ZigBee - Make or buy? Introduction

ZigBee - Make or buy? Introduction
ZigBee® - Make or buy?
The life cycle cost of a ZigBee HW solution
Introduction
ZigBee is the only global wireless communications standard enabling the development of
easily deployable low-cost, low-power, monitoring and control products and the interest to
add ZigBee capabilities to products are increasing rapidly. There are already hundreds of
hardware products and software services enabling the market.
Adding ZigBee capability to a product involves designing both the appropriate hardware
(HW) and the firmware (FW) including an application. This paper addresses only the HW
development of ZigBee solutions.
Because of the cost and risk involved in a chip-based design, many companies are reluctant
to add a new feature like ZigBee to their product. The paper will show how adding ZigBee
can be done at minimum cost and minimum risk. For companies starting to develop ZigBee
end products now, the time to market will also be an important factor.
There are both numerous chips and module available for ZigBee. What difference the
modules from the chips is that they contain a small PCB with all necessary components (One
piece solution). The RF design and layout (often including antenna) is already done and the
module have been manufacturing tested and tested towards conformance to standards.
The final evaluation of make or buy (chip or module) is left to the implementers. This paper
will help the decision makers to get the full picture of the cost and other relevant factors, and
therefore be able to take a better decision.
The Myth of chip cost
Today a developer can choose between different degrees of integration of the required RF
hardware; from a dual chip solution, via single chip solution, to a module solution. Many
believe in the myth that basing their design on a single chip always will give lowest cost, but
for most applications this is not true. When starting to calculate the life cycle cost of a ZigBee
HW platform one will see that the “single-chip” itself is only a fraction of the total cost. Even
so-called single-chip solutions require several handfuls of carefully chosen passive
components, critically specified crystals, antenna including its matching, shielding can etc., in
addition to a very careful layout on a high-frequency board or substrate to give the promised
performance.
Cost
The cost of the ZigBee hardware is divided between the bill of material (BOM) and the
engineering cost associated with the development, manufacturing and testing over the
products life cycle.
The BOM of a ZigBee HW solution is not the same as the chip cost. In addition to the ZigBee
chips (RFIC and MCU, or a SoC) there are crystals, antenna, balun, shielding can and a
handful of passive components. Depending on the solution it might also include flash,
EEPROM, power amplifier, low pass filter etc. Adding all this can make the BOM cost 50100% higher than the pure chip cost.
In addition to this, there will often be handling cost associated with each component. This will
include shipping, logistics, etc. and will give a small cost increase per component.
Development cost
The development can be divided in the pre-study phase and in the engineering phase. The
pre-study includes specifying the performance and functionality the ZigBee HW shall have.
The pre-study will also include making a component survey. Which components are
available, what is the critical performance, what is the size etc? When datasheets are
hundreds of pages, this work can be time consuming. In addition a purchasing engineer is
needed at this point to collect price offers and negotiate prices.
The engineering phase will includes schematic, RF layout, prototype manufacturing
(including component purchase), and qualification. In this phase the RF competence of the
engineers will determine success or failure. At 2.4 GHz every line width, pad size and
distance to ground matters and the layout should not be left to layout engineers with a pure
digital design background.
In the qualification phase good RF instrumentation is also important. Companies without RF
experience and RF instruments often ends up developing the ZigBee HW at an external
design centre.
The work of the design engineer is not over when the product is in mass production. There
will be support activity towards production that includes: evaluate new and cheaper second
source components, analysing yield, process changes etc. This will not be a full time job, but
still a cost to be reckoned with.
Development cost is a one time investment and when calculating the development cost per
device, one tends to believe that a design lasts forever. As ZigBee is a relative new standard
and there are many companies working on new chips, so the lifetime of a ZigBee HW
solution should not be expected to be more than 2 years. Today’s chips might be available
longer, but in two years there will be better and cheaper solutions available.
Testing – the hidden cost
Testing is a cost that often is forgotten when making a radio design. Test cost includes the
test development cost, instruments and the actual testing of each device. RF instruments are
expensive and will costs $10k-50k, even if bought refurbished.
For smaller volumes a test engineer can manually test each device, but for large volumes
there is a need for an automated test system. The development of such a system can often
be as complex as making the ZigBee solution itself. The test times per device should not go
above 15 seconds and the test coverage (percentage or errors discovered in test) should still
be above 95%. This will require an advanced test jig and also carefully written test program.
Many believe that a simple functional test will be sufficient, but this will lead to many products
in the market with poor performance.
Chip solution vs. module
Adding all the costs related to making a ZigBee HW solution with the BOM-cost we can
compare a chip solution with a module. For a low volume customer the number could be as
shown in table below.
BOM
Test cost
Yield loss
Chip
Module
~$ 12 (1k)*
$1
$ 0.3
$ 21 (1k)*
$ 0.2
0
Life support cost
$ 20 000
HW Development cost
$ 150 000
Test development
$ 100 000
Incl. RF instrument
Total per device(1k)
$ 283.3
* The numbers are based on published internet prices
$ 1000
$10 000
0
$ 32.2
The same calculation can be done for higher volumes and the relationship between total lifecycle cost per device and volume are shown below.
Chip
$100
Module
$10
$1
1k
10k
100k
1M
10M
Figure 1 Estimated LCC for chip vs. module solution
A good RF performance
Adding ZigBee hardware to a product does not only involve the risk of faulty functions, but
also the risk of poor RF performance. To get the specified range, a good antenna design, a
good grounding and a good RF layout must be achieved. If some of these fail, the
performance will be poor and the product will not be a success.
It is not only the performance that is critical with regards to the RF design. It’s also the
conformance to standards. In Europe your device must be according to the R&TTE directive
with its underlying EN standards and ERC recommendations. In USA the FCC sets the
requirement the wireless devices in the ISM bands. In addition to understanding all these
standards, each design must also be verified against them. For FCC-compliance these tests
must be done at an approved test house.
Yield is another factor that can influence the life cycle cost. Analogue designs and
components have an inhabited variance and the higher the frequency, the more difficult is
this variance to control. If the radio design doesn’t deal with this tough variance simulation,
analysis and appropriate margins, the result might be that the number of unit that fails during
test is high.
If one of hundred units fails in test, the fault must be identified repaired and the unit retested.
It this operation cost $50, a 50 cent extra cost is added per unit.
Based of the above factors the importance of getting a good RF and antenna design can not
be overstated.
Time to market
A short time to market is a critical factor in order to be successful in this business.
Developing RF hardware is an activity with risk and a long lead time. An RF design with
development and prototype manufacturing and testing will often take 6 months or more. If the
performance is poor additional reruns might be needed. Such a long design time might be
the difference between market failure and becoming a market leader. One extra week in the
development process means 1 less week presence in the product market.
RISK
All HW developments have risks associated with them. Will the RF performance be as good
as you specified? Will the design conform to the standards? Have you chosen the right chip?
In addition to this there is the business risk of your own product: Will the new product with
ZigBee functionality be a market success?
These factors might leave a manager reluctant to use a large amount of money to add
ZigBee into the end product. When faced with this risk, it might be more appealing to the
manager to add a ZigBee module into the product with much lower initiating cost (<10%) and
lower risk.
Making your own module
When a company introduce ZigBee capability in their products, it is usually not only one
single product, but a line of products involved. It is inefficient to do the ZigBee HW design
with critical RF-layout more than once. They want to do this part once and reuse the design
in several products, also for certification reasons. And the best way to do this is by making a
module that can easily be copied from product to product. Therefore companies often end up
with their own module, even if they initially decided to make a design based on single chips.
The argument that the module level adds to the cost, does therefore not apply.
Business models
Starting with a module does not prevent changing to a chip solution at a later stage. When
the new end product has been successfully introduced to the market and the sales are
increasing, then it is the time to look at optimizing unit cost. By employing this strategy the
best from both modules and chip-based solution can be achieved: Low risk and development
cost from modules and still getting the lowest cost for large volumes.
But optimizing cost will not necessarily mean throwing out a module and starting from
scratch. Working together with the module provider still might have strong benefits. By
continuing to work with the module manufacturer when entering high volumes there are
several paths that might give the best overall solution:
-
-
-
High volume discounts
o When you have an established product in the market and the prospects of
high volume arise, modules with volume discount might give the lowest
total cost.
Design-in of the module on your board
o This would include copying the module design onto your board and adjust
the design. This could dramatically reduce development time and cost, but
would normally require some sort of module design licensing
Optimized custom module
o Adjusting the functionality of the module to best fit the application
Design buy-out
Conclusion
This paper has investigated the LCC cost of a ZigBee hardware solution. It has shown that
the cost of the single-chip itself is only a fraction of the total cost and that for low and mid
volumes a module would give the best solution with regards to total cost, time-to-market and
risk. This paper has also shown how modules could ease the market introduction even for
high volume (>100k) customers without compromising cost per unit.
ZigBee® is a registered trademark of the ZigBee Alliance
Author:
Ørjan Nottveit, project manager at Radiocrafts
Contact Information:
Web site: www.radiocrafts.com
Email: [email protected]
Address:
Radiocrafts AS
Sandakerveien 64
NO-0484 OSLO
NORWAY
Tel: +47 4000 5195
Fax: +47 22 71 29 15
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement