Slashdot Mirror

This isn't true. Check out Texas Instruments. For example, all their Sitara processors are available up to 105C.

http://www.ti.com/processors/s...

I didn't even look at other lines, I just clicked the first product lines that would do the stuff in the question, eg push 32 bits around moderately fast.

Electrical Engineering is not my speciality, but would it take much to wire up 2 microphones, facing opposite directions, into an operational amplifier like the 741, or perhaps better yet the LM386? The difference between the two signals would be his speech with background noise subtracted out & such a circuit could amplify just his speech.

As for #2, feed it into a computer, run a Fast Fourier Transform, tweak the signals, and convert back.

He might have to use speakerphone with such a getup, but...

Congratulations for proving you have poor reading comprehension.

He said not allowing the owner of the device to upload code was wrong, NOT that allowing anyone to upload code was right.

He also gave an example for an authentication mechanism to determine who can upload new code. Which is absent in the vulnerability you're referring to.

Further the vulnerability you're referring to would have still been present if they had required manufacturer signed firmware only. Which guess what? They do if it's enabled by the manufacturer of the device. Which means any vulnerability would be protected by the manufacturer and could only be fixed by the manufacturer, even if you knew exactly what bits were bad and how to fix them. The only difference in this case is that this particular vulnerability bypasses this restriction when enabled. Rendering the protection useless.

So no, his "complaint" is different.

Many Android phones had three little metal dots on the side. Those output HDMI. Likewise, many phones now have USB Type-C, which can carry audio, HDMI, and wired Ethernet (the specification allows particular pins to carry things not encapsulated by USB protocol--it even provides straight four-pin analogue stereo audio with microphone in both Nokia and Standard pin-outs).

You probably would need the extra battery, although I can't imagine this adding much weight: the Asus R11 Chromebook uses a 3490mAh lithium ion battery, while the OnePlus Six uses a 3,300mAh lithium ion battery. Seriously. The same battery in your phone runs the Chromebook for 10 hours. That's an 11 inch screen, not a 13 inch screen.

We live in a strange world. If you actually understood how computers work, you wouldn't believe it. It's like someone trying to convince people Harry Potter is real.

A codec is a device or computer program for encoding or decoding a digital data stream or signal.

https://en.wikipedia.org/wiki/...

for example:

http://www.ti.com/product/PCM3...

But you'll never hear a TV, DVD player, watch,

I have a watch that is a computer for real. It is the EZ430 which has an MSP430 microprocessor in a watch form factor, and comes with software to program it. Not just tell it what functions to do, but actual programs. If you buy the full IDE you can modify the entire watch program. The free version is limited and has software for "watch", but not the full functionality of the installed demo program.

The difference between TV, DVD, walkie-talkie, and the EZ430 is that you do not program the computer in the former, but you can in the latter. You tell the already programmed computer what of the preprogrammed functions it should do for consumer devices that contain computers; you program the computer in a computer. And yes, the iPad has a processor that you can program, so it is, indeed, a computer.

Demeaning your opponents. So when you can't win with facts, you try to win with insults. It does shows me what kind of person you are however.

While 32-bit bytes might be non-existent, 16-bit bytes are not: http://www.ti.com/lit/ug/spru2... . Check out section 5.3. Here's the note from the bottom of it:

Note: C55x Byte is 16 Bits By ISO C definition, the size of operator yields the number of bytes required to store an object. ISO further stipulates that when sizeof is applied to char, the result is 1. Since the C55x char is 16 bits (to make it separately address- able), a byte is also 16 bits. This yields results you may not expect; for exam- ple, sizeof (int) == 1 (not 2). C55x bytes and words are equivalent (16 bits).

Guess the engineers at ISO and Texas Instruments are full of crap as well.

You are not an opponent, you are nobody. The Texas Instruments engineers are bright people, who are not part of this conversation. They did not utilize 9 bit bytes, nor did anyone else for that matter. You fucked up by spouting off about 9 bits and you haven't been able to recover. You've desperately tried too, but you cant get around the fact that a 9 bit byte doesn't exist. In the future, feel free to sling your garbage as it pertains to any real byte implementation.

