Slashdot Mirror

http://www.analog.com/processors/processors/sharc/ index.html

There is a whole family of DSP chips from analog devices with that name(Super Harvard ARChitecture, more info about what that means . SHARC's get used in lots of things, particularly audio equipment like creamware's dsp cards, behringer's digital mixers, etc along with many other uses. For some additional info about DSP's you can read this intro

It's neat that Vanu is doing this on Linux, but it's not like it's a revolutionary technology breakthrough.

There's still an analog RF radio involved; all the digital processing is at the IF frequency. Digital signal processing of raw RF in the gigahertz range is still a bit out of reach. (And it will require an A/D with huge dynamic range.)

It's not clear that it's a win to do this using commodity PC hardware. Most of the crunching is in tight signal-processing loops that don't use much memory. With custom boards, you can have more CPUs on a board. Squeezing the physical size down matters in this application. If you can put the gear in a box on the pole, instead of needing a little shed, that's a big win. PCs also tend to use more power, and thus generate more heat, than DSPs per MIPS. Cooling all the gear is a constant headache in the cellular business. It typically doubles the power consumption, and the air conditioners themselves are maintenance headaches. What the industry wants is gear that doesn't require air conditioning, at least for smaller sites. Qualicomm has been shipping pole-mounted CDMA base stations since 1997.

It's also not clear that introducing a network between the radios and the processors helps reliability. If the radios are flexible enough that one can take over the job of another, it's easier to fail out a radio/processor pair and switch in another one.

None of this matters all that much because the cellular base station equipment industry is in the tank. The industry overexpanded based on forecasts of huge needs for 3G gear, and that didn't happen.

2) All the four digit model number BlackBerries (GSM/GPRS, CDMA/1xRTT and iDEN) use ARM processors.

It uses an ARM7TDMI as well as an ADI DSP in this chipset.

an Intel 386 32-bit chip

That's strange - I thought they use the ADI Softfone chipset. This would seem to imply that they do...

an Intel 386 32-bit chip

That's strange - I thought they use the ADI Softfone chipset. This would seem to imply that they do...

I work with ADCs and DACs all day. Your first pass answer of 96dB is correct for DC characteristics. However, sinewaves introduce some differences.This ADC is a darn good performer. You will notice that the SFDR (Spurious Free Dynamic Range) is -101dB, while the THD is -99dB. Also, its Signal to Noise is -92dB, while the theoretical best is -98.08dB.

In fact, a small amount of noise actual can improve the signal representation! But that is a rather long discussion.

FPGAs are really neat, but waaay to expensive for a consumer product. (High-end FPGAs run $1k+, the cheapest moderately compex ones* are still $50-150). Audio is still relatively simple in processing and data thoroughput requirements - an FPGA would be overkill. For expandability, it would be best to get a cheap massively-overpowerful DSP and do it all in custom software. For power, an ASIC would be most efficient (also cheapest, but the least flexible)

----
*Yes, with a lead-in like that, I must have pulled that number out of my butt!

To the original project. I have some samples of the latest model 2G and 10G accelerators from Analog Devices, the ADXL202JE and ADXL210JE respectively. They are in a smaller package now, which means they should fit in there even better, but I haven't yet got the surface mount caps that I need to implement the hack inside my Palm Pro with 2MB upgrade. Still, it's on my list.

I just don't understand how they're going to pull this off. If you've ever taken any kind of digital design class, you understand how ubiquitous and relatively inexpensive A/D converters are. I've had a whole lot of fun with the AD670
from Analog Devices and if you look at the pricing, I just don't know where they're going to fit a "cop chip" into this thing. At $10 a chip or thereabout, putting a big chunk of digital logic into it seems unmanageable.

Also, maybe I'm wrong about this, but for the "cop chip" to tell the ADC whether or not everything is good, won't it just have to make a pin high or low to signal it to shut down? I mean, there is a limited number of ways that one chip can efficiently signal another chip. How hard would it be to just tie the "cop chip" lead on an ADC to Vcc and not worry about it again?

This doesn't even get into the fact that there is no WAY, I mean NO WAY for any over-the-counter ADC to pick up a watermark. I mean, its not like in the Flintstones where there is a little bird pecking shit out on a stone tablet and the bird says "Hey, that's copyrighted... SQUAWK!" The ADC doesn't know how its being used. I could be throwing anything at it in any format. Hell, I could have run all sorts of analog signal processing on it beforehand...

Here's a good example: I was building a TTL-based robot for a class that needed to see using an NTSC camera. In the class, we weren't allowed to use any pre-fab processors - we were required to build our own from scratch. So I built a little special-purpose 16-bit RISC processor. So memory wasn't exactly in abundance, due to prototyping limitations, and clock speed wasn't exactly stellar, so the two big video-processing options were out: not fast enough to process NTSC in real-time and not enough memory to store the whole image and then process it. The way I got around this was by using the camera as a sort of memory device and detecting each horizontal sync off of the ADC, then just storing the individual line and processing that. It was an extremely good compromise. My point is this: how would the MPAA or whoever have any idea whether or not there was a watermark on the thing I was pointing the camera at? Particularly since I ran the whole thing through some signal processing before the ADC in order to widen the band of the most relevant image data.

