James and the Mooshimeter
Meet the one man show who manufactured the first 2000 Mooshimeter multimeters by hand and is in charge of engineering, distribution, and customer service for Mooshim Engineering.
The Mooshimeter is a modern multimeter that leverages your smartphone as the display, via bluetooth. Not having a display on the multimeter allowed Mooshim Engineering to invest more resources into the components, creating a high quality product (5–6 noise free digits), and allowing users to measure two different inputs at the same time.
Mooshim Engineering was started by James and his friend Eric in 2013. They had a successful crowdfunding campaign via Dragon Innovation in February 2014, and shipped the units a year later in January 2015. Eric went to grad school shortly after, and James has been running every aspect of the company since. I met up with him at his shared office in the San Francisco Dogpatch, to learn from his manufacturing experience.
Manufacturing wins
When James designed the Mooshimeter, he knew he would have to make it with manufacturability in mind, because he was planning on assembling the first 2000 in-house. The Mooshimeter shell has the connectors as inserts in the injection molded housing. Spring contacts mean that it can be easily snapped together. The enclosure is secured with two screws. It has pads on the bottom which are used for calibration. James spent a month building a calibration stand, and after that was built, calibrating the Mooshimeters became a pretty mindless task.
Many friends were bribed with food and beer to help with things like packaging.
Barriers to shipping
In preparation for the crowdfunding campaign, the Mooshimeter had been through 5 rounds of prototyping, and James and Eric were pretty confident in the design. “I thought we could ship by August, but I was wrong.” They didn’t start shipping until January of the next year. The biggest delay came from unexpected results during regulatory testing.
Two Types of Regulatory Testing
There are two types of regulatory testing: radio testing, which means making sure you don’t go outside your allocated channel, and safety testing, since it’s a handheld device. The full test suite is performed by a lab, and cost them fifteen to sixteen thousand dollars. For retesting, the daily rate for a testing chamber was in the ballpark of a thousand dollars.
The first chapter of their regulatory saga began with a small bump. From the blog: “In the lab, when blasted with radio waves in the neighborhood of 200MHz, the meter would erroneously display a large current (up to 2A). That’s not acceptable.” Eric and James added the necessary filtering components to rectify the noise.
The next bump was the housing dielectric test, required for every handheld measurement device rated at 600V. This test simulates a lightning strike, which is an interesting standard for the Mooshimeter, as it is battery powered and wireless. However, it can be held by the user, which would give it a path to ground. This test is done by wrapping the meter in foil and running 5400 VAC through the terminals. The first time this was tested, an arc ran from the screw boss to the ground on the PCB. “It just popped!” So, they moved the screw boss further away from the edge of the PCB.
They also had to widen the enclosure and the space for the SD card slot, because with the super high voltage, the air in the slot was close to electrical breakdown. Lastly their custom ABS plastic was mangled by the oven test, so they replaced their housing material with polycarbonate. This meant that they had to pay about a thousand dollars for an expensive machined polycarbonate enclosure for retesting, since they couldn’t 3D print their prototypes anymore.
After making all these changes, the lab emailed Eric and James to say that they were 99% confident that all the tests had passed. But when the lab was filling out the paperwork, they noticed that they had missed something: Pre-humidity conditioning is required before the high voltage electrical tests. They performed the conditioning, and the Mooshimeter promptly failed the high voltage test, because there was water in the SD card slot. When James and Eric reviewed the test, they realized that the pre-humidity conditioning is not actually supposed to leave water on the device afterwards, which meant that their expensive CNC machined polycarbonate enclosure had been destroyed by an inconclusive test. They made another one, but this one had a defective enclosure — there was a bubble in the plastic — and it failed the electrical tests too.
Machining and testing another enclosure would take two more weeks. After some testing with insulated tape on the SD card slot, Eric and James decided to “take the conservative route and close the slot”. To make matters worse, when the lab said they were 99% confident all the tests had passed, Eric and James had told the contract manufacturer (CM) to go ahead and make the injection molds for the enclosures, and the CM was ahead of schedule. So they had to design an enclosure with a fin blocking the SD card slot that was compatible with the molds already milled.
All in all the regulatory testing took several months longer than anticipated. I highly recommend reading the Mooshim Engineering blog for more details about topics such as emissions testing, or general tips for regulatory testing.
