I am a beginner in electronics design. I have some experience with fairly complex PCB design. I want to design a product that will hopefully sell a lot. How do I ensure that the design is inexpensive from manufacturing perspective? I mean not for a single PCB fabrication but for mass production. I use commonly found microcontrollers like atmel, Texas instrument. Is this the way to go for mass manufacturing?
Brand based answers are usually frowned on BUT Microchip with their PIC microcontrollers do a superb job of not obsoleting hardware products that are decades old. If you look in their catalog and selection guide pages you will find products that are not recommended for new designs but which they were making in the early years of their operation. Upgrade paths are also often relatively pain free compared to some alternatives.
PIC processors have for decades had a very large % share of the bottom end market - in equipment like cable TV set top boxes, toys appliances and more. In Asia they are competitive against direct contenders. For high volume manufacturers I understand that they offer untested products at a significant discount, allowing suitably capable manufacturers to run their own acceptance programs.
For an entry level processor in high volume you will certainly be able to get standard PIC parts down to around $US0.20 and I imagine substantially below that - but I have no way of accessing such information as it would be on a 'need-to-buy' basis.
Another approach is to look for generic processor families that are manufactured by a range of suppliers. eg Various 8051 variants
Atmel are specifically not good at maintaining long product lifetimes.
Update - January 2016: It has just been announced that Microchip has arranged to buy Atmel, so it will be interesting to see what changes this causes. These could range from discontinuation of some Atmel products, through to improvement of long term availability.
You can get lots of 50 or 100 made, but it will be more expensive than lots of 1000 or even just a few 100. Unless you have a previously established relationship with a manufacturer in China, and you are happy with that relationship, I would not try to get your product made there. It takes a lot of time and effort to set something up that you can eventually rely on. It may sound easy, but it's definitely not. Going to China is not cheap unless you know what you are doing, have the time and patience and expertise to invest in getting good manufacturing set up for yourself, and are willing to get personally envolved. I'm sure Russell has a lot to say about this. He has more experience with this than I do.
Frankly, from the little you've said it doesn't sound like your business case makes any sense. This sounds like a fairly simple device. The chance that you have stumbled upon a major need that nobody addresses is quite small. This is therefore a niche product at best. How much are you going to be able to mark up each unit? $20 maybe? That means at best you make $1000 profit on your whole lot of 50 boards. How much time did it take to design the circuit, enter the schematic, lay out and route the board, create the BOM, test the result, write the firmware, etc? Now you're talking about assembling them yourselves, which will include purchasing and kitting. Each one has to be programmed, tested, and possibly calibrated depending on what these things do. Divide $1000 by all the time you have spent and will spend doing the build, and you'll be paying yourself pennies per hour. If you can't see at least 1000 of these things getting sold, this is a waste of time unless you're doing it for the fun, challenge, entertainment, etc. If so, the method of production doesn't make a lot of difference.