronics to get the initial build of 250 MultispeQs underway. There have been lots of ups and downs, but we finally got the motor running.
We can now make about 24 units per day, which means we can fulfill the initial 250 in about 11 working days. Here are some pictures of the first 24, ready to be put in a box.
Thank you so much Jared, Paul, Jan, Cathy and everyone else at Lectronics for pushing through and getting 250 units through completion… it would probably surprise people how much we had to touch every single one of these devices to get them out into the world. Hopefully the next 250 are easier than the first 250 🙂
For the manufacturing uninitiated, a manufacturing facility can be pretty impressive. Robots, half-million dollar pick-and-place machines and assembly lines with people diligently working on making stuff produce a lot of ‘ooohs’ and ‘aaahs’.
But, automation is only part of the story. There is a lot of decision-making and little details to make things actually work. Our first week in manufacturing has had some setbacks with a few misplaced parts and similar issues, but our manufacturing partner has beefed up their support and Sean, Dan and I have working side by side with them. We’re refining our production process and ironing out errors. In all, we are still making decent time.
All production processes look different, but here’s a peek at ours:
1) Pre-assemble case. Many parts are more efficiently pre-assembled in bulk – placing the small light-blocking o-rings in the main case, for example, or installing the battery. These pre-assembly steps are largely done for all 500 units, so once circuit boards are ready the device will snap together pretty quickly.
2) Assemble + test circuit boards. The 2 circuit boards are connected, bluetooth and DACs are configured, EEPROM (memory) is zeroed out and lights are tested.
3) Place boards in case. Circuit boards are installed in the case. The PAR sensor, which contains several very small parts which fit precisely around the TCS light sensor, is assembled at this time as well.
4) Calibrate + validate. We have 7 individual calibration steps, as well as a final validation on 3 different plants of varying leaf colors and thicknesses. All of this data is saved to the website and you (the user) can follow up on your devices factory calibration (if you want).
5) Pack and ship. Now the final devices are packed in their boxes and shipped to labs around the world.
As of today (Thursday), we have almost all of the pre-assembly work done, and are in phases (2) and (3) above. Most of next week will be spent calibrating devices and boxing them up to ship.
Today we started producing the first 250 of our 500 initial production batch at Saline Lectronics, a contract manufacturing company out of Saline MI. No major hiccups, though we had one or two LEDs which were switched and a filter which was incorrectly sized, but those are fixable in the next day. Luckily we found them after the first 30 units, and not after the first 500 🙂
We spent most of today prepping the case itself (as you can see in the picture), and did a dry run of a single unit through the entire process (from assembly to calibration) to ensure we had everything we needed. Bluetooth continues to be challenging (connecting mostly), and it’s now my least favorite communication standard by far! But once you get it connected once it gets easier after that.
Expect to see more updates in the next few days as we continue to move forward. The current estimated ship date is March 3rd. If there are problems or setbacks I will update you as soon as they happen.
I was hoping to have more specific information for you guys before creating this post, but we need to provide an update either way. Here’s the update:
The bluetooth module was mis-shipped again (yep, not kidding). So we managed to find a local supplier. This time we confirmed it’s the right part by ordering it and installing and testing it first. The expected delivery date for the full quantity we need is Feb 10th.
We are waiting on a response from the factory on when we fit into their schedule once those modules are delivered. We have been planning on late January, so they’ll need to adjust their schedule to get us in in mid February instead.
Everything else is in place to fulfill all of the pre-orders. Specifically, we have:
All parts except the bluetooth module
Factory calibration process for each device
Factory Assembly instructions
Other packaging (instructions, box, etc.)
Confirmed precision and range of key measurements on a a subset of 10 – 15 devices
Updated website and android app
As soon as we begin manufacturing, I’ll update the blog again with details, progress and pictures.
The MultispeQ ambient PAR correlates well to a LiCOR LI-250A PAR meter (LiCOR = MultispeQ * 0.95 + 3.99, r2 = .996, using 13 devices with 206 measurements). Devices showed some offsets from the actual PAR, but the noise at any given light level for each device is relatively low (high r2).
Validation included measurements in different spectral conditions: cloudy day, sunny day, LED, and fluorescent lights.
Additional spectral conditions (for example, inside a dense canopy) should be added to improve the quality of the validation.
The MultispeQ has to replicate ambient PAR inside the measurement chamber. This is quite tricky because there is some error in measuring ambient PAR and some error in converting ambient PAR to a defined LED intensity. Those two errors combine to increase total error.
The MultispeQ’s conversion of ambient PAR to LED light inside the leaf chamber is ActualPAR = 0.94*LEDPAR + 12.17 (r2 0.979) using 10 devices, across a range of spectral conditions and intensities, with a total of 253 measurements).
MultispeQ LED PAR tended to be shifted higher slightly compared to the actual ambient PAR.
Green, red, and blue LEDs (lights 1,2, and 4) correlated the best with the LiCOR LI-250A. The orange LED (light 3) performed the worst.
We’re going to continue to validate and check devices as we produce them, and post it through these same projects so everyone can see it. Expect another update regarding production status in the next few days.
You will find comparisons to commercial devices, calibration, and use and application in the field in collaboration with Beta test partners.
It’s important to note that this is paper is about the MultispeQ Beta device, NOT the MultispeQ v1.0 which we’re working on currently. The MultispeQ v1.0 will have its own publication. While the v1.0 has many improved features, better quality, and higher accuracy, the v1.0 is otherwise pretty similar to the Beta and this publication is a useful starting point for understanding either device.
Congratulations to Sebastian, Dave, Dan, Marty, Robert, Isaac, Donghee, Mitch, Kevin, and Pro for their hard work on getting this paper out into the world and the many sponsers which supported us along the way.