Demeaning your opponents. So when you can't win with facts, you try to win with insults. It does shows me what kind of person you are however.

While 32-bit bytes might be non-existent, 16-bit bytes are not: http://www.ti.com/lit/ug/spru2... . Check out section 5.3. Here's the note from the bottom of it:

    Note: C55x Byte is 16 Bits
    By ISO C definition, the size of operator yields the number of bytes required
    to store an object. ISO further stipulates that when sizeof is applied to char,
    the result is 1. Since the C55x char is 16 bits (to make it separately address-
    able), a byte is also 16 bits. This yields results you may not expect; for exam-
    ple, sizeof (int) == 1 (not 2). C55x bytes and words are equivalent (16 bits).

Guess the engineers at ISO and Texas Instruments are full of crap as well.

HDMI and DisplayPort were USB-C alternate modes before Thunderbolt 3 came along. That's why the 2015 MacBook had only USB and DisplayPort supported on it's single USB-C port

This prompted me to do a bit more research and you're mostly right. According to the alternate mode specification, the 3 modes for USB-C include USB, DP, and power delivery; there is no HDMI support in the USB-C spec. Non-Thunderbolt Macs with USB-C (e.g. the 2015 and later non-Pro MacBook models) require an adapter for HDMI and Apple's adapter contains active hardware (similar to DisplayLink) to provide simultaneous HDMI (or VGA depending on which adapter you get) and USB, which is not possible when using an alternate mode. Alternately, it seems you could use a DP to HDMI adapter to get HDMI (and better display performance than Apple's HDMI adapter), but at the cost of any other use of the port (e.g. USB or charging).

Interesting, none the less, and probably why Intel chose the USB-C port for TB3. At any rate, a USB-C to DP cable isn't likely to confuse a lay consumer. USB-C to HDMI, though... since the HDMI mode is part of the TB3 spec, someone with, say, a 2015 MacBook might wonder why a USB-C to HDMI cable doesn't work.

I was with you at the beginning but this discussion and the research I've done has shown the logic behind why USB-C is what it is. [...] Thunderbolt does away with this USB baggage by leaving it behind as much as it could. The need for USB 3 fallback is pretty limited and is there so long as one chooses their devices and cables carefully.

Unless you have a USB-C host without Thunderbolt, which includes the non-Pro MacBook line (including current models) and every AMD system with USB-C. Then, well, you might like to have devices that can utilize the higher throughput and lower latency of Thunderbolt 3 where it's available, but still work where it's not. Portable drives with both Thunderbolt 2 and USB 3 are popular as hell, Apple sells the living shit out of them, in case you're wondering if that's a use case anyone actually cares about. My wife, a graphic designer, has a couple of them and uses TB2 with her iMac and MacBook Pro and USB 3 with her PC laptop that doesn't have Thunderbolt. Another advantage with Thunderbolt (all generations, mind you) is that it's not CPU-driven like USB, which is why it might be preferred where it's available, even if the additional throughput would be superfluous.

I can look forward to a thinner and lighter laptop in my future and not have to be concerned if I have the right ports on it.

As long as you choose Intel (as they won't license Thunderbolt to other x86 chipmakers), that is. With AMD being ahead in performance per dollar, performance per watt, and overall performance at the moment, that just seems... limiting. Also keep in mind that the lay consumer often buys on price which, combined with the recent release of Ryzen and the just-release Ryzen APUs and upcoming Ryzen mobile processors, means the lay consumer is likely to not have Thunderbolt 3 on their laptop in the very near future. Especially if they're shopping for better performance, longer battery life, and a cooler-running system, because AMD has all of those boxes ticked with Ryzen, along with the price box.

In short, this will be an issue for the lay consumer because the systems they're likely to be buying are not going to support Thunderbolt 3.

