On Tue, 09 Mar 2010 22:53:25 +0000, you wrote:

>-----BEGIN PGP SIGNED MESSAGE-----
>Hash: SHA1
>
>M. Adam Davis wrote:
>> Idea:
>> 
>> Take one oscilloscope probe, and one USB cable. Cut the BNC connector
>> off the oscilloscope probe, and the USB B connector off the USB jack
>> and splice them together.
>> 
>> Mount all the USB <--> analog circuitry inside the scope probe body
>> and/or the USB A connector body. It would essentially compete with all
>> the other USB oscilloscopes out there, but it would be in a single
>> cable design - no extra boxes, etc.
>> 
>>    1. Is this a product you would like?
>
>Definately.
>
>>    2. What are your minimum specifications for this type of product?
>
>At least 50MSps for the work I'm currently doing with a realtime trace
>and half decent triggering. Comfortable in the hand!
>
>>    3. What are your ideal specifications for this type of product?
>
>200MSps would be pushing it, no?
>
>>    4. Would this be any more useful or desirable than the current USB
>> oscilloscopes available?
>
>Yes. Low power and less desk clutter is never a bad thing.
>
>> My biggest concerns are:
>> 
>>     * Obtaining the speed necessary to be reasonably useful without
>> requiring a lot of components
>>     * Synchronizing multiple probes into the computer (Getting nS
>> accuracy without wiring the probes together seems problematic...)
>>     * Fitting it all into the form factor and power limitations of a
>> USB port (ideally 100mA unpowered hub ~0.5W)
>>     * Keeping the assembled cost low
>> 
>> I'd appreciate feedback on any or all of the above (what you want vs
>> how to implement it). Ideally it'd be completely open source, but
>> using very tiny surface mount parts it might not be easily assembled
>> by hand.
>> 
>> -Adam
>> 
>> ---
>> http://chiphacker.com/ - EE Q&A site
>
>I have definitely thought about that idea a few times. The main problem I
>found was being able to isolate the USB power from the signals in a small
>and cheap way.

I think the answer here is probably to isolate at the ADC interface, as you have more control of the
nature of the signals at that point, and the DC-DC doesn't carry the load of the digital stuff. Also
helps isolate the digital noise from the analogue stuff.

>The next fun part is choosing an MCU with good enough raw data->USB
>throughput for any kind of decent sample rate. The more you actually
>delve deep into the delays associated with getting that data into the PC,
>the more a small FPGA (Actel IGLOO, etc) and USB2.0 PHY seem desirable.
>
>Anyway, if your design gets further than just initial scribblings you
>deserve a pat on the back.

The FTDI FT2232H USB2 hi-speed chip might have some interesting potential for a cheap high-speed
scope. 
A small FPGA with  enough buffer RAM to cover latency and add timestamps, probably wouldn't need an
MCU at all. 

I would see this as mainly a tool for use on-site on a laptop rather than on the bench, so isolation
is maybe not a huge issue. 

I'd say the ability to sync multiple probes is essential even if it does need an external lead. 
An alternative might be a second analogue channel with a small connector for a second probe. 

If it's all in the probe, it would be good to be able to attatch a 'proper' scope probe, i.e. the
end is a BNC into which you can plug either a short plug-mounted probe or a proper probe or other
BNC lead. 

I think you'd probably often want a proper probe, so maybe a 'lump in cord' solution isn't too bad -
it could have a USB B socket so you aren't carrying around a dedicated attatched USB cable.
lump in cord also makes the possibilities for interconnect of multiple channels easier - the 'lumps'
 plug together somehow. 


-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist