Board Assembly. In house or outsource?

T Chap

I have a product that is low volume, high price, so it is not a simple issue to send out boards at the moment. It requires from 5 boards to 7 depending on options. One of the smartest things I have done was to develop a software and CNC mod to do assembly which makes the process easy and cheap, but the system is manual feed, one tape at a time. It is simple and easy to build boards with it, 0603 parts mostly, effortless other than grabbing the parts as the app calls for them. But, to grow, I need to remove myself from the shop. I know some successful companies that have large 100k machines and they do everything in house, when it would be easy to send the boards out when they are spending that kind of money on machines. Then there are people that send boards out and go that route. If I had the money, I think my choices would be all in house. Design change? No problem. New board concept? Very little time and cost to test it.

One particular fellow I chatted with online on another forum owns a very large company that makes a popular video graphics card. He explained his entire process of how he grew his in house assembly operation, and even as CEO, he is/was very hands on with the assembly and soldering processes in great detail.

What are the factors in deciding? Ease of prototyping and design change? Cost per board? Complete control of product and QC? Right now, after getting quotes from Advanced and Screaming, it is not possible to farm out the work with either, but soon I have to make some choices and bite the bullet: hire someone in house or go out. Any insight?

WBA Consulting

The position you are in is frequently referred to as the "make or break dilemma" for a PCB based product. Building in house may cost you more upfront (equipment, learning curves, time), but can be very beneficial if your product will have a long lifespan with a steady run rate that you can maintain. Outsourcing provides you with many options that you cannot afford to do yourself, but increases the per board cost.

Some basic calculations can help you decide. Factor the relationship between your monthly/annual run rates and the expected lifespan of your product. With this, determine the cost of the life of your product by building in house and outsourcing. When factoring building in house, keep in mind the initial cost and resale value of the equipment. (in other words, if you buy the equipment, only factor the depreciation of the equipment of it's lifespan). Note: A contract manufacturer will have relationships with distributors that will allow them to get better pricing on fabs and components.

For example (values are plucked from my experience for this example):

• In house:
  • Equipment cost, $30k, $4k depreciation per year
  • Fab cost for board set: $40
  • Component Cost for board set: $100
  • Your "labor cost" per board set: $40
  • Annual run rate: 200
  • Lifespan: 3 years
• Turnkey Outsource
  • Equipment cost, NA
  • Fab cost for board set: $30
  • Component Cost for board set: $85
  • Their manufacturing cost per board set: $90
  • Annual run rate: 200
  • Lifespan: 3 years

So, lifespan cost is:

• In House:
  • Total cost per board set (40+100+40=$180) times
  • 200 per yr across 3 years (600) plus
  • Equipment depreciation (3 yrs at 4k=$12,000) equals
• $120,000

• Outsource:
  • Total Cost per board set (30+85+90=$205) times
  • 200 per yr across 3 years (600) equals
• $123,000

Ok, so I admit I massaged the numbers from my original "off the top of my head" because the numbers were a $11k apart at first. But, I hope I have helped make a point of all of the factors involved and how to look at long term costs to make a well informed decision.

For some other tidbits on outsourcing, check out this thread that has a lot of information.

ps. In regards to control of your product and QC, you will always have control when working with a CM. However, depending on the CM, there may be fees associated with changes made to your assembly that the CM must implement, you will be dependent on the scheduling of your builds in their factory, etc. Also the CM will be building to a standard (typically IPC-610) and you can hold them to that with any issues you may have with quality concerns. How they handle quality concerns should be a topic of discussion during your audit stage of a CM.
s

T Chap

Thanks for the in depth break down! Everything I have come up with over the past few years leads to think in terms of in house. This will be an ongoing thing for many years to come barring any unforeseen event, so I have to approach it for the long haul. Tonight while getting online quotes, there is no way to consider outsourcing anyway for the immediate future. One main consideration is that someone has to assemble the throughole parts, wiring harnesses, enclosure, screen print, etc. That means it is mandatory to bring in someone for those aspects, so I might as well train someone for the complete process. The point you make indicate, that at small runs there is no significant gain to outsourcing cost wise.

Erik Friesen

I used OURPCB to assemble 25 of these. It cost right at $25 each (assembly only). However, it was not without some issues, mainly with parts substitution, and some minor issues that were my fault. Some of the parts in question are SM resistors, I suspect they are about 5-10% tolerance. I have had on .1ohm shunt resistor on one of these boards burn out, which seems odd to me. Another substitution was completely out of line, I finally just told them not to put it on, it was a 3.8mm header connector that was cheap, and completely incompatible with what I needed.

The problem I face, is that most of the USA board house engines I put my information into spit out $125 - $200 per board(assembly only), for the big board only! I can HAND assemble (with a stencil) one of the big boards and pay myself decent wages(in this economy) rather than pay someone $200 dollars. To get the $125 price I had to order 100 boards.

T Chap

Nice work Eric. Yes I get the same rates for my largest board, 175 or so for several week turn around, quantities of 5 minimum. I can make it in an hour with little effort and no waiting, no multiple email exchanges with the assembly house, no dealing with mistakes due to miscommunication.

As far as SMT low value resistors like .1, or 1.0, if they are involved in some power related issue, I always double those up to double the power rating limitation, since SMT resistors rate well compared to TH versions. Now I only use TH 3-4watters just in case.

I think the bottom line is, if you are willing to roll up your sleeves and either build a system or buy a system, you are better off in the long run if the project is expected to have a life span.