Which leads me to believe the problem will solve itself as those same consumers learn to avoid the Thunderbolt logo, then advise their friends and family to do the same. As I said before, Intel killed TB3 with they chose to use the USB-C port for it and not license it to AMD.

These annoying little adapters aren't so annoying since I know I can leave them behind and have a tiny little laptop when I'm done working, rather

At one extreme, TiESR https://gforge.ti.com/gf/proje... is a fairly simple to use. Not state of the art, but it does use Hidden Markov Models (HMM's) and has some noise compensation built in. It comes with word and language models, so it's fairly easy to use - for US English at least. I haven't been ambitious enough to figure out how to build new models.

At the other extreme, Kaldi http://kaldi-asr.org/ is the most advanced open source recognizer that I'm aware of. Neural Nets and all the other goodies researchers have been working on the last few years. Definitely not easy to compile or use, though. And don't even think about trying to design a neural net without a graphics card to use as a math accelerator: one of the examples ran for days and wasn't even close to finishing when I gave up.

Anybody else have suggestions for another toolkit?

You mean the Amazon Echo:

https://en.wikipedia.org/wiki/...

https://www.ifixit.com/Teardow...

With 256Mb RAM, 4Gb NAND, and a 32-bit chip:

http://www.ti.com/product/DM37...

that's based on an ARM Cortex-A8:

https://en.wikipedia.org/wiki/...

Which is a 32-bit CPU?

Yeah. You're a twat, who thinks that the bigger the number the better and you "can't possibly" do stuff with anything else.

I mean, seriously, did you even spend two fucking seconds thinking about it, given that one Google search and the first couple of Wikipedia articles show you how much rubbish you're talking?

That's a good point, but the thermite need not apply extreme heat to the entire device -- just the flash memory chip. A strip of magnesium with oxidizer might also do the job -- both could be adjusted so that the user sees just a bit of smoke coming out of the phone, without personal injury. See section 3.1.2 in this document about the relation of temperature to data retention.

Shorting out the battery through a coil around the memory chip is likely to make it hot but not necessarily hot enough to truly erase the information (I may be mistaken -- I haven't done the calculations), You might even explode the battery and still keep the information. This also would not necessarily produce the electromagnetic pulse you're thinking of. Even if it did, EMP is unlikely to destroy the contents of flash memory. (Note that I am talking about actually destroying the information stored on the chip. Destroying the small gold wires on the package which connect the chip to the rest of the circuit is easier -- but an entity with enough resources can still recover the information).

Here is another writeup on the effects of temperature and storage time on flash memory (it seems baking the chip at 125 C for 10 hours will still not necessarily erase everything on it). You will need much higher temperatures than you would get with a polymerization reaction to destroy memory in span of several seconds. Both thermite and magnesium can produce thousands of degrees Celsius, which any data stored in flash memory is unlikely to survive.

Replying to myself, I found this that says NOR erases are mega slow. So perhaps NAND on SIM cards.
http://focus.ti.com/pdfs/omap/...

I recently wrote an App for a company that's building AI models of organic and non-organic substances.
The App works over bluetooth with a pocket sized, battery powered, NIR spectrum analyser made by Texas Instruments.
TI DLP® NIRscan Nano

This technology will be the next big thing on smart phones which are crying out for something new to revitalise the market.
Pick a time frame. I'd go for 5->7 years from now that right next to the camera on all smart phones we'll see an NIR scanner.

Imagine being able to scan anything and get a report of it's contents and make up?
Fake, or out of date pharmaceuticals, will be a thing of the past.
Nut allergy? Scan the meal you've just be served before eating it.
Consumer products like Aloe Vera Gel not containing Aloe Vera? Your phone can tell you that it's a fraud.
The list of applications is endless.

Here's a demo video of the App identifying real and fake viagra pills.

Tricorder Trekkies? :)

I agree that 'smart glasses' fall into the "not at all clear we can actually find a market for these" category; but 'risky' seems like an odd term.

