Oscilloscope Bandwidth!

"Myauk" a écrit dans le message de news: snipped-for-privacy@31g2000cwt.googlegroups.com...

You are a couple of years late... With sampling scopes (ie working in undersampling mode on repetitive signals) the best scope you can find are as far as I know the Lecroy Waveexpert series... with a bandwitdh of 100GHz ! For realtime sampling you can find products with bandwidths up to 15GHz/40Gsps (TDS6154C from Tek or similar from Agilent), or even 18GHz/60Gsps with more specialized serial data analyzers (Lecroy SDA18000). Of course this all depends on how deep your pockets are...

Friendly, Robert

Nope, try 100GHz analog bandwidth for repetitive sampling scopes, like this one:

Huh? Same way they always have! :->

Repetitive or single shot? Do you know the difference?, it's important... In either case you do it with a very careful probing, and specially designed high bandwidth probes.

You want to measure a 4GHz clock signal? If that is the case then you need a VERY serious oscilloscope, a VERY serious amount of money, and even more serious probing. If you have to ask this sort of question then I'd question whether or not you actually have the ability to build or measure this sort of stuff, even if you had the right gear. And no, a 4GHz bandwidth scope will not allow you to view the characteristics of a 4GHz clock signal.

Dave :)

Hi Myauk,

You will never have success for probing a 4 GHz microprocessor clock, because this signal is not available on any pin outside the chip. Even if you find a clock signal > 500 MHz outside a processor be aware and don't touch this. You will destroy this clock with your probe. The useful frequencies for monitoring wit an osci end at about 20 to 100 MHz depending on your equipment and knowledge. All above is very special for very high sophisticated users solvinf´g very sophisticated problems with a huge budget.

Marte

I am not working on it this time. I am just curious to know about it because the application note from Tektronix wrote something about high speed circuit designs. The signals I am measuring are some audio signals, video signals, and the clock signal not higher than 100MHz for the signal processing chipset. Any way, I am learning from your answers. Thank you. Regards

Fet probes now go into the low GHz with fractional pF capacitance. Right now I'm using a Tek SD-14 sampling probe, an ebay find, which has better than 3 GHz bandwidth at the probe tip and about 0.3 pF loading.

John

In the early 1990's, EG&G/Energy Measurments came up with an analog oscilloscope tube (not a sampling 'scope) and the accompanying amplifiers and interfaces that was good to 18 GHz, and would still function out to 54 GHz (the highest frequency the Naval test lab could produce). These oscilloscopes were to be used in nuclear weapon testing, but the US quit testing nuclear weapons in 1992. Five were built. I was privileged to see one of these units in operation in

1993. To see a 55 pico-second one-shot (not sampled) rise time was pretty cool.

In the 1980's and 1990's, some French company had also produced analog oscilloscopes with 10+ GHz bandwidths for use at CERN and other high-energy physics experiments.

Bottom line: fast analog oscilloscopes exist, but at a high price.

A 55 ps risetime computes to about a 6.5 GHz bandwidth.

My friend Bernard still makes it:

It uses a traveling-wave-deflection scan converter tube inside. It used to be relabeled and sold by Tektronix.

Tek 7104's (1 GHz, microchannel plate CRT) are available on ebay. Nice scopes.

John

Massive Precious Knowledge for me. Thank you so much. I got initiative to own my first oscilloscope. I hope I need to buy one for my experiments! Regards

Massive Amount of Precious Knowledge for me! Thanks.

John Lark> A 55 ps risetime computes to about a 6.5 GHz bandwidth.

John, I'm getting a different answer. From what I can remember, 55 ps would give a 250 GHz bandwidth:

55 ps risetime (10%-90%) computes to about 25 ps 1/e time (or "RC time constant") 25 ps time constant computes to a bandwidth of 1/25ps = 40 G-rad/s = 2*pi*40 GHz = 250 GHz 18 GHz should correspond to a 770 ps risetime.

Am I figuring this correctly?

Mark

The rise time and bandwidth of a signal are approximately related by the following equation:

Signal bandwidth = 0.35 / Signal rise time

For a 55 ps signal we have:

BW = 0.35 / 0.055 ns = 6.36 GHz

Now, to *measure* this signal with some useful accuracy we need an oscilloscope with a much higher bandwidth than 6.36 MHz.

--
Regards,
Costas
_________________________________________________
Costas Vlachos  Email: c-X-vlachos@hot-X-mail.com
SPAM-TRAPPED: Please remove "-X-" before replying

In my last sentence I meant 6.36 GHz of course!

--
Regards,
Costas
_________________________________________________
Costas Vlachos  Email: c-X-vlachos@hot-X-mail.com
SPAM-TRAPPED: Please remove "-X-" before replying

= 40 G-rad/s = w = 2.pi.f

so f = 40G/2.pi = ~6.5GHz.

(You multiplied by 2 pi instead of dividing).

No :)

Don't worry, I did exactly the same thing a few months ago. For a while I was trying to design a photodiode amplifier that was 40 times as fast as it needed to be. Doh.

--

John Devereux

That was the fastest rise-time pulse generator we had in the lab, at the time. I was told the 'scope could do much better (i.e. showing a real-time sine wave display at 54 GHz at some Naval test and measurement lab).

Thanks for the link. I always wondered about who 'they' were.

The EG&G 'scope, I am told, was a breakthrough for the guys in Woburn, Mass. because they changed their way of thinking about the electron-gun and deflection system: from fine tuning a TWT, to treating the system like an electron linear accelerator.

Right! I forgot about the 7100 series systems.

Tom

I've finally become a full convert to digital, color, LCD scopes. My personal scope (in my office) is a Tek TDS2012, which handles 90% of what I do. The digital storage, averaging, and FFT things are neat, as it the fact that you can hold it in one hand.

John

Frackin' A, you're right. Doh!

Thanks,

Mark

Thanks Costas, I've got it now. My 2*pi factor went the wrong way.

0.35 = (ln(0.9)-ln(0.1)) / (2*pi)

Regards,

Mark

Most of the useful energy in an edge is contained within the bandwidth defined by (roughly) 0.5 / tr which in this case would be 9.1GHz, so a

10GHz scope might suffice.

Dave :)

I carried my pair of 7904's from MIT to Worcester square, in the south end, with a backpack full of books. I regretted it the following day. Incidentally, I bumped into Win there that day, also looking at books, although I'm sure he doesnt remember (although I waffled about the 3MJ flywheel I was working with) :)

Cheers Terry