Very fast oscilloscopes

Bleeding edge equipment gets old the fastest.

Their 32 GHz ones are nearly $300k. You can get a 50 GHz 11801C with lots of good modules for a couple of percent of that number.

Cheers

Phil Hobbs

Thanks Phil, I found a price list in their product announcement here:

I wouldn't be surprised if the *probes* were in the high-four digits either -- maybe even hitting $10k+!

It is kinda ironic that the 1st page there shows an eye diagram where you likely could get by just fine with a sampling scope...

...but I do expect it's pretty cool to see, e.g., a non-repetitive high-speed bitstream like SATA or PCI-E or similar at, say, 6Gbps go marching by...

---Joel

If you can trust the seller that the sample heads aren't shot. This can be a major problem with older Tek spectrum analyzers. Sometimes the first mixer is shot and so far I haven't heard from anyone of a successful low-cost fix, seemingly because this device is jammed into it really deep.

The last decent oscilloscope HP built was the 130C. It has been steadily downhill ever since.

Their idea of "trigger" was a horse.

I said that once, in a rather grumpy way because the thing wouldn't trigger. Turned out the guy standing next to me had been on the design team of one of those ... whoops ... I wanted to vanish into the next hole.

$30K for the new Agilent probe, I've been told.

New car or new probe?

John

An 11801 series frame with an SD-24 is a fabulous instrument for around $2500 maybe.

ftp://jjlarkin.lmi.net/DSC01371.JPG

That's a 1 GHz square wave being displayed.

I've had pretty good luck with used sampling heads from ebay. I must have 30 or 40 of them, theoretically worth about a half a million dollars.

We've been successful in fixing one common failure mode, a dead serial eeprom.

John

Nice. I'd put the vise up there somewhere else, in case the big one hits. Wouldn't want that falling onto my head ...

If nothing is busted in the wire bond area that is good. But for many sellers it's probably hard to say whether the head is dead because of a bad EEPROM of because someone had exposed it to the spike from hell.

ttp://cp.literature.agilent.com/litweb/pdf/5990-5271EN.pdf

I worked for EG&G Energy Measurements in the 1980's, up to 1997. Another EG&G organization near Boston built ultra fast scopes for use in underground nuclear testing. They had their corporate knowledge of how to build fast real-time scopes but the limit of their expertise was about; 10 GHz bandwidth, rise-times of less than 350 ps, non- gaussian. Then, in the early 1990's they hired a guy who applied electron LINAC techniques for the beam steering and they broke the 10 GHz barrier.

I got to see one of their new creations in 1995. They took it to the Naval Standards Lab (now NIST) and it responded in real-time to the maximum frequency the lab could generate: 54 GHz; with a sensible, measureable display. We hooked it up to a Berkely super fast pulse generator with a claim of sub-100 ps rise time, and between the scope's rise-time and the pulse generator's rise-time, measured 165 ps.

Needless to say, I was very impressed.

I understand they only made 2 of these, but I left to work for another company shortly, thereafter, so I could't keep up with the latest internal news,

Tom Pounds

For what purpose? You would need tons of storage to capture some meaningfull information. Not to mention trigger algorithms.

If you have signal integrity issues you could input a continuous square wave and check the eye diagram with an 'ordinary' sampling scope.

"Real time"?? Does that mean it can do single shot events giving a continuous (analog) trace of the event?

Watching for glitches, runts, or other improprieties that an eye diagram (or sampling scope) wouldn't catch.

Correct. (Although that analog trace is digitized, of course.)

Thanks; that reference was way too big for dial-up.

It's a little scary that such an expensive piece of precision hardware runs Windows. Whatever happened to using simple embedded operating systems that don