Dangerous Prototypes: Open Source Hardware Seeding 48
MojoKid writes "Dangerous Prototypes is a two-year old organization with the stated mission of producing 'one new open source project every month.' In its nearly two years of existence, DP has created about 30 projects, such as the Flash Destroyer, which tests the limits of solid state storage by writing and verifying a common EEPROM chip, rated for 1 million writes, until it burns out. The projects themselves are being sold by another interesting company, Seeed Studio. Seeed is a contract manufacturing/sales channel for hire. It helps hardware designers get their ideas manufactured in China and sold worldwide with a service called Propagate for manufacturing small quantities (100+) of open source hardware."
Really, Flash Destroyer the best example? (Score:5, Informative)
Dangerous Prototypes doesn't just produce shitty projects for the purpose of proving something that can be found on any flash manufacturer's datasheet. They make plenty of very useful projects too. Not just useful for an end user, but useful for the hardware tinkerer. There two most famous projects by a long shot are the Bus Pirate [dangerousprototypes.com] and the Logic Sniffer [dangerousprototypes.com]
The bus pirate is a small device that connects to the computer via USB and allows you to use a terminal to talk all sorts of weird and wonderful protocols like SPI, 1-wire, I2C, or UART. Great for debugging a design, or reverse engineering. It is also capable of sniffing out the commands on a bus.
The Logic Sniffer is a cheap 16 channel Logic Analyser, which while no where near as good as a commercial unit comes in at 1/100th of the cost as well.
Both are fantastic tools for anyone hacking away at microcontrollers and both have saved me lots of headaches at some point. The best example was when the Logic Sniffer was released with firmware that wasn't very upgrade friendly to say the least, I used the Bus Pirate to flash new firmware to the Logic Sniffer.
They also make a JTAG programmer / debugger [dangerousprototypes.com], a Infrared I/O board [dangerousprototypes.com] for a computer, and a fully functional tiny Web Server [dangerousprototypes.com]. They are much more useful than the summary makes them out to be.
Re:Can you say... "I'm a douche" any more clearly? (Score:5, Informative)
If you knew anything, you'd know that none of these projects have been abandoned. Some are so useful and popular (like the Bus Pirate) so as to be in their 5th update. More than a few more-traditional electrhobbiest retailers carry one or more DP "products".
While the OP did a miserable job conveying what DP does (and picking what I'd consider their least useful/interesting product to highlight), your proud and tenacious grasp of ignorance dwarfs the lameness of the OP. Had I not been so familiar with DP and SeeedStudio beforehand, I probably would have emitted a "WTF?!?" when I first saw this thread, as some of the other less-smegtastic responders have
DP is probably the best example I know of small-scale cooperative distributed engineering in the hobbyist space. While Adafruit and SparkFun certainly dwarf DP in terms of sales and recognition, DP is an order of magnitude more tightly involved with their target audience in the definition and execution of their projects, and have a much higher ratio of work done by their customers than they do themselves. To date their "customers" are just as involved as the greater Arduino, and now Kinect, communities.
Re:Really, Flash Destroyer the best example? (Score:5, Informative)
Logic analyzers are collecting dust now, for many reasons. Originally they were envisioned to capture the complete state of a complex logical circuit, and perhaps they were useful at that time. But that time was about 20 years ago.
Today all the hardware that logic analyzers attempted to debug is located on the MCU die. You don't have access to it, and generally you don't have to - the manufacturer of the MCU did that already, and the peripheral is likely to work (modulo the errata.)
If you are building a full size clone of IBM/370 then you can benefit from the device. But if you are doing it in an FPGA then Xilinx already has a logic analyzer there, called ChipScope. It's an IP core, so you can remove it when you are done debugging.
What you really need today is a half-decent digital scope, with at least 1 GS/s (better 4.) You don't need lots of channels - one is enough, two is plenty. If you need four you probably don't know what you are doing. Note: the trigger channel is also a channel, it's there for a purpose.
