First kit - appropriate waiting time

[sharp]

Hey everyone.

I am in the process of designing a pretty ambitious project for a first kit. I have the thing prototyped on breadboard and am in the process of writing firmware. I have the schematic laid out in geda-gschem, and am using the same suite of software to produce a PCB.

Anyway, I am guessing that when it is done, it will cost ~$20 for all the parts from Mouser, including power supply - but that doesn't include the price of the PCB, which is looking like it will be about 2"x10", double sided, one sided silk-screened. My question is this: if I am a college student who is relatively broke when it comes to spendable cash, how acceptable is it to make folks wait for their orders. For example, is it ok to wait until I get 20 orders before putting the order in with mouser and whomever I will use to do PCB fab? (I would of course be up front with the customer about this.)

Most of my interest is in this particular kit, but would it make sense to start with something smaller to fund this one?

Thoughts?
-sharp

[adafruit]

in general, its a bad idea to take people's money before you have something in hand to sell. so we dont recommend it.

[sharp]

Makes sense. In that case, how possible is it to start very small? Like ordering 3-5 times the amount of kit of of mouser, and then having only 3-5 kits in stock. Is that a bad idea?

[sharp]

Or perhaps some kind of short-term x0xb0x style waiting list?

[adafruit]

sure

[chuckm]

I've got a few thoughts on this.

The first is along the lines of prototyping versus production. There are two big factors to consider when 'kitting' a build (this is the process of making all the parts available to build the finish product, in your case if its a kit anyway then there is no 'assembly' step). The first is lead time on parts, which is to say that some parts have really long lead times when a distributor runs out. Flash is notorious for this and if you try to buy flash chips just after Nokia or Motorola has started production of some new handset you may find lead times in the months rather than weeks!

The second is that price goes down hugely when you order in quantity rather than singles. Lets take for example our favorite chip, the AVR ATMega 328P, soul of the Arduino and its clones. If you're building a kit that has a '328 in it, and lets say your buying from Digikey (a pretty common distributor) then buying for 5 kits and the 328P costs $4.30 when you buy 1 to 5, but it only costs $2.86 when you buy 25 or more. That's about 35% off on what is likely one of the most expensive chips. Also consider the crystal, a 16Mhz parallel cut crystal the CTX-415 (low profile, through hole) is $0.75 each but in 10 or more its $0.642 (another 15% savings on a part for buying 10) if you buy 50 its only $0.535 or a 30% off.

The third thing is that the cost of making a printed circuit board at a production house (as opposed to a one-off shop like Olimex or APC) generally has a large "setup" charge which is the cost to make films and to 'panelize' your board (make the maximum use of the board stock), and then a per-board cost which is usually quite small. Your total cost will be ($setup cost + $per board cost * Num of boards) / Num of boards, and if you've taken Calculus you can see that the limit as the number of boards approaches infinity the cost per board approaches the per board cost. Another aspect of manufacturing PCB's is that the manufacturer passes on the cost of "waste" board so if your design panelizes evenly into a single blank you will get the best cost per board.

What these add up to is that the cost to you of making 25 versus the cost of making 5 may literally be 2 - 5 times. I've got a relationship with a board house in the Bay area and I can tell my rep what I'm targeting as a per-board cost and he can tell me at what volume I'll achieve that cost. He will also keep my films and tool files on file so that I can come back and get another batch run without paying any setup fee at all.

Given all of that, here is my advice (money back if you don't like it : Price out the quantity discounts for making 10, 25, and 50 kits. Tell your potential customers that you're going to make them if you can get enough people signed up, and collect pledges (no money, just a promise to buy one when you make the run). Once you have enough pledges (some folks might buy 2 or 3), send out an email to your pledgers and say its a go, please reconfirm that they are in on the purchase.

At this point, if you can swing it, you place your orders. Once your stuff arrives you start kitting it, sticking it into an envelope and charging your customers (or cashing their checks, or depositing their cash). Its generally best on only charge/deposit when you can also show proof of shipment.

If your selling your kit is for $20 and you need to build 50 to get the best price, then you'll need a credit card with a credit limit of at least 50 * $your_cost.

