Thanks Sean
I am launching a pulse into the air and measuring time of flight down  
to .0001 inches .
Since sound travels one inch in ~ 10**-4 seconds , I presume 10**-4  
inches will require 10**-8 resolution.
Gus

>
> On Aug 28, 2009, at 10:45 PM, Sean Breheny wrote:
>
> Hi Gus,
>
> Are the pulses 8ns long or are you saying they are much longer but you
> need to know the pulse width to within 8 ns?
>
> You will need to be concerned about not only sampling rate but also
> rise time (which is directly related to bandwidth).
>
> If you need 8ns pulse-width resolution, then the rise time/fall time
> needs to be less than about 1/4 of 8ns. This is because the pulse your
> scope sees will be reduced (in width) by 1 rise time and 1 fall time.
> You want this sum to be less than half of your required resolution.
> 1/2 times 1/2 = 1/4. So, 2 ns rise time.
>
> It takes about 3 time constants of a first-order system to rise (or
> fall), so you need a 2/3 ns (666ps) time constant. Bandwidth is
> approximately equal to 1/(2*pi*Tc) where Tc is the time constant. So,
> you need 240 MHz bandwidth at the probe tip. In other words, the total
> bandwidth of your circuit driving your probe driving your scope must
> be 240MHz or more.
>
> Then, you want the sampling rate to be 5 times this (if you don't want
> to do anything fancy) or if you are willing to do some careful
> interpolation between samples, you could get away with 2 or 3 times
> 240MHz. So, you are probably looking at a 1GSample/sec, 240MHz BW
> scope. The good news is that almost all modern digital storage scopes
> have sampling rates higher than 1 GSample/sec. The bad news is that
> the cost of one that can do 250MHz BW is much more than the typical
> low-end 100MHz or 60MHz model. Also, you will need a fast probe,
> perhaps even an active probe, to get this.
>
> It is pretty obvious, but I would look at Tektronix, Agilent, and
> LeCroy scopes. I have never rented one so I do not know how much that
> costs - I'm sure it's pretty costly. You could probably buy a used,
> working scope which is capable of this for about $1000 to $2000:
>
> Check out the following eBay items:
> 280389126103    LeCroy 9350AM Oscilloscope, Digital: 500MHz,1GSa/s,2ch
>  buy it now $1200
> 200348744074    LECROY 9370M DIGITAL OSCILLOSCOPE DUAL 1GHz + PP062
> (w/1GHz probe) buy it now $2000
> 160352930409    HP AGILENT 54520A 500 Mhz, 2 Channel Oscilloscope
> (warning: as-is) buy it now $650 or best offer
> 200318155708    LECROY 9354AM DIGITAL OSCILLOSCOPE QUAD 500MHZ 2GSa/s
> buy it now $1580
> 390087113847    TEKTRONIX TDS380 400 MHz 2 GS/s DIGITAL OSCILLOSCOPE
> buy it now $950
>
> Are these pulses always the same amplitude? There are other kludges
> you can do to get a pretty good handle on the pulse width without
> actually sampling them. For example, you could measure the average
> signal voltage during an interval, along with how many pulses there
> were in that interval, and then get the average pulse width from that.
>
> You could also build a little circuit which used a very fast
> comparator to charge a capacitor during the pulse on time. After the
> pulse was received, the voltage on the cap would be proportional to
> the on time. 1nsec comparators are obtainable for about $5, but there
> would be considerable design skill needed to get the whole circuit
> working right.
>
> Sean
>
>

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