I will show you how. (It's all about fit and quantity).

Let's say your product contains a Commercial Off The Shelf (COTS) controller, for instance a PLC or a single board computer like a PC104 product. Say now you need some special added functionality, like humidity measurement. To get that you will have to locate and pay extra for a special purpose product to fill that need. Something like a humidity transmitter could cost more than the controller itself.
 
If on the other hand the special function could be integrated directly into a custom controller, it may cost very much less. Not only will you be saving on the raw function, you will also save on mounting, wiring, space and inventory.

A real example

This is a real customer (let's call them Acme Apple Processors - AAP) who manufactures a product that contains a custom SPLat controller. I cannot identify them, but I can give you some real numbers.
 
AAP's controller, which we make for them, has the following attributes (this part is real):

1. An integral humidity and temperature sensor;
2. One analog input, 0-10V, for an external setpoint variable from a potentiometer;
3. One analog output, 0-10V;
4. Four digital inputs from external control switches;
5. Two relay outputs rated 2A @ 24VAC for driving external contactors;
6. Expandable via our full range of I/O expansion boards;
7. Able to perform an elaborate set of mathematical calculations combining environmental readings with setup constants via several cascaded formulae incorporating transcendental functions;

8. One 8-way DIP switch for setting up site related parameters;
9. A companion program running in a PC which was invaluable for monitoring performance, adjusting operating parameters and logging long term performance data during product development;
10. Push-fit terminal blocks for quick and easy wiring (could have been whatever suited AAP's manufacturing preferences)
11. Three LED indicators for diagnostic signalling;
12. Contained on a single board about 3.5" square;
13. Runs off 24VAC because that already existed within AAP's product design.

The following estimates are based on the lowest prices I typically come across in my many wanderings around the Internet. As such they are not "real" but I do believe they are representative. More on this later ...
 

The control system built with the above components will fall short of the customer's needs on points 8, 10, 12 and 13. Point 11 may or may not come out in the wash, perhaps by using the indicator LEDs on otherwise unused outputs. Point 6 is considered met if the PLC is expandable.

The custom controller we built for AAP uses a high quality temperature and relative humidity sensor component, the Sensirion SHT71 This is mounted directly onto our small board, which in turn is cleverly integrated into the end product so the air being measured passes by the sensor.
 
In production quantities (some hundreds) the controller, complete with the sensor, costs AAP less than $150. It meets all the requirements of the "specification" above.
 
The initial NRE charge was under $20,000 for the hardware (I'm taking programming to be about the same cost as for a PLC)

Nope! Not unless you consider some of the world's most seasoned electronics engineers in Melbourne, Australia to be off-shore. If you are in the US, I guess it is. Sort of. But we do speak your language, and what's more you can talk directly to our designers. The $20K tag is not because we work for a pittance. It's because we have been doing this for a long time, and our SPLat technology provides a well defined framework and a set of boilerplate design elements. That means we can make up a design with comparatively little effort compared to starting from scratch with a blank CAD screen. Also, it's the ongoing production we are after, not selling our time for $x per hour.
 
By the way, our approach reduces the risk factor. Our designs are more like re-mixes than full custom designs. Our success rate is 100%. We have been in this business since 1981, and intend to stay in it for a while longer.
 
Our team (Christmas 2008). Left to right: Fred (Boston office FAE), Keith (Director, Engineering), David (Managing Director), Rob (Testing), Richard (Software engineer and chief barrista), Alan (Firmware engineer, 1-man skunkworks), Chris (Testing, tech writer), Judy (shipping), Carla (Director of Manufacturing), Angela (QA manager), Linda (Accounts), Adrian (Software engineer).

For this set of example figures we can work out a break even point, i.e. answer the question "how many must I build before I have repaid my NRE investment?"
 
The saving on each unit is $475 - $150 = $325 (68% saving)
 
Divide that into the NRE: $20,000 / $325 = 61.5 units

It is very hard to provide firm guidelines on economical quantities. Our cost structures are such that the unit price will rise quite steeply on small batches. You would not get the $150 price cited above for batches of 50 or 100. On the other hand, for batches of 5,000 the controller in question would be well under $150.

Good question!
 
From the point of having a totally clear and agreed upon hardware specification, we can generally ship an "alpha" prototype in 6-8 working weeks. Allow 3 weeks for customer testing. After that a small number of "beta" prototypes might take 5 weeks. So the total is in the order of 16 weeks to stable prototype hardware.
 
In our experience the customer is never, if ever, waiting for us. All the other aspects of product development mean we are frequently left waiting for the customer to be ready to test.

If you have a product or project you think may benefit from a custom SPLat, please contact us. We are happy to evaluate your needs. click here to go to a questionnaire that will help you provide the kind of information we need. Alternatively you can read more about our structured process in our knowledge base here.

The above PLC costing is an estimate. The PLC cost is based on looking a several models of PLC from a low cost online (direct sales) vendor. The biggest challenge for the PLC is that it needs analog and it must be capable of some reasonably serious floating point math (or be able to use several interlinked lookup tables).
 
For the humidity sensor I found 3 competing models online, selected the cheapest at $199 1-off and halved that figure to allow for quantity discounts.
 
I believe my estimate probably errs on the low side.
 
I welcome any alternative costings, and will use the lowest one received providing it will fulfil the same functional requirements and all prices and model numbers are verifiable online. I will accept pricing for quantities up to 1000 per order.
 
Any solution based on Commercial Off The Shelf (COTS) products is acceptable. It doesn't have to be a PLC, but it must be COTS.
 
I'd like to hear from you. Just use our contact form to send in your costing.
 
This challenge was first posted 20th October 2009