--Chuck

[stbtrax]

I know everyone is saying not to do it but I have seen it done successfully. Successfully meaning that this 1-man operation is getting lots of orders, and seems to be fulfilling them, not 'successfully' in the sense that it is the best strategy. He has about a 3-month waiting list for his niche electronics. But he does have a pretty good reputation in his community so I am not sure how that will work for you just starting out.

[sharp]

Thanks Chuck, that helps. Perhaps when my prototyping is done I'll use the "pledges" idea. Maybe once I get 25 or so signed up, I'll fill those orders. If I wasn't too stressed about kitting that amount, and others are interested, I can fill bigger orders on my waiting list, and move somewhat exponentially from there.

That's if I ever complete this prototype. I wish my town had better electronics stores.

-sharp

[chuckm]

That's if I ever complete this prototype. I wish my town had better electronics stores.

Don't feel too bad, I live in the heart of Silicon Valley (Sunnyvale) and there aren't any decent electronics stores.

--Chuck

[mctaylor]

In my opinion, your design is still in prototyping / debugging stages until you have beta testers build the kit from small-run prototype boards with your beta instructions.

The prototype boards can be homemade or bare bones without the solder mask and silkscreen labeling to keep your investment low.

What I would suggest is, if there are no obscure or difficult to find parts, don't do a entire kit, either do a partial kit (pre-programmed MPU, any hard to find parts, and PCB) or just a production quality PCB (with solder mask and silk-screening) for your first venture, and re-invest the 100% of the "profit" back into the business as capital for future projects. I would expect that you would likely need 25 boards to be practical (that's a general assumption based on 'gut'), and ideally hope to sell 50-100+ if the interested customer base is large enough. For example, if you can start with a kit that is appealing to either a typical Make magazine reader, Hack-A-Day follower, or amateur (ham) radio builder, you are more likely to be successful, IMHO.

I know some of the kits from non-profit amateur radio clubs offerings have 2 or 3 prototype versions of the PCBs sent out to beta builders (~2-5) before offering the kit to the wider community.

To me personally, the reputation gained from doing a small(er) well-done project with no (or minimal) waiting lists with high quality instructions, testing procedures, and good customer support (both technical and sales / shipping) is worth a lot more in terms of reputation than having too many Oops trying to overextend yourself with a major project that you can't afford to screw-up.

Having a robust reliable design with accurate and clear instructions is worth a lot. Personally, I could build most of the kits AdaFruit offers myself, but the convience of having a tested design, a correct PCB, saves me my most valuable resource, my personal time.

From a business point of view, this is considered "sunk cost" of development or NRE (non-recurring expense) that is an up-front cost regardless of how successful a product is, but good development reduces the ongoing costs, both in manufacturing and customer support, returns, and repairs.

[macegr]

Don't feel too bad, I live in the heart of Silicon Valley (Sunnyvale) and there aren't any decent electronics stores.

You don't count Jameco, HSC/Halted, Weird Stuff, Ace Components, and Fry's? This is as good as it gets.

[chuckm]

Don't feel too bad, I live in the heart of Silicon Valley (Sunnyvale) and there aren't any decent electronics stores.

You don't count Jameco, HSC/Halted, Weird Stuff, Ace Components, and Fry's? This is as good as it gets.

Hmm, JameCo lets see do they stock an Atmega 328P? Nope. They have Z80's though, go figure. HSC/Halted (well Halted went out of business leaving HSC (the other brother)) which recently dropped half their retail space. Now don't get me wrong if HSC has what you want you'll get a good price, if they don't have it they may never have it. LM555 timer IC? Opps not this week check back next week. Its one of the challenges of being in the 'surplus' business, they only get parts if someone decides to dump their excess and most places don't do a lot of through hole or even 603 surface mount. Weird stuff has no components, Ace Components has retreated to Oakland with narrow business hours and only surplus parts, Frys? C'mon, they were great when they started, now they mostly have NTE (for those of you who don't know NTE stands for Nearly Twice as Expensive) parts. Radio Shack is competitive with their one rack of 1/4w resistors and switch assortment.

Sunnyvale Electronics, Quemont, Jade, Advanced Products, Quest Electronics, Zacks, all dead. I'd take a Radio Shack from the 80's
http://www.radioshackcatalogs.com/catalogs/1983_small/ even. Sad sad sad. Lets say you wanted to build an Arduino from "parts" and a breadboard. There is no where in the bay area where you could buy all the parts. This makes me sad.

--Chuck