TI would be more than happy to refer you to somebody who sells suitable displays in small quantities; and Apple engineers presumably have access to quantities of iphone/Apple watch dev boards, so getting a prototype up and running would probably be cheaper than some excessively long meetings.

Doesn't mean they'll necessarily have any better luck getting a useful product out of it; but it is a trifle difficult to call it 'risky' when potential losses are so small compared to the company.

Type C can be used to pass an analog signal by way of two configurable sideband channels (SBUs) in Audio Adapter Accessory Mode. Whether anyone is doing so yet is another question.

http://www.eevblog.com/forum/r... - I am unsure if this link will work. This shows a teardown of a tag.
On eevblog, under the title "weird "energy-harvesting" broadband (?) antenna ".

This shows a 450mAh rechargable lithium-ion battery.
http://www.ti.com/ww/en/wirele... - this is a TI device which is designed for sensor tags, and sports a 1 year battery life reporting once a second over BLE with a 2032 battey. This is one half the capacity of the lithium battery used.
It also won't be sending data at one second intervals.
It seems entirely reasonable that the comparatively large battery will last five years without any wifi charging at all.

Here's a electric toothbrush reference design from Texas Instruments.

Here's the MSP430G22x0 microcontroller used in the design.

Here's a list of software tools for that microcontroller. The list includes something called "GCC", which they say is an "Open Source Compiler for MSP Microcontrollers".

Here's a page from Renesas about electric toothbrush designs.

Here's a list of software tools for Renesas processors; they list C compilers for the R8C and RL78 microcontrollers, as mentioned in the previous page.

So don't assume all the code in your toothbrush was written in assembler language; some of it may have been written in C, although some of the low-level library routines might be written in assembler (or an asm in the C code).

Here's a electric toothbrush reference design from Texas Instruments.

Here's the MSP430G22x0 microcontroller used in the design.

Here's a list of software tools for that microcontroller. The list includes something called "GCC", which they say is an "Open Source Compiler for MSP Microcontrollers".

Here's a page from Renesas about electric toothbrush designs.

Here's a list of software tools for Renesas processors; they list C compilers for the R8C and RL78 microcontrollers, as mentioned in the previous page.

So don't assume all the code in your toothbrush was written in assembler language; some of it may have been written in C, although some of the low-level library routines might be written in assembler (or an asm in the C code).

Elon may also be betting on technology getting good enough to no longer need LIDAR. Such as this: http://www.ti.com/lsds/ti/sens....

My favourite programming language is solder.

That was Bob Pease.

I know this is a sarcasm thread, but 24 bits is actually a lot for an ADC. You're talking 0.2uV/LSB with a 3.3V reference. Even getting close to that requires careful attention to noise sources and PCB layout. 16 bits is pretty hardcore in its own right. 8-, 10-, and 12-bit ADCs are far more common.

But it's about headphone jacks, isn't it? No ADCs involved.

DACs for audio are cheap as chips, and tiny: something like TI's PCM1795 outputs two channels of 32-bit samples at 192kHz with a 120db SNR, costs less than a buck in bulk, and comes in a minuscule package (something like 1 x 0.5mm). Add another buck for a clean signal path plus a solid Neutrik-style connector and it's well underway. But the question is then, where could they put in all the DRM?

I know this is a sarcasm thread, but 24 bits is actually a lot for an ADC. You're talking 0.2uV/LSB with a 3.3V reference. Even getting close to that requires careful attention to noise sources and PCB layout. 16 bits is pretty hardcore in its own right. 8-, 10-, and 12-bit ADCs are far more common.

But it's about headphone jacks, isn't it? No ADCs involved.

DACs for audio are cheap as chips and tiny: something like TI's PCM1795 outputs two channels of 32-bit samples at 192kHz with a 120db SNR, costs less than a buck in bulk, and comes in a minuscule package (something like 1x0.5mm).

I know this is a sarcasm thread, but 24 bits is actually a lot for an ADC. You're talking 0.2uV/LSB with a 3.3V reference. Even getting close to that requires careful attention to noise sources and PCB layout. 16 bits is pretty hardcore in its own right. 8-, 10-, and 12-bit ADCs are far more common.

