[Noisebridge-discuss] cheap USB Analyzer, or signal integrity tools?

[Michael Prados]

One of the reasons I liked the comparison review I cited above was
that it did a mini-teardown on the three USB analyzers it considered:

http://www.summitsoftconsulting.com/UsbAnalyzers.htm

Each of them seem to have a pretty similar architecture:

1. USB transceiver- takes care of the input differential waveform and
any shenanigans on the line.
2. FPGA- breaks down the USB transactions, ties everything together
3. SDRAM- buffers the USB data
4. USB controller- to talk to the host PC

For the cheapest option they test (the International Test Instruments
1480A hardware, which does 480 Mbits/s and retails for $500,) those
components appear to be:

1. An NXP (now ST-Ericsson) USB transceiver like the ISP1504A.  $2.00
@ quantity 1
2. An Altera Cyclone II FPGA like the EP2C20F484C7N, $56.30 @ quantity
1, with a dev kit available at $199
3. Some 32 MByte SDRAM, perhaps something like the Micron
MT48LC64M4A2P-75:D TR, $12.67 @ quantity 1
4. Cypress FX2LP USB controller, $11.40 @ quantity 1

So, we are looking at a BOM starting at $82.37.  I'd hazard a guess at
$60 for the rest of the components, and a 4 layer board of this size
at $200 @ quantity 10-20.  So perhaps $350 for a small run.

I'm a mechatronic engineer, which means I am an electrical engineer
once every three days :)  The PCB design on this would be challenging,
because you have to sweat the signal integrity a bit.  Getting from
the current state of open FPGA cores to what would be necessary to
implement this is certainly a little intimidating.  The host side
software is also a bit of work.

This is all a bit beyond what my time budget currently allows for, but
it could be a worth and potentially profitable endeavor.  I don't
think an open source product like this could blow the competition away
on a features for the price basis, but I agree that the open source
nature would be selling point.

Of course, none of this addresses signal integrity testing.  To go
there, we'd need a high bandwidth ADC, a real clean analog path, and
probably more horsepower out of an FPGA or perhaps DSP.

-Mike



On Mon, Nov 8, 2010 at 9:45 AM, miloh <froggytoad at gmail.com> wrote:
>
>
> On Mon, Nov 8, 2010 at 8:28 AM, John E <neurofog at gmail.com> wrote:
>>
>> What about using a FPFGA + High Speed AD Converter to make a DIY one?
>> John
>
> I checked opencores.org, and found a few quasi-related high speed USB
> projects, but It looks like there's a lack of good open hardware/open core
> signal analyzer projects  there (I haven't checked the forums for discussion
> yet).  A new project with the same price range as the ones linked to in this
> thread (~1-5k range) would still be more attractive if it was rooted in open
> design with *nix compatible software components.
>
> It would be great way to get more folks sharing digital design skills and
> developing drivers for things like SANE at Noisebridge.
>
>  Lets sketch it out more, it sounds like a very big project with some
> preliminary questions to answer:
>  1) Why aren't there widely known open source high frequency analyzer
> projects?
>  2) Why are the cheapest high Fq. (nS resolution) signal analyzers still
> around the  $1k price point (and why are the cheaper ones cheap)?
>
> Are there any new families of FPGA's that *are* fast enough for 2.0 and
> still affordable to work with?   Are the analyzers linked previously all
> using asics or what?
>
>
> -r. miloh
>
> _______________________________________________
> Noisebridge-discuss mailing list
> Noisebridge-discuss at lists.noisebridge.net
> https://www.noisebridge.net/mailman/listinfo/noisebridge-discuss
>
>



-- 
[REMOVE THIS TEXT BEFORE SENDING AN EMAIL!]