This has been a very interesting read. I only know a bit of electronics but I can understand the distinction between a radio and it's frequency range and the rest of the stack above that, that would control what can be actually done. As for software controlled don't think software and firmware. They are the same thing. Both are software so yes it is software controlled in the end. The only difference is where they run and how they are coded up.

I hope I haven't given any more confussion to this thread.

SD69, I'm not sure its true, but I remember HTC had the 850 frequency locked in some WinMo devices, and the xda devs unlocked it, so its possible, but I have no idea myself.

I really appreciate you guys. I totally understand the story now. We get frequency and channel mixed up. From the looks of things, 2100 supports both bands, and the device caters to far more people than I originally assumed. It also tells alot about Nokia's faith in the growth of the at&t network.

Either way, now kenny and I know the facts, and we can stop arguing about this. I think texrat cleared it all up, and I'm moving on.

BTW, Kenny, you assumed I meant N900 in one post, but I'd editied it to 900 MHz. Sorry for the confusion.

texaslabrat provided a guess in the affirmative but I thought I'd just confirm it. The answer is yes, if there are different radio configurations (and to an extent, software configurations), Nokia designates them with different labels.

For instance:

Original Euro/Asia N95-1 = RM-159
NAM N95-3 = RM-160

Euro/Asia N86 = RM-484
NAM N86 = RM-485
China N86 = RM-486

Euro/Asia N97 = RM-505
China N97 = RM-506
NAM N97 = RM-507

This is why when some people were saying that ordering an N900 from an importer would result in an EDGE only experience with T-Mobile USA, it was just plain bogus. The N900 that passed through the FCC had a model designation of RX-51 and was tested on the AWS bands. The N900 previews that had been popping up on the web (mostly from European sites) all had the RX-51 labeled clearly on the device as well. So in other words, there was no "NAM" specific N900 that you had to buy to get 3G on T-Mobile USA. The only reason you'd want to buy a "NAM" version was because of return/exchange/warranty purposes. My experience with Nokia USA is that they won't honor the warranty on a device purchased abroad.

Chris, I assumed you meant N900 because you typed N900. And btw, it's texaslabrat on this thread, not texrat......he's another guy (who I think you know.)
Anyway, SLOW DOWN man, you're supposed to be a journalist. There's enough confusion surrounding the NITs already. As you know, this forum is and has been the place for answers......but it's often a long process. There tends to be a rather low signal-to-noise ratio here with a lot of newbies, a lot of guesses, a lot of un-knowledgeable "facts", etc.
.
Also, could you still please clear up my confusion concerning your posts in that other thread ? See post #98: http://talk.maemo.org/showthread.php...84&page=10that

Thank you very much.

I'm a simple guy, I'll just try to find obvious markings from the producers (like what you've confirmed above) to find out whether there may be difference in spec\capability between products. If the SKUs are the same, then they should have same product inside

Can't channels be different because they have a different centerpoints, bandwidth and guard bands, even though the same frequency falls within both channels?

Plus, you asserted that the software can tell the radio to tune to any arbitrary frequency "XYZ". It can select channels by tuning to the 2135.6 MHz centerpoint of the channel, but it can't tune the chip arbitrarily to 2139 MHZ. Again, this is AFAIK

Of course, I didn't say that. I'm talking about the chips (and you're talking about the radios). Let's keep the focus on the chips that the semico delivers and the extent to which Nokia can control them via software. I don' think there are any UMTS chips that Nokia, via software, can control to tune to some arbitrary frequency. Again, pls feel free to point out any such UMTS chip.

Ummm... These don't support your main assertion, only the sub-assertion (which I've never disputed) that the UMTS chips have firmware that can be reflashed. Here there is not even a claim to be able to control the tuning of the UMTS chip via software; only to unlock the bands which it seems that the chip provider disabled for some reason or another. Whether either of these attempted hacks even worked to that extent seems unclear.

And I don't see how they show how an OEM like Nokia can tune the UMTS chip to some arbitrary frequency from the software stack.

Yes, please feel free to cite some IEEE Communications Society material, or better yet, just link to any commercially available UMTS chip which can be tunable by the software stack.

Yes, I've read about stuff like that. It is always unfortunate when devices or components get crippled. Unfortunately, once the chips have been contracted, an OEM like Nokia has no ability to overcome such measures but I would hope that Nokia would not deal with vendors that offer crippled technology.

Yes they can. But the physical radio doesn't know about it...it only cares about the frequency with which to oscillate the antenna. The mapping of channel->frequency happens higher in the stack. And the choice of what channel to use happens even further up the stack than that.

The software *CAN* tell the radio to tune into any arbitrary frequency, though there are some caveats provided by the WCDMA specs that must be followed. The main one is that everything must be in multiples of 200khz. That, plus the absolute min/max frequencies supported by the hardware, are pretty much it as far as the RF chip is concerned. Again, the other constraints are imposed further up the stack (such as the begin/end frequecies of the band that the phone is trying to communicate in...which is controlled by a firmware-enabled SoC).