But it's about headphone jacks, isn't it? No ADCs involved.

DACs for audio are cheap as chips and tiny: something like TI's PCM1795 outputs two channels of 32-bit samples at 192kHz with a 120db SNR, costs less than a buck in bulk, and comes in a minuscule package (something like 1x0.5mm).

I know this is a sarcasm thread, but 24 bits is actually a lot for an ADC. You're talking 0.2uV/LSB with a 3.3V reference. Even getting close to that requires careful attention to noise sources and PCB layout. 16 bits is pretty hardcore in its own right. 8-, 10-, and 12-bit ADCs are far more common.

But it's about headphone jacks, isn't it? No ADCs involved.

DACs for audio are cheap as chips and tiny: something like TI's PCM1795 outputs two channels of 32-bit samples at 192kHz with a 120db SNR, costs less than a buck in bulk, and comes in a minuscule package (something like 1x0.5mm).

If you look at the TI chip for USB C power delivery, here, overcurrent and overvoltage protection is specified as a feature. This is not to say that I have first-hand data on how he blew his Pixel C or what chip Google used. I do notice, though, that Google is not offering a protection policy for Pixel C as they do for the phones sold in the Play store. I hope the design isn't problematical.

Here's a 71 page reference called "Caller ID (CID) Algorithm User’s Guide". The service providing names with numbers is technically separate, called CNAME. Wikipedia has some basic info. But seeing as Caller ID was developed in 1984, far before most tech we're currently using and made to work with basic switching systems (in the teleco sense, not modern networking sense), it's quite primitive and easily spoofed. Anyone with an old modem can make scripts to spoof it.

Just like the 2N3055.. Yes, old. but in some cases, absolutely perfect for the job. Usually as linear pass transistors for power supplies :D

There is plenty of reason to avoid the 2N3055, well actually to avoid the TO-3 packaging. The package flexes when mounted, which can cause poor thermal transfer to its heat sink. TO-3 are manufactured to be curved so as to help ensure good thermal contact, but the curvature isn't preserved if the TO-3 is removed.

References: OnSemi's Application Note 1040, and Burr-Brown (TI) Mounting Considerations for TO-3 packages.

Is TX Instruments somehow related to this story?

Yep.

Who do you think builds radiation resistant electronics? Radio Shack?

And TI's LaunchPad boards: http://www.ti.com/ww/en/launch...

The primary advantages of this thing seem to be a somewhat more compact and breadboard-friendly form factor and less work to set up a development environment and get started with the board (the STM32F4 Discovery Board sample code is neither comprehensive nor particularly suited to being incorporated in other code).

I think you're going to wake up one morning and realize the internet-of-things revolution happened quietly around you. Either that or you're going to get dragged kicking and screaming into an internet-of-things world much like the textile workers of the early 1800s who opposed industrialization.

For the most part, the necessary tech artifacts you're talking about already exist. You can already order a mesh-routed, IPv6 aware radio IC for pretty cheap (6LoWPAN, example part by TI). It's been 4 years since NXP Semi demo'd occupancy-aware lighting modules. For me at least, intelligent lighting is a big deal because lighting costs are the third highest contributor to my electric bill.

The hardest parts, in my opinion, are pushing for standardization of interfaces to keep complexity and cost down, and ever-important though higher-visibility now, security and access control. There are already significant working groups dedicated to these tasks, for example, the goog/nest, ARM, samsung, et.al. in the Thread group. But there are a ton of different and incompatible ways to do the same thing; ANT+, bluetooth LE, zigbee, and 6lowpan are just the low power ones I can think of off the top of my head. And that's just the physical through network OSI layers, it doesn't begin to address announcement of features (zeroconf, etc.) to each other or standardized interface presentation to the user (????).