After they changed the plastic from a softer ABS to a harder polycarbonate, James noticed cracks in the injection molded enclosures that the contract manufacturer sent them. The enclosures were fine when the manufacturer looked at them but cracked after going through the mail. James asked the contract manufacturers to try putting the enclosures in the freezer, and sure enough the enclosures cracked. “The area where we found cracking is doubly vulnerable because it’s (a) at the junction between a thick and a thin wall and (b) next to an overmolded brass part (the connector).” Fortunately they were able to adjust the molding parameters to reduce internal stresses.
Once the product was sound, there were still barriers in shipping. Ideally all the units would be shipped ‘ready to use’. Because the Mooshimeter doesn’t have an on/off switch, shipping it ‘ready to use’ initially meant shipping a little live radio, constantly looking to connect via bluetooth. James was alerted by a blog reader about how doing this would likely become problematic: “Customs in many countries find it suspicious if a package is sending out radio packets”. He added ‘shipping mode’ functionality to the firmware, which means that the unit is shipped in a state where it only wakes up every few seconds to check if the resistance measurement has been shorted, and turns on when it detects this.
The final barrier to shipping was labeling regulations. James decided to only include what was necessary in the package. This meant that the unit didn’t come with any extra instructions or labels, because “this is 2018, just check online!” Unfortunately the Mooshimeter was dinged by customs in Germany, for not having proper safety labels (the icons that tell you what dangers to expect from the product). James started to ship a safety card in the packages, preventing them from being banned in Germany.
Manufacturing losses
The Mooshimeter has a metal battery contact, which is manufactured by stamping sheet metal, bending the resulting piece, and deburring it. The manufacturers were supposed to bead blast or tumble the battery contacts to prevent leaving sharp bits of metal (called flashing) on the edge where the metal was stamped. Unfortunately this was not done adequately, and on some units the flashing would cut through the battery label, causing the battery to short and drain more quickly than normal. From the first manufacturing run up until 2017, James has had to refurbish many units that were returned, by deburring the battery contacts by hand.
Apps are hard
Since the Mooshimeter doesn’t have its own display, not only did James have to write the firmware, he also had to write apps for both Android and iOS. On Android in particular, Bluetooth development is challenging because the implementation of Bluetooth on each device is up to the device manufacturer. This means bugs!
James has really great deep dives on the Mooshim Engineering blog about his debugging process. Here are two articles about diagnosing and fixing disconnects on iOS (via firmware updates), as well as a really great read on a firmware update failure due to Android randomly corrupting Bluetooth output data. Gnarly.
It’s a bird, it’s a plane
When the first Mooshimeters started shipping, Eric left Mooshim Engineering to go to graduate school, and the company was solely run by James. We’ve discussed how James is a manufacturing engineer and also developed the Android and iOS apps. Additionally, he maintains the Mooshim Engineering website and forum himself on an Amazon EC2 instance, is in charge of 100% of the customer support that the company provides (including repairs and replacements), deals with all the Mooshimeter distributor relationships, and does all the blogging and outreach.
For most of 2017, James didn’t take weekends. An average day for him looked like waking up, answering emails on a laptop, walking down Potrero Hill to work at the office, walking back up the hill to cook dinner at home, and ending the day with more emails and administrative work.
I ask James how many hours he works per week. “All of them? I don’t have an office culture to fit into so it’s like the division between your personal and work life evaporate. It’s not a healthy thing.” He talks about how friends break up with him when they’ve tried to hang out with him for months but he doesn’t return their calls. In the new year he’s been rethinking how he approaches work and personal life: turning down contracts and trying to make more time, and setting concrete goals for his mental health. He used to take every contract that came his way, but he’s very grateful that he’s become established enough as a contractor that he doesn’t have to go looking for work. On the mental health side, he says: “I try to keep a notebook. Being able to bullet things has changed my life. For example, I can keep track of things like, if I answer 10 emails in a day that’s my quota and I can, without guilt, go on to do something else, even if there’s still something left in my inbox.”
As for what’s next, James says that he doesn’t want to work for a Silicon Valley company that’s just inventing problems to solve. “2018 is such an important political year that I want to find a way to be more involved. In theory I have more control over my time than most of my friends with full time jobs working for a company. I live pretty frugally, I’ve stashed away enough, and I’ve been doing this solo, flexible, feast or famine, get-work-when-you-can thing for 5 years so I’m comfortable with it. So if I can’t take the time to do some sort of political thing, then who can?” He’s not exactly sure what he can do yet, but he was inspired by a talk about radio at a recent meet-up, and will be getting his amateur radio license. “The implications of being able to broadcast at 1.5 kilowatts are huge!”