You can debug nearly all the amateur level hardware with just a scope. More test equipment is needed if you are into RF. The problem, as you point out, is that cheap equipment is often a waste of time and money. If you have to, go out and rent a decent instrument, it's not that expensive on a daily basis. You don't want to fight the test equipment that is too slow for your signals - you won't see a thing anyway, and the joy of owning a piece of junk is not worth contemplating :-)
Re:Really, Flash Destroyer the best example? (Score:5, Informative)
Knowing the limitations of your tools is important whether they're $50 or $50000. I use my DP Bus Pirate every so often at work for programming the odd flash device or simulating an I2C master to debug I2C slave code on a microcontroller. I have a "professional" I2C/SPI pod that the company purchased that can't do many of the things this $30 board can.
Re:Really, Flash Destroyer the best example? (Score:4, Informative)
The sub-$500 PC-based logic analyzers like the Logic Sniffer and Saelae Logic still come in handy for low speed buses like I2C and SPI
In my experience it is essential to see the analog side of the signal, especially on the I2C (which is passively pulled up.) A lot of degradation can occur there, and a cheap logic analyzer will show garbage, if anything. You simply need to synchronize your scope well; sometimes you need to output something on a pin. A reasonably good digital scope will capture all you need to know, at any delay from the trigger and at any speed once the instrument triggers. Often you can see the trouble after one run.
With regard to profiling, I personally use one GPIO pin for that. On an AVR it takes one simple machine instruction to set it or clear it. Connect an oscilloscope to it and observe your timings. If you have two pins you have everything that is necessary to measure all the timings in the design (one by one.)
I don't know how you could use the abysmally slow I2C, or somewhat faster but still glacial SPI for profiling. That would require a fast-running timer (like on AVR32) and a cache of profiling data in RAM; but if you have that then SPI is just another debugging channel, not any better than a spare 3.3V UART that you connected to a 3-pin header.
One of good things about those busses (I2C and SPI) is that they are under your control. You know what is sent because you sent it. You know what was received because the hardware received it (and you will do well if you connect a debugging facility to that, if you need to.) A logic analyzer on such a bus is needed primarily when you are not sure what's happening. Again the roots of the thing go back into the world of one foot by one foot PCBs with SN74 DIP logic on them. Then you had to monitor tons of signals because you didn't know what is wrong. But if a 2-wire bus toggles, chances are your master is OK. If you see ACKs then you know that your slave is also OK. If you have issues with the nature of the data, it's datasheet time.
Still, as you say there are uses for everything. It's just I haven't seen much use of logic analyzers recently :-) You can't even connect them to modern busses like DDRx (and that's one tough bus to get timings right.) If you could they'd cost $50K, like PCIe bus analyzers, and you still need to instrument your board with MICTOR or like connectors that don't ruin your design right away.
Re:Really, Flash Destroyer the best example? (Score:4, Informative)
A good digital scope costs hundreds if not thousands of dollars
Sometimes you can get lucky and grab one for free or nearly free. I got a Tektronix 2440 [ebay.com] this way. It wasn't completely working, mind you, but it's something one can fix... if not then probably you don't need the scope anyway :-) There is a lot of old, well used and maybe a bit broken equipment around that nobody in a business wants. You just need to make contacts, look around, visit your neighborhood Weird Stuff [weirdstuff.com].
But there is something else you can do. Build your own high speed oscilloscope. Today it's not that difficult. Take AD9601, for example - it's a 300 MSPS A/D with dual (interleaved) parallel bus. You need also an FPGA to capture the data - some Spartan probably will do. Then you need a simple USB MCU to fetch the data from the FPGA and slowly ship it into the PC. Total three ICs, not counting the analog front end which is not a rocket science. You can generate the sample clock with a DCM in the FPGA. Build such a thing and it will be a great exercise. Such a scope will be not a toy, it will be a very useful, small device. Logic Shrimp is a logic analyzer, but this is a real scope - in all its 10-bit glory. You actually can measure analog signals with it. You can use even a lower resolution A/D, like AD9484. (Bits are necessary when you are doing DSP, not when you are just looking at things.)
DP and Seeedstudio (Score:4, Informative)
I have a few of DPs products. The webplatform is cheaper and more useful to me than an arduino/ethernet shield or even the new EthernetPro http://www.sparkfun.com/products/10536 [sparkfun.com]
Seeedstudio's fusion service is good price wise, but I will not have PCBs made through them again. Too many bad traces. Pads lift if you try to re-work them.
Slightly more expensive, http://dorkbotpdx.org/wiki/pcb_order [dorkbotpdx.org] , but I have never had a bad board and the quality is much much better.
-JC