what is the typical new product design process ?

[jayachar88]

Hi,

Hope that the professionals here can help with suggestions, pointers and guidance. I would like to understand the typical / usual "new product design" method / process, involving embedded computing. Please bear with me, for a long text as I try to explain what I've understood so far, and what I'd like to know.

We need to design a console which interacts with users (through a small graphical touch-screen), plays out some audio messages etc., interacts with some electro-mechanical devices (actuators/sensors) etc. Pretty typical of an embedded system, nothing too fancy. Given our requirement, we've zeroed down on ARM926EJ series processor, because those seem to be the most popular, meet our requirements and closest to the target price-band. However, we would prefer to take the SBC route if we can (as we don't have board-level HW design expertise). Why we don't go ahead and do so, is because I do not know if this is the right approach for something which you hope to sell commercially in large numbers (say 10,000 units over a year -- of course, optimistically speaking). Another challenge / concern is, the SBC today costs me about $100 at source (retail), after shipping/import-duties, it comes to about $170. Add some of the sensors, actuators, a casing, packaging, cables, manuals, battery-backup, the system price is already close to $300, without even including profit-margins. I believe that I can sell this only if the total system cost is around $150 (incl. my profit margin). Looking through the various volume pricing tools (s.a. at Mouser), I get an impression that most electronic components in volume, of 1000 units can easily attract a 50% discount (without further negotiations). So, my single-unit BOM of $300 should be doable at $150, including a some profit margin. Is this a reasonable expectation or that is a very naive way to look at things [won't be surprised].

Also, what the SBC's offer today, are lot of extra peripheral connectors, which I am very sure that I don't need, and yet, some modules which I need to integrate (which SBC's don't have by default). Have tried to speak with bunch of ARM SBC design-houses, they are not willing to talk to me (since I am not a big company, and can't place a 1000 unit order today). The ones that talk give very wishy-washy answers on how much can I bring the per-unit cost down in 1000 unit volumes, if I customize the SBC. Is such a thing even possible / usual ? I can understand that it is not easy to give a firm response on cost of customization without a deep study/evaluation, and cost of final product per unit -- but for people who do this for a living, day-in-and-day-out, I think it's not that difficult / usual either ! So, what is the ideal way to approach this and make some real progress ?

Our constraints --
1. We are a very small, very new startup (staff strength less than 5 at the moment, less than 2 months in existance).
2. Our ability to make large upfront investments or commitments isn't there.
3. We have some light expertise in embedded HW (but mostly hobbyist level -- with Arduino, so not really hard-code), but have loads of software development experience, thus HW integration we can do in-house is very limited.

Thanks in anticipation.

regards,
Jayanth.

 

[andre_luis]

The first step before decide isert new product in market, is perform a business case study.
You need know the capacity of absortion of your equipment by market.

+++

 

[jayachar88]

@andre_teprom, appreciate your response.
We have conducted a fair bit of market-survey, including studying the demand for alternative products and are confident of some numbers.
So, what are the next steps ?

---------- Post added at 20:17 ---------- Previous post was at 20:03 ----------

@andre_teprom, on 2nd thoughts... if there are some suggestions on how to do a better job of 'business case' study, they are most welcome.
We understand that the kind of business-case study done for software products, may be quite different where HW is involved, because of the order of magnitude difference in initial investments, impact of wrong decisions early-on etc.
Our study so far has been based on study of market for similar products which are semi-automatic, products addressing similar problem in another domain, sold at similar price-points with similar alternative products etc. We've also done a statistical survey using interviewing/questionnaire method. We also have access to some market research reports.
Coming from a software background, much of modern web or web-centric products follow the mantra of "release early - release often", which is in some ways an iterative market validation in itself. I believe, such a thing won't be possible where embedded HW is in the picture. So, we are limited by our knowledge/experience in how to do a better business-case study. We are willing to do it further, but wanted to reiterate that it's not that we haven't done it .

 

[keith1200rs]

I would certainly look at a custom design for those quantities. If it saves $10 and you make 10,000 then that saves you $100k.

Also, pick the processor for reasons other than 'it is popular'. Cost effective is probably the most important criterion.

Keith

 

[jayachar88]

Thanks @Keith. The arithmetics & economics of using design-services is absolutely clear. However, if I may -- I'd like to draw your attention to our constraints , i.e. small/new startup, limited funds, and bunch of guys with very little HW-design experience, and to top it all, the multiple custom design-shops I've tried to talk to, end up in lack of interest from the other end. Umpteen followups, and still I can't say that I've got a good lead which looks hopeful. Anyhow, shall keep trying.

My main question (to put it in a nutshell) is to figure out, what is the right / usual approach of working with / engaging such design houses ? Do they work on upfront payment basis, retainer basis, signed-up agreement on volume-purchase (since many of the design-houses actually like to sell you boards, not their designs).

Largely unrelated to my original post, would also like to understand how does the design licensing work. If it's a custom design, do I get exclusive rights on it's usage, or does the design-house retain rights to resell this design to others (including my competitors) etc.

regards,
Jayanth

 

[keith1200rs]

I can only really comment on what I would expect in the UK. Most design houses would work for upfront money and you should own the design when the job is complete and they have been paid. Unfortunately there are plenty of companies around who are not competent and also quite a few who will try to retain some ownership of the design which to my mind they shouldn't expect if they have been paid 'the going rate' for the design.

You may be able to find a manufacturing company who will do a custom design for nothing provided you commit to a certain quantity of the board. The problem there is that you may not be able to commit (financially). Also, they would own the design. I know of a reputable Chinese manufacturer who will do the designs for almost nothing provided there is $1M/year of potential production and they are prepared to take the risk (i.e. they don't need a firm commitment). They are unusual though.

I think that unless you can find someone to do the work for next to nothing in return for a share of the potential rewards your best bet would be to start with a standard board with the intention of replacing it with a more economical custom board as soon as you can afford it (and assuming you can find a decent company/person to do it without ripping you off).

Keith

 

[horace1]

Although I have used SBC modules (Microchip explorer 16 in this case) in commercial products the system cost was of the order of £10K and speed of getting a working system was critical.
For a low cost unit I would usually be looking to design a PCB but I can see you problems if you don't have the expertise in this area.
I think you should look at your specification and determine a low cost system that would support the requirements - certainly not a $150 per SPC.
For example, Microchip's Microstick is low cost (typically $25 in small quantities) and should meet your requirements
**broken link removed**
It has an onboard programmer/debugger and includes a header than can plug into a PCB or breadboard that would connect to the external devices.
We use it extensively for undergraduate and postgraduate projects in engineering and computing.

 

[andre_luis]

jayachar,

The constraints are clear, but you didn´t summarized the questions, so I will point somethings :

> It is very important that persons involved in some project be fammiliar with target area product ( commertial, technological, etc... ). There are particularities that we commonly discover too late.
> Production in large scale, is not only multiply by a scale. There are a lot of concerning when you work at large scale. A Production spacialist in electronic may show how to minimize human interference and reduce cost. There are details from logistic, management, contract, testing, assembling, etc that can be considered on development. The project itself is only part of care.

+++

 

[keith1200rs]

Boards like the Explorer 16 are great for development but I wouldn't use them in any significant quantity - they wouldn't be cost effective. However, if you develop your prototype using something like that it is pretty easy to then turn that into a custom board - Microchip provide all the circuit diagrams and you just keep the bits you need. The connector on the Explorer 16 is pretty much the same as the micro I/O pins so conversion is easy. I have developed a product that way - Explorer 16 with a custom add on board to demonstrate the principle then a custom board which combines the PIC and custom circuitry on a single board for the product.

Keith.