So where are the products? Well, Nest gen2 thermostat is IoT-enabled. Fitbit monitors all wirelessly update your stats and profile. Apple's [i]watch and the moto360 smart watch are both network-aware. Even companies outside of the consumer electronics sphere are getting invested, like Chevorlet's automotive lte/wifi.

Granted, these aren't the groundbreaking, for-every-person products you're talking about, but the tech infrastructure is coming into its own. Product development takes time and age is only going to make the baseline models cheaper, more capable, more standard, and more prevalent. There's a lot of work to be done yet, but given the number of people and companies invested in IoT consumer electronics industry-wide, it's hard to imagine a world where everyone simply gave up on the tech instead of working out the problems.

I did a bit of reading on the subject from TI, which has FRAM integrated into some of its MSP430 microcontrollers. If anything, the technology seems to be well-suited to the space environment, because bit storage is accomplished via a crystal structure change (polarization), rather than through charge storage.

>Power? Volts? Boost? Huh???

These must be the same people that wonder whether a product runs on batteries or electricity.

datasheet for up-converter
http://www.ti.com/lit/ug/sluua...

distributor announcement video
https://www.youtube.com/watch?...

I wonder what sort of human to d.c. transducers are planned. Flexing watchbands is one thing, but the /. crowd might be more excited over wind, combustion, and fuel cell power from flatulence.

Then there's the magnet implant paired with coils around the fingers or?
"Could you give me a hand charging my phone?"

Say it with me now... Voltage is not a measure of power!

Looking into the data sheet a little shows that this DC-DC converter maintains decent efficiency from a few microWatts to several hundred milliWatts;

http://www.ti.com/product/bq25...

Now, to bring this home with a car analogy (and a moderately controversial one, although it shouldn't be), quoting the voltage of an electrical power source is not unlike quoting the torque of a car engine. Neither is a measure of the power available. Torque can be changed by going through a gearbox without changing the power delivered apart from the efficiency of the transmission. Similarly, voltage can be changed without changing the available power apart from the efficiency of the DC-DC converter. In both cases, load conditions matter. It would be like looking at the specs of an air wrench that has up to 450 lb-ft of torque and saying "Wow, that's as powerful as a Corvette engine!"

AC is correct. Most forms of subharmonic oscillation are caused by slope compensation issues, but pole filtering can also be an issue . For those who want the nitty gritty details, see this. http://www.ti.com/lit/ml/slup2...

I really appreciated the Cell BE too. I do hope that that architecture, with cache coherence (local stores are a pain to manage), becomes more common. Have you taken a look at Texas Instrument's KeyStone II? It's ARM + crazy DSPs. It doesn't seem that anyone has really noticed it though. http://www.ti.com/dsp/docs/dsp...

Ten years later, the base model still has 480 kilobytes of ROM and 24 kilobytes of RAM, its black-and-white screen remains 96x64 pixels, and the MSRP is still $150

I really hate it when people pass off misinformation.

As tempting as it is to call the black and white version the base model, it doesn't appear to be manufactured any longer.

Which means that the current base model is the version that has with 3.5 megabytes ROM and 21 kilobytes RAM, with a color screen that is 320x240 screen. The calculator also has a rechargeable battery (type unknown) and an MSRP of $140.

You can find this information (except the MSRP) on this chart.

Incidentally, Amazon US currently sells the color LCD model (black) for $104. Other colors seems to cost more.

BeagleBone Black is that board. Better GPIO, no broken USB stack and other Broadcom SOC BS.

Without Ethernet, I don't see how this is supposed to be competitive against RPi considering that TI has the nice EK-TM4C1294XL Tiva C Series Connected Launchpad for $20.

http://www.ti.com/corp/docs/ki...

TI is notoriously bad at documenting their non-CPU cores.

Eh? See below, you may be complaining about the wrong company.

... Is there some commitment from them to open specs for the OMAP5 SoC?

The OMAP543x Technical Reference Manual has been available from TI's usual documentation pages for over half a year now (look for document SWPU249Y or newer). In typical TI fashion, that's 6105 pages of extreme detail, isn't it enough for you?