Um, all of them currently shipping today? That's the whole point I've been trying to demonstrate..the channel-frequency mappings are in firmware that can be modified. People have done it..this isn't theory. I have given you examples of where this very thing that you claim to be impossible has actually been accomplished. If the "allowed" frequencies were hard-coded into the hardware IC, there would be no reason for them to be enumerated in firmware which can be modified, now would there? If the "allowed' frequencies were hard-coded in hardware, the U.S. 5800 3G launch issue would have turned out very different.


Many of them have been successful as I (and others) have pointed out.

Because they can write the firmware which controls the tuning functions. These things are not "black boxes" in inmutable silicon...they are small computers on SoC packages. The SoC talks to the RF chip via a serial interface which sends commands and data to the portion which converts the signal to an analog wave (and tunes the radio) for transmission out the antenna (or conversely tunes the radio to await a signal from the tower). If you look at the SoC "computer" part, you'll notice that it is an ARM-based computer with it's own firmware stack separate from the ARM-based computer that runs the user-facing software. It is here that the tuning decisions are made and where the DSP lives that handles all the complicated decoding/encoding stuff. In my example that I linked in the prior post..this portion would be a Qualcomm PM7500 which is serially connected to the radio (RF chip) of varying types depending on the frequency ranges one wants to support (eg RTR6275/RFR6500..etc)

I did...I might have included it via edit after you started this response...check out my latest thread again.

I mean, if you are just trying to be pedantic about the whole thing..then yes you are right. The radios are restricted to tuning to certain frequencies because the raster spec of 200khz so tuning to 2110.111 Mhz would not be allowed because it is not a smooth multiple of 200khz. However, what you keep missing is the radio itself is COMPLETELY oblivious to the label that we have assigned a given frequency. This means that the radio is fully capable of tuning into "fractional" channels since such fractional channels are allowed by the WCDMA spec even if they might not exist in practice by being deployed by a service provider. So, if you are trying to say that, through following the WCDMA spec, that by default every possible channel allowed by the spec is "burned into the silicon" due to the restrictions imposed by the raster and the min/max frequencies of the various bands..then sure I guess you could claim that..though the safeguards in place from one controller to another might vary so you would have to looks at each individual design case-by-case to veryify it wasn't implemented in firmware instead.

But instead what you SEEM to be claiming is that there is some hardware-based database of channels (and I mean "channels", as in the derived labels, and not the physical frequencies) in use in each market that the phone is restricted to tuning to..and that's simply not true as has been demonstrated through decompilation of radio firmware. Further, the choice to utilize "BAND IV" channels is a firmware issue if the radio is already spec'd for "BAND I" as long as the corresponding 1700 radio is also present in the system. There isn't a special version of the 2100 radio that is "band iv capable" versus a 2100 radio that is "only band I capable". If *that* is your argument..you are dead wrong. Sorry...nobody is going to invest in creating millions of "BAND I ONLY 2100Mhz" radios, and in a separate bin have "BAND I + IV 2100 Mhz" radios when one part can do both jobs. That would just be bad business and, as I've pointed out, fabs make their money through volume and scale..having 2 sku's that are capable of the same job is a loser. They are one single part that gets used in various system configurations to support the bands a given phone is meant for. Now..are there hardware restrictions (ie burned into the IC design) in the controller chip that would prevent tuning to some undefined channel outside of known bands or that didn't follow the frequency raster constraint? Say a new band that has a frequency range outside of any existing UMTS band class. That I won't argue either way since I'd have to be on the design team to know for sure...and both "yes" and "no" are plausible since such restrictions could be implemented anywhere the designers saw fit (hardware or firmware). But that's not what we're dealing with here...we're dealing with chips that are aware of the full breadth of frequencies in use in terms of blocks of continuous spectrum and have connections to radios that have, in turn, the appropriate analog components to be able to utilize those frequencies.

So I guess all this really comes down to the fact that, although it could theoretically exist, you can't BUY a "BAND I Only 2100Mhz" UMTS radio for use in conjunction with a UMTS SoC. They don't exist through volume suppliers like Qualcomm. What you CAN buy is a "2100Mhz" UMTS radio. You then program the controller to use the appropriate channels for your market via firmware. If you have also selected a 1700Mhz radio for your system, then you can add the Band IV channels to the mix and tell your controller, via firmware, that they are fair game. Likewise, you can't BUY a "BAND I Only" UMTS controller versus a "BAND I + IV" controller as dictated by customized transistor circuits. They don't exist (these days, anyway). You can buy an integrated package of a generic UMTS controller included with frequency-appropriate RF chips (ie radios) and you program it with firmware to talk to the networks you are interested in (ie which bands you plan to connect to in the markets the product is intended for).

As has been demonstrated by our Brazilian friends..you can try and tell your radio, via firmware, to use frequencies that there is no analog hardware for. The controller will happily comply..but the analog portion isn't designed for it so in the end it fails to connect. If all of that isn't proof enough for you, then I suppose you just can't be convinced of the facts. Tilt on, Mr. Quixote...Tilt on!

I'm thinking the main confusion is that some people think that firmware != software... that is wrong...

Firmware is Software just runs on a different level.