At the end of the day, this is simply impossible. Maybe companies that have a vested interest in this, such as Sony, can integrate it into their higher-level hardware, but I can't imagine that the Analog Devices of the world are ever going to pull this off sanely.

Just my thoughts.

Check out Analog devices; their 2181 demo has echo cancellation as a part of the included software; source included.

The comments above re bandwidth etc are correct. Using the cable modem tuner our bandwidth is effectively limited by the SAW filter in the tuner module. It's sets an IF bandwidth of about 6.5 MHz.

The best SDR type A/D I've come across is the Analog Devices AD6645. It'll do 80 megasamples/sec at 14 bits. It's also capable of bandpass (aka direct IF) sampling.

...there has to be a translation from protected digital to unprotected digital before it is converted to analog. All I have to do to pirate is capture that stream before it goes to analog.

While technically true, all it would take would be a D/A converter that accepts encrypted input. An example of a chip that does this can be found here (although it's for voice only, right now). You'd need to probe up a dozen spots on the chip, and each spot could be protected by metalization above the trace you're trying to probe. This would raice the cost of digital ripping so high that it would be worthless to most people. DVD pirates would (and do) just make copies of the encrypted digital content.

The fact that Vorbis is still an 'evolving standard', meaning that a player made now could be obsoleted (or at least require the user to reflash something) when a newer version of the codec is released, is probably also a deciding factor. Developers would have to hinge their product on the hope that nobody adds a new CPU-intensive enhancement to the codec before it is finalized. (I don't suppose they would, but then again, engineers who suppose also tend to get sacked.)

You are letting your obvious bias get in the way of what is going on. AMD should have some sort of internal thermal protection system to stop this. The P3s do, the P4s have the best implementation. Price is not a factor here either. There are many, many very rudimentary amplifier chips and power regulators have thermal protection built in and they cost less than a dollar. I'm not expecting a throttling design like the P4, but at least a freeze like the P3 would be nice.

This isn't directed at you but I think its funny how a few months back everyone was bitching cuz the P4 would throttle back when hot.. hehe.. my how the times have changed. Guess, thats the difference between understanding and knee jerk reactions.

JOhn

The accelerometers were Analog Device's3-axis evaluation modules. The outputs were fed via a tether to a custom interface card which plugged directly into the host PC. A set of six 20Mhz high-speed Motorola video ADC's (sorry, can't find link) converted the acceleration data. The host PC was responsible for querying the card.

The software used freshman-level kinematic equations to integrate/interpolate the data to the the position. There was a fair amount of hysteresis in the software, but I attirbute that to only having functional hardware for a week before the project had to be presented...

From analog.com:

Some people have discussed 'reverse ADSL' - simply swapping two modems so that the high capacity direction is from the home to the CO. Unfortunately, in most cases this is not going to work. ADSL relies on all the 'loud' signals being located together (eg downstream sends are all at the CO), and all the weak received ones being located in a different frequency area, and physically separated. If you reverse this, then at the CO the loud 'send' of everyone else's downstream will be right where your reversed system is trying to listen to a very weak will be right where you are trying to receive the attenuated noisy weak high capacity 'upstream' - drowning it out. Conversely, your transmit signal will swamp everyone else. Given spectral compatibility constraints and 'good citizenship', this will limit reverse ADSL to perhaps 1000ft. Of course, up and down are arbitrary - what matters is everyone has to operate in the same direction. It is a little like driving; in the US people drive on the right; in the UK they drive on the left - either is fine, so long as you are consistent!

So it looks like you'd have to convince the telco to configure the entire DSLAM for reverse aDSL - for all the customers connected to it. Most likely not going to happen.

- Eric

Creative soundcards aren't good, they're good enough. Without competition there really isn't any way to judge 'better' or 'worse'.

There are plenty of other sound card manufacturers out there. Look at AOpen, Crystal and Analog Devices. They all make sound cards, but why are they not leading the market? Because Creative cards are just plain better.

I have never had problems installing a Creative card. Plus, they are still improving their cards (even when they don't have to). The new series of Live cards prove this. They are amazingly above their old Live series and the competition. I have tried the other cards, and my judgement is that Creative leads the market because their products are insanely superior to the competition. Besides...how much more can one add on to a sound card???

"Light and Matter Physics" High School/Community college level.

"Handbook Of Applied Cryptography"

"Numerical Recipes in {c, fortran}"

"The Scientist & Engineer's Guide to Digital Signal Processing"

"Using Z"

"The Red Book"

etc. I'm sure there are a ton of others.

