Tips to get a first market test batch production of a simple product ASAP ?

[fab672000]

So from the concept I do have now a prototype, that i use everyday at the office ATM.
I developed this one at home,  in about 3 week ends and time includes the Windows host software, the embedded software and the hardware (on a soldered breadboard for now!). My colleagues suggest to move to the next step and sell it, but I feel reluctant to get to the manufacturing aspect ; in particular pcb + enclosure design + packaging aspects are where I am stuck at and not frankly being very motivated to investigate. The idea has not been done already and happens to be useful though probably not for a million units market.

How about a #29 blab on how to get a prototype manufactured, packaged and selling on the Web as easily and fast as possible for market testing ?

I'm interested only by any tips and trick to shorten the manufacturing aspects, in the context of getting only a market test batch of , say 20 units ; done first and nothing more complex yet that would lead to more professional workflow practices.

The idea being that if the market tests does not show much interest, I don't want to spend too much time and energy on manufacturing and better switch to another prototype idea for a similar market test  and repeat the process.
 
As an example, I would like to challenge and not be assuming I need a pcb design  yet: electronics is simple and compact (My current device would probably fit on a 60x80x40mm enclosure) .

Similar thinking for enclosures, there may be some shortcuts that I may be able to use to avoid spending too much time on that.

But then I don't see how I could avoid spending design time and specific enclosure design in order to access my device's ports and specific modules (the box would need holes for connectivity at least).Though about investing in a 3D printer but would it be suitable for 20 units to build as opposed to few prototypes ?

Until now, these problems did not exist on the prototype and it was all pure fun & delight to design something that really works and also a fast process (few week-ends); I suspect it is the end of the fun part or is it not ??

Any real  tips sharing  of  post-prototyping experiences to get to a home-made a produced product, would be greatly appreciated.

Sorry in advance for my newbie questions, I usually develop software not much hardware and started to become very fond of learning more about electronics, which surprislingly was not the main obstacle I had to deal with in my current projects.

I also realize the question is a vast subject, and I apologize for not sharing much about my prototype.
I don't expect specific solutions to my problem, but instead any working tips that if applicable, could shorten the manufacturing phase of a small product market test batch.

[Howardlong]

Something I learned was when designing real products, as opposed to tinkering about, you need to comsider the form factor and envelope much earlier, as that will frequently define your board dimensions.

Trying to jury rig a random board into an enclosure it wasn't designed for will almost certainly be messy.

You can of course have custom enclosures made, but tooling is invariably going to be very expensive. Usually for low to medium low volume, it's common to use an off the shelf enclosure, so you pretty much have to design your board around the chosen enclosure's envelope. In many ways that latter method makes things easier because many previously flexible parameters are effectively forced on you, so you spend less time procrastinating.

3D printers, unless you have access to a seriously expensive one, are OK for prototypes and rough fit checks, but are not for commercial products. Sure you can use the parameters to have your enclosure made properly, but unless it's high volume or high value, I'd use an off the shelf enclosure, or else you'll find your enclosure costing the majority of your BOM. Depending on your market though, the target audience might demand a custom enclosure, but for low volume I'd avoid it. If I had industrial design skill sets, I might have a different conclusion though!

One thing I would say is that there is a world of difference between having something working on the bench and having it manufactured. If you're "not... very motivated to investigate" then my first thoughts are maybe you should stop right now.

Selling on the web is easy, essentially it's a Wordpress website and a Paypal button. Much as there is a vociferous hatred of Paypal, and much of it is justified, its ubiquity means it's the most sensible way to start receiving money. Setting up an eCommerce website and credit card payment provider interface, especially for international use, is a very significant undertaking, which is why I recommend a simple Paypal button on a web page.

Dealing with Paypal though is a right royal pain in the ass when you first start bringing in significant funds, particularly in that they withhold funds sometimes for a couple of weeks even though you've delivered product. Once you're established, payment time becomes immediate, but it takes a month or two for that to happen. Even if you have an established personal account, that counts for nothing as far as Paypal are concerned. There are several other gotchas with Paypal, such as don't access your account from abroad, as that will trigger your account back into a two week payment delay mode for example. (I access via my home VPN to avoid this).

You might also want to consider whether you want to find a retailer or distributor, but whatever you do don't work to silly margins unless you think what you're doing is a charity. If it costs you, say, $35 COGS/unit in volume you should be looking at at least $100 retail. The retailer will be paying you around $70-$75 per unit. You have to keep a roof over your head. You are the one taking the risk by investing in the product. The product needs to pay its way, and it is right to expect a return on your investment. Don't be afraid of making a profit. It is a lot easier to reduce price than it is to raise it. If you have a unique product that people want then people will pay.

For 20 units, I assume you'll make those yourself. My first product, I made the first hundred or so myself, which had about 120 parts per board, all smd. It was not a career I wanted to get into, hand placing parts. You need a board assembler and the best way to find one is on personal recommendation. Do not be tempted to go to China just because it's cheaper, because I guarantee you that for small/medium volumes in the end it won't be. There is a different culture, you think you have a deal, they will use all the parts as specified on your BOM, and then they go and substitute crappy parts. It happened to me, it happened to Raspbery Pi, and it'll happen to you!