Here's a direct link to their highly informative block diagram for the SoC -- OMAP5432 Block Diagram.

Documentation is one thing that's never been a problem with TI, as long as the information is theirs to give. If a GPU core is licensed from a 3rd party though, for example the PowerVR SGX530 licensed for use in AM335x, then TI isn't legally allowed to document it openly --- your complaints are best directed to PowerVR.

TI has a well earned reputation for providing the most detailed reference manuals in the business for any parts which they actually own.

LDOs aren't that expensive, and certainly wouldn't dissipate that much power.

I mean, this circuit topology is very recent:

http://newscenter.ti.com/2013-...

Are you claiming that now every single LED lighbulb out there uses this now? Every single one?

The classic 74181 4 bit ALU shows how it can be done (page 5). It shows the schematic of the chip in gate form. Page 4 shows the 'opcodes' (really operation selection line combinations) that this simple chip can perform.

The mighty (!) 32xx series of minicomputers from Concurrent Computer in the 1980s/90s used a bunch of these chained together to form a 32 bit ALU.

You do not have to design your own...
Your algorithm probably needs vector units, fast multiplication, wide memory bus, largeish...

I will take you and that company many years to make a good dsp, and it will not be better the the Texas Instruments ones you can buy...

Or, you can implement it in an FPGA, but a DSP with handtuned code is more energy efficient...

Here is one for vision:
http://www.ti.com/lit/wp/spry251/spry251.pdf

This is not true. The MSP430, for instance, does not have its own PID, though TI offer to sublet their VID and assign you a PID. See this which says:

The use of TI's Vendor ID and an assigned Product ID can be shared with those who prefer not to obtain their own via the USB Implementer's Forum. This VID- sharing program is here to help your project get up and running as quickly as possible.

The Schematic for this board has 27 pages ...

An the reference manual for the TRE's CPU (AM335x) runs to 4,500 pages.

To paraphrase Nancy Pelosi, they had to build the Beaglebone to find out what was in it.

The AM3359 is a far more flexible chip than the BCM2835 of the Raspberry Pi, just look at his datasheet: http://www.ti.com/lit/ds/symlink/am3359.pdf. And at USD 23, the AM3359 is a very good deal.

Exactly, I believe the TI DLP system has been in use for this for quite a while. Last time I went looking they had DLP chip in the lab that was able to handle almost 15 watts of laser power without a problem. http://focus.ti.com/pdfs/dlpdmd/Using_Lasers_with_DLP(r)_Technology.pdf for one of their earlier papers on this.

You use a VFC, obviously. For example, you could use an LM331 to be moderately old-school about it.

Lets do the math:
FTTN will need power and optical rolled out into suburbia.
Each node will be ejecting fancy new vectoring or better tech into existing final very old copper runs of 200~2000m.
Australia will have have to look hard at each adsl user. That long run of existing adsl copper from the 500m-4 km exchange/rim (~digital loop carrier) will have fight with the new nodes.
What will a new 300m-2km run of vectoring copper do to existing adel 1/2 users?
Hint - every user will have to get a node connection for internet if vectoring is used near existing adsl connections:)
Thats a lot of nodes to build out in suburbia per 500-2000m suburban copper loops.
The short runs of copper are corroded, crushed or have a few too many joins and will need ongoing care.
Australia did over provision copper, but that was a long time ago ~100% redundancy (~2.5 pairs per home) is now very low.
What is left is over used or of unknown quality re low number of working pairs.
http://www.abc.net.au/technology/articles/2012/11/27/3642266.htm
Our solder joints are old, oxidisation is ongoing, alien crosstalk (ATX)....copper diameter (in Australia 0.4mm is common)... the list of copper issues in the ground is Australia is not like some 'new' lab network.
http://www.ti.com/sc/docs/products/network/vdslwp.pdf has some numbers over longer runs on page 32.
http://delimiter.com.au/2012/04/30/fttn-a-huge-mistake-says-ex-bt-cto/