"This seems like an expensive and inefficient solution for the problem you describe. PCs are actually pretty bad at doing signal processing; there's a whole class of chips (DSPs) that are optimized for it. There are almost certainly DSP-based hardware boards/widgets that will handle a lot of these effects for you, and it wouldn't surprise me to learn that there was a general-purpose programmable DSP card on the market too. Researching this before you spend money on a cluster would be a Good Idea."


Yup. Here's some links:
Symbolic Sounds Kyma system, a DSP-based programmable hardware solution for audio.
Details on the Kyma system (drool!):
http://namm.harmony-central.com/SNAMM00/Content/Sy mbolic_Sound/PR/Kyma5.html

Hardware DSP hacking tools:
Analog Devices EZ-SHARC/EZ-KIT DSP experimentation board (which screams for Linux support, incidentally)

The author of the review is wrong when he talks about the freestyle pro using gyroscopes! It uses a nifty little chip called an accelerometer made by analog.com that is found in many products like car alarms and the like. It is a very cool chip (I have some myself) and they will send you some samples if you want. The chip is called the ADXL202

SuperID

If you need really a fast floating point real time, why not use a DSP? Although DSPs are primarily built for signal processing, the new processors handle other operations like interrupts pretty well. Many of these DSPs can perform floating point inversions in one cycle. A lot of them are Harvard Architecture machines and are primarily built for fast Multiply and Accumulates. But, these processors run real time OSes and are capable of doing multithreaded operations. Some newer processors have really fast ports that can talk to other processors and can be easily scaled to a multiprocessor system. And - manufacturers like TI and Analog Devices have been making really fast DSPs lately (TI just announced its 1GHz DSP). They are not that expensive either (They are comparable to the embedded processors made by Motorola). And, the best part is that the new processors really little power. So, why use Intel or AMD for real time processing when you can use a really fast DSP which costs less.

• System Clock: 300 MHz
• Co-Processor: FPU (Floating Point Multiply Accumulator x 1, Floating Point Divider x 1)
• Vector Units: VU0 and VU1 (Floating Point Multiply Accumulator x 9, Floating Point Divider x 1)
• Floating Point Performance: 6.2 GFLOPS

The PS2's vector units are running at three times that speed and are almost two times longer. Plus, there are two of them in there.

Two words: Holy shit.

Another player in the field (Intel) only pushes the others toward excellence and is a Good Thing (tm). IMHO.

BTW, Intel doesn't make any analog-to-digital (A/D) or digital-to-analog (D/A) hardware, so systems OEMs integrating the Intel DSPs would have to go elsewhere for those. A/D's & D/A's are required for most DSP apps. So: monopoly? bah!

Hardware Guy by day, Linux slacker by nite,

QuantumHack

The mouse would need a capacitance sensor, though, so that it wouldn't drift if your table is at a slight angle. It should probably sense change-in-acceleration, rather than absolute acceleration, for the same reason.

This would be an amazingly precise mouse, that would never jam or get dirty, would respond to your slightest touch, be reasonably cheap to produce, and would just be cool as hell.

I know, though, that Microsoft will patent touch mice, now. I suppose I could still build one for my own use -- which is all I really wanted to do, anyway.

thad

• Phillips 87c055 - 8051 microcontroller for television and video, you can control it through a serial or parallel port.
  
• Analog Devices AD722 - RGB-to-NTSC converter
  
• Phillips SAA5252 TV Data Display (this is what I recommend for titling, it costs ~US$8, is controlled serially, and takes composite or RGB input and outputs RGB (which can be converted using the US$8 AD722 above). You can use a parallel or serial port to tell it what to display.)
  
• There's more out there, try Phillips first and try also Analog Devices who gives out free samples (I got a ~$100 ADC from them a while back).
  

• Build This Video Titler, Dan Michelson, Electronics Now, May 1995, p.49
  
• Hardware Hacker: Low-cost TV data displays, thoughts on brain implants, ..., Don Lancaster
  
• Hardware Hacker: Lamps and lighting efficiency, Don Lancaster
  
• Tech Musings: Video, Video Editing, Character Generators, and more, Don Lancaster
  
• Hardware Hacker: Laser printer repairs, sync-separator circuit, GPS navigation update, video interface module, and hacking Super Nintendo!, Don Lancaster
  
• Hardware Hacker: Dye-based solar energy, Neo Geo interface cables, another sad patent story, and RGB monitor fundamentals, Don Lancaster
  
• Tech Musings #110: NTSC TV fundamentals, Don Lancaster
  
• Tech Musings #134: PIC video overlay displays, Don Lancaster
  
• Captain Video's Secret Mountain Laboratory, Don Lancaster
  

Funny, but I was under the impression that was exactly how my Logitech TrackMan Marble worked. Same principal at least.

The ball is patterned (default is a spotted ball), and the sensor tracks the shifting pattern as it moves. Best trackball I've ever used; can't stand mice for long-term activity.

They both still need a surface though; should have used accelerometers instead.