I am not sure what the "market tests" are for? Put your gadget up for sale, see if it sells.

Regarding packaging, if it's a gadget, frankly I'd find a reasonable existing plain box that can be had in bulk for pennies and use that. Print some stickers out and but them on the box.

Lastly there is the thorny issue of certifications and type approval, but it very much depends on your particular device and the markets you intend to sell into as to what that will entail.

It's not that it's not fun, but a lot of it can be challenging and frustrating, particularly when manufacturers let you down or parts become unavailable. The best bit was being able to give up my day job.

[fab672000]

Hi Howard,
Lots of useful information here so first thanks a lot for your answers on my naive noob questions!

Yes, I would be producing these units myself, eventually I know a guy I worked with that could build them and thanks again for the suggesting not using the chinese way for the assembling choice, as I was wondering if it is worth it for small quantities ; and it seems not.

Not that I am particularly pessimistic, but for now I don't even plan to get a medium/high volume of a product ; I am more thinking of testing a good idea and if it seems to get  more interest that the one by my colleagues ; eventually sell the entire idea to one of the companies that I know could be interested and provide them with it.

Yet I thought that even in that last case, it would be very nice to have devices to show and not only one hand made ugly prototype (yet 100% working) so that's where I'd really like to continue the manufacturing adventure, in my spare time only for now !

I cannot see certification issues for this particular simple idea, but again that is a great point that I should also consider. I am aware of BOM related issues and already have substitution parts in mind in case that would happen, and you are also right that I should also think of these implications on the PCB design (and the embedded software too)  as early as possible.

[ebastler]

Many great comments from Howard. Just two aspects I'd like to add to:

(a) Forget about 3D printing... The affordable (amateur) 3D printers are not meant for producing anything, they are meant for playing with a 3D printer.  

An off-the-shelf enclosure is probably your best bet. Or, depending on the components in your circuit and your target audience, you might not need an enclosure at all. If it is a small, battery-powered gadget for a techie audience, a bare PCB might work. You can play some nice tricks with free-form PCB outlines, a clean top side which doubles as the "front panel", while most (SMD) components are on the bottom, etc.

(b) Don't forget about, or underestimate, the compliance aspect! If it is a small and battery-operated device, electrical and mechanical safety or fire hazards are probably not an issue. But if it uses a clocked, digital device on board, electro-magnetic compatibility (radiated emission) probably is an issue! And the cheap enclosure options -- plastic box or bare board -- are no help on that front.

Getting EMC compliance established and verified is expensive... You may decide to just ignore this and fly "under the radar", or sell your gadget as a kit if that works for the target audience?
 

[tpowell1830]

MacroFab.com

[Howardlong]

I forgot to mention testing. Spend some time thinking about ways you're going to test your device, and how you might want to build some of that into the design. Sure, for 20 units it's unlikely to be a big deal, but as soon as production ramps up, you'll thank me.

You need to be in a position to test every component and signal path, so there may well need to be multiple tests to accomplish that. Automation is the name of the game!

For example, in one device I have it tests the RF front ends by injecting a series of about 50 different signals from a built in signal generator and measuring its own results. A preprogrammed device in the unit puts the device in test mode the first time you turn it on. The whole test takes about five seconds. It presents the results on a computer through a USB interface, and if it passes the device writes to a fuse so that next time it turns on it skips the test mode.

In a couple of my devices I also have an LED inside whose sole purpose is to give an indication of any failure in case not even the USB interface starts, for example if a voltage rail doesn't come up or a device isn't responding on the I2C bus. It flashes out a two character code in Morse code (!) to indicate the device state.

While you can self-test 95% of a device, you also need tests which do end to end testing perhaps including the use of calibrated test equipment. For those secondary tests, you need to come up with a test jig.

Also as part of the test, make sure that the full end to end test determines any physical problem such as malformed or bad connectors. Some of my test jigs use quick-connect (none-screw) jacks onto screw sockets to save time and reduce hand cramp (really!), however there is always the chance that the screw thread may be malformed, so I do at least one test with proper screw connectors.

Returns are expensive and that cost needs to be built into your business model, so testing is crucial: consider that as soon as you have a return, you may well have burned your profit on the sale.

My worst facet is that I struggle to let go of units that fail testing. Say you have a unit that costs $35 to make. How long should you spend getting it to work? The answer is not long, and 95%+ of the time I come across failures that I've seen before and can be reworked in a few minutes. But then there are bastards that aren't so easy, often they have multiple faults. which confuses things. I might have three or four goes at fixing them, perhaps three or four hours of work. That is a rotten pay rate!

For one product in particular I have a small box of fails, perhaps 30 units out of 22,000 that I just can't get to work fully to spec but I live in hope that one day maybe I will. What I should do is throw them away!