Rumors have cropped up again Friday about manufacturers preparing a CDMA version of Apple’s iPhone 4, which a range of lazy journalists are using to speculate that the iPhone is coming to Verizon Wireless sometime soon.

But even if Apple is building a CDMA iPhone using Qualcomm’s chips, that doesn’t make it a Verizon iPhone. According to the CDMA Development Group, there are 100 different carriers using the most advanced CDMA EVDO Rev A technology. Verizon is currently the largest with about 90 million subscribers, but China Telecom is neck and neck and coming up fast.

Other major CDMA operators include Sprint here in the U.S. and KDDI, with 32 million subscribers in Japan as of August 2010. A collection of CDMA operators in India had more than 120 million subscribers total earlier this year. Even little MetroPCS wants to bring a budget iPhone to its 7.6 million subscribers.

In other words, preparing a CDMA iPhone may just mean that Apple now feels comfortable enough to run two parallel phone development efforts, not that it’s suddenly closed a deal with Verizon. Now, I’m not saying they haven’t closed a deal with Verizon, just that eliding automatically from “CDMA” to “Verizon” without additional Verizon-specific evidence is lazy and wrong. I’m pretty sure that for a while now, the lack of a Verizon iPhone hasn’t been about technology, but about two companies that can’t come to terms because each one wants to be the leading player in any relationship.

While various studies say Verizon could gain subscribers by offering an iPhone, Verizon has been very successful selling its Droid line without the hassles of dealing with Apple. Verizon isn’t desperate, and nobody at the carrier wants to become a “dumb pipe” for Apple’s experiences if they don’t see it as absolutely necessary to gain market share. Apple, meanwhile, has been selling iPhones like hotcakes. That means for both companies, the other is a “would be nice to have” rather than a top priority partner.

I’ve always been bearish on the potential of a CDMA iPhone because CDMA is a dying technology, and Apple hates dying technologies. Apple is usually the first company to cast off old standards and adopt new ones, whether that means ditching the floppy drive on early iMacs or knocking out DVI ports in favor of Mini DisplayPort. While CDMA networks will be around for the next ten years (and may potentially grow, especially in the developing world), cutting-edge CDMA operators are looking to 4G LTE as their next step.

As recently as yesterday, Verizon CEO Ivan Seidenberg shot down the potential of a CDMA Verizon iPhone, encouraging journalists to think of any Verizon iPhone as a 4G LTE device. That’s what I’ve thought for a while, too.

Pegatron Technology is expected to begin volume production of a CDMA-version iPhone 4 in November with total shipments likely to reach 3-4 million units in the first three months, and 10 million by mid-2011, according to industry sources.

With shipments of iPhone 4, and orders from Asustek Computer, Acer, Toshiba and Sony as well as tablet PC orders from HTC, Pegatron is expected to snap up handsome profits in the first half of 2011, the sources noted.

But industry sources have predicted that Pegatron’s gross margin and utilization rate for the third quarter of 2010 will fall below its targets.

One importance between CDMA and TDMA is that

In CDMA, the UE transmits and receives continously during all the frames and slots.

In TDMA the UE is only active during one slot in a frame.

Then, during the rest of the slots, the receiver can be tuned to other frequencies and perform measurements.

With CDMA, this is not possible to make interfrequency measurements unless the receiver structure is duplicated.

Do spread:

symbol sequence X  (XOR) some code Y => some signal Z

Do despread:

some signal Z (XOR) some code Y => symbol sequence X

—————

2. Let’s assume there are 10 users who want to send its data X1, X2, …, X10.

Let’s assume they have their own codes Y1, Y2, …., Y10.

Let’s assume they send data at the same time.

Let Z be all summation of Xi (XOR) Yi .

I’m only interested in X1 signal. But, how can I extract X1 from all summed signal Z ?

If Yi are othogonal (i.e., Yi XOR Yj => all zero if i and j are different), then

Z (XOR) Y1 => X1

This is the fundamental concept of spread spectrum.

Second, remember that a chip duration is fixed.

Third, what is symbol? Each symbol can represent some amount of bits. e.g., QPSK, each symbol represents 2 bits.

SF: 2 SF vs. 512 SF
Time: 2*delta vs. 512*delta
Data: 1 symbol vs. 1 symbol

where delta is chip duration.

With 2 SF, you are sending symbol fast (my wife cooks very fast, so she can make lots of food)
With 512 SF, you are sending symbol slow (I cook very slow, so I can can make less amount of food than my wife)