Re: [SI-LIST] : What speed scope should I consider?

About this list Date view Thread view Subject view Author view

From: Chuck Hill (chuckh@altaeng.com)
Date: Fri Nov 12 1999 - 11:17:24 PST


At 09:34 AM 11/12/99 -0800, Mark Randol wrote:
>Larry McMillan wrote:
>> They do. It's a 2GHz model called the TDS794. We have one and it gives
>> very satisfactory performance. By the way, the Tek 580D is exactly the
>> same as the 784 except that it has a black and white display instead of
>> color and it costs a whole lot less. We have that one too, and it is
>> also very good. Got rid of a couple of our 1GHz LeCroys when we got
>> these, and now only use our two remaining LeCroys when both Tek scopes
>> are already occupied. Tektronix doesn't make a black and white version
>> of the 2GHz 794 or else we probably would have gotten it instead too.
>
>OK, maybe I don't get it being a mostly analog type, but why is a
>2GHz analog bandwidth with a 1-4GSa/s rate better than a 1GHz one
>with a 2-8GSa/s rate? Sure, according to Nyquist you could see
>up to 2GHz with a 4GSa/s rate, but if you're going to be able to
>see much, you need a sample rate about 5-10X the analog
>bandwidth.
>
>Going with ROT (rule of thumb) BW~=0.35/trise, a 2GHz BW means
>you can see a
>trise>=175ps. But to hit it with a single sample point you need
>a >5.7GSa/s rate. If you get lucky and hit it on the start and
>end of the slope, you've got a pretty good measurement then, but
>how many points on the slope do you need to really get it? Lets
>just assume that 2 points on the slope will do it. To guarantee
>you get at least 2 points on the line, you'll need twice the
>sample rate, or >11.4GSa/s, or tsampling<=87.5ps. If just having
>two points 50% of the time is ok, then 131.25ps or 7.6GSa/s is
>sufficient.
>
>If you have the luxury of being able to use repetitive waveform
>sampling, you can get away with a lower sample rate at the risk
>of missing some events. Personally, even though I know it can
>work, I don't much trust it if it isn't "real time", or I have
>other ways of verifying it.
>
>Also, being mostly an analog type, I'd be concerned with aliasing
>the frequencies over 0.5GHz when running the scope with the
>1GSa/s rate on each channel. In reality, I don't figure that a 1
>or 2GHz scope (analog or digital BW) in my mostly analog
>applications is much use above a couple 100MHz anyway.
>
>So what advantage is there to a 2GHz analog BW vs 1GHz when the
>sample rates is halved? What am I missing? (I'm comparing a
>TDS794 vs an LC584)

From one "mostly analog type" to another, the aperature width of the input
sampler determines the maximum input bandwidth, as well as the preamp
chain. The repetition frequency of the sampling pulses are the samping
rate (of course). Aliasing can occur.

Repetitive random sampling is useful. I'm careful about measurements in
general unless I have other ways of verifying it. Repetitive random
sampling is a technique used when Nyquist is not met by the sampling
frequency and input bandwidth (or aliasing can occur). An example is the
HP54120, input bandwidth 18GHz and sampling rate of 40MHz. How can you
build a digital memory with >36Gbytes/sec transfer rate? If the input
signal is not repetitive, then you get funny results. But that is also
true for spectrum analyzers (non-digital) because a short transient pulse
gets smoothed out by the IF filter. Network analyzers measure steady state
impedance, not transient impedance. By the way, for certain kinds of noise
problems, my Tek465 completely analog scope can see stuff my digital scopes
cannot.

Bandwidth is the most obvious performance measure for scopes. But there
are many other parameters which are important and may not be specified. I
"fixed" a noise problem once which turned out to be crosstalk between
channels in the scope (I won't mention whose scope it was). What's the
noise floor? Dynamic accuracy of the ADC? How good is the trigger?

My advice on determining which scope is to talk with the sales people.
I've had good results from guys from LeCroy, Tek, and HP. You might need
to talk with their more technical folks, though.

Charles Hill, consultant

>
>Thanks,
>--
>---------------------------------------------------------------
>Mark Randol, RF Measurements Engineer | Motorola SPS, Inc.
>(480)413-8052 Voice | M/S EL381
>(480)413-4150 FAX | 2100 E. Elliot Road
>ryvw50@email.sps.mot.com | Tempe, AZ 85284
>---------------------------------------------------------------
>
>**** To unsubscribe from si-list: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list, for more help, put HELP. si-list archives are accessible at
http://www.qsl.net/wb6tpu/si-list ****
>

**** To unsubscribe from si-list: send e-mail to majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP. si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list ****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Tue Feb 29 2000 - 11:39:46 PST