Many of us will have casually eyed up the idea of turning a project into a product. Perhaps we’ve considered making a kit from it, or even taking it further into manufacture. But building a single device on the bench is an extremely different matter from having a run of the same devices built by someone else, and in doing so there are a host of pitfalls waiting for the unwary.
[Erika Earl] is the Director of Hardware Engineering at Slate Digital, and has a lengthy background in the professional audio industry. Her job involves working with her team to bring high-quality electronic products to market that do not have the vast production runs of a major consumer electronic brand, so she has a lot of experience when it comes to turning a hacked-together prototype into a polished final device. Her talk at the 2017 Hackaday Superconference: Manufacturing Hacks: Mistakes Will Move You Forward examined what it takes to go through this process, and brought her special insights on the matter to a Hackaday audience.
She started her talk by looking at design for manufacture, how while coming up with prototypes is easy, the most successful products are those that have had the ability to manufacture as a consideration from the start of the design process. Starting with the selection of components, carrying through to the prototype stage, and through design reviews before manufacture, everything must be seen through the lens of anyone, anywhere, being able to build it.
At the selection of components for the Bill of Materials level, she made the point that high quality certified components can be the key to a product’s success or failure, contributing not only to reliability but also to it achieving certification. In her particular field, she often deals with components that can be close enough to the cutting edge to be prototypes in their own right. She mentioned the certification angle in particular in the context of exporting a product, as in that case there is often a need to be able to prove that all components used to meet a particular specification.
When it comes to the prototype stage, she made the point that documentation is the key. Coming back to the earlier sentence about anyone anywhere being able to build the product, that can only be achieved if all possible stages of manufacture are defined. She mentioned an example of a product in which the prototypes had had PCB fixing screws tightened by hand; when the factory started using electric screwdrivers the result was damaged PCBs and broken tracks.
The design review should look at everything learned through the prototype stage, and examine everything supplied to the manufacturer to allow them to complete their work. She describes finding support documentation containing a poorly hand-drawn schematic, and seeing an electronic assembly in which a piece of gum had been used to secure something. She also made the point that another function at this point is to ensure that the product is affordable to produce. If any parts or procedures are likely to cost too much, they should be re-examined.
After the talk itself as described above there is a Q&A session where she reveals how persistent and cheeky she sometimes has to be to secure sample parts as a small-scale manufacturer and delivers some insights into persuading a manufacturer to produce prototypes at a sensible price. And yes, like most people who have tried their hand at this, she’s had the nightmare of entire runs of prototype boards returned with a component fitted incorrectly.
The talk is embedded in its entirety below the break, and represents an extremely interesting watch for anyone starting on the road to manufacturing, particularly in the electronic world. If this describes you, take a look!
[Erika Earl] standing image: David Williams (with permission).
“… it’s really easy, just look at the datasheets…”
“… I did not even know what regulatory standards were…”
Really have the feels for the person that does the compliance engineering in her organization. Data Sheets are just the start. They have no meaning for other than the designers. Spec sheets have no basis for any presumption of conformity.
She obviously knows the important elements of DFM and the basic product design processes, but no other useful information. Suppose this would be of interest to those who are curious about professional design procedures and infrastructure, but do not understand why this is relevant for a ‘superhacker’ conference.
Also, someone should point her to one of the local IEEE chapters that does seminars for technical speaking.
DFM is mostly automating the Production, QC, and import/export shipping logistics.
The last contingency is manual labor, as every screw adds $1 and every component/jumper-wire adds $3 to your production cost.
Her contract manufactures probably keep their mouth shut over the ridiculous small orders — given her team haven’t botched an order yet.
Luck is something engineers should never dwell on — as relying on a single supplier of anything is bad even if your CM moves 50k units a month on 15% commission in China.
https://www.youtube.com/watch?v=G2y8Sx4B2Sk
Yes, you are correct. Looking back on it I believe I was probably too vague and maybe too general. Thank goodness the sub-title of the talk was Mistakes Will Move You Forward! Thanks for the feedback. Also, funny clip.
I’d cut her some slack, ten years ago she was a “tech, receptionist, shipping & receiving person”, and now she’s a director of hardware design. She clearly has management skill, as evidenced by Slate’s willingness to put her in charge, but she most likely hasn’t dwelled in the bowels of a hardware engineering organization for decades as most engineering directors have.
She’s also a respected audio engineer in the industry. Audio engineering is *not* the same thing as hardware engineering, despite the similarity in title.
Hi Brian, to be clear… I am the person that does the compliance engineering for our organization. I also manage our technical files. It’s true, I started from a baseline of zero. In addition to attending seminars, studying up, and making giant mistakes, I made friends with Mark Montrose (the guy who literally wrote the book on EMC) and a Senior Compliance Engineer at Apple who has spent many weekends mentoring and teaching me in his spare time. I discovered when you start by recognizing which components are already compliant early in the design phase you can avoid expensive design revisions later.
“I made friends with Mark Montrose (the guy who literally wrote the book on EMC)”
That man has a name.
His name is Henry Ott.
DFM and the basic product design process is extremely relevant to the hacker interested in moving an idea into production.
Data sheets may not be acceptable documentation of conformity, but any reputable company listing conformance to a particular standard will have corroborating documentation that you can request. That information is not useless. The same can’t be said for your comment.
Do between 7 and 20 submittals to various NRTLs/SCCs/CABs each year for power conversion systems and components. Have been and/or am currently accredited by CSA, TUV Rheinland, and UL per the various client data and test programs. As such, write about 100 Technical Files, construction and test reports, EMC reports, declarations, etc each year. Review on the order of 100s of technical spec sheets each month; of which approximately 50% are technically incorrect, and approximately 15% indicate the mark of a NRTL/SCC or some other conformity assessment body that is not merited, some by marketing error, and some may have been intentional misrepresentation. The compliance engineering community is small, and we talk to each other about this stuff, and tell each other when our published data sucks. Typically, marketing idiots are the culprit. But oft it is a product manager that knows enough buzz words to be dangerous. And some of the messy or incorrect data comes from Fortune 500 companies. Have seen much more than one “reputable company” that has made some big mistakes in their published specifications.
Here is a short list of the more common CAB databases to verify certifications of components and products:
database.ul.com/cgi-bin/XYV/template/LISEXT/1FRAME/index.htm
certipedia.com/TUVdotCOM/home.jsp
csagroup.org/services-industries/product-listing/
Luckily, most of the world has a more pragmatic approach than full-time compliance engineers. Otherwise we wouldn’t be able to buy any small volume products for less than $1K.
How would a datasheet suffice as evidence of conformity to a safety standard?
You go to UL’s site and you look up the product / part’s file number. If it ain’t there, call the manufacturer and ask for it. Not every part requires this. The list of parts are identified in the Appendix of your end product standard.
In general fuses, terminal blocks, relays, transformers, capacitors for across the line usage, PCB materials have UL component recognition requirements if their failure would lead to the risk of fire, electric shock or injury to persons.
However, in many end product standards, if the component used does not already have a safety recognition you can use it as long as the product passes the end product standard, and you do not substitute it.
If a manufacturing engineer or manufacturing representative was not present at your first design review toll-gate, you fucked up.
How big should the PCB panel be? What design features add to the cost of the producing it? What design features cause a lower first pass yield? Which components have a lead time risk? Are any parts going EOL in the next 18 months?
The factory started using cheap electric screwdrivers without torque limiting, which made the result damaged PCBs and broken tracks. Tell the factory to use proper torque limiting screwdrivers. This is a very common thing in many industries and seems like a fairly easy problem to solve.
I think you may not have been in that position. It’s not uncommon to tell a factory in China to do something, they say yes, we understand and we’ll do it that way, then once you are off the premises they start doing it another way because they think they can make a cost saving. And you find subsequent batches of your product are borked, so you come back to them and say you *really* want them to do it your way. I am certain many readers will have been burned that way, I know I’ve certainly seen it happen.
That was a tough watch as I believe there was about 12 minutes worth of value-added content in a 37 minute video. I agree with a few others regarding training for presentation, but
I may have a stronger opinion regarding how to communicate and /or train others about proper DFM since I have 26 years of experience in Electronics Manufacturing and the last 18 have been with 2 different CMs, but I feel she simply glossed over many of the basics regarding DFM. You could get the same level of detail with a few google searches. She does at least have some real life experience with wins and fails, which is of value however. The example she gave regarding the electric screwdrivers damaging traces is horrible as it shows two key failures: 1) the PCB design obviously had traces too close to mounting holes as over-tightening a screw should never be able to cause trace damage, and/or, 2) the engineering counterparts she worked with at the CM were severely under-qualified.
Watching videos like this really makes me wish I had the time to make a YouTube channel centered around Electronics Manufacturing. The line between professional manufacturing and hacker manufacturing has become fuzzy in many ways and since I dabble on one side for a day job and the other side for a hobby, I think I could communicate what I do professionally in DIY/Hacker language to make it valuable. I did an “Outsourcing 101” presentation at an event a few years ago to a similar audience and it was very well received. I covered the basics needed to work with a CM to launch a product.
When you do, let us know: tips@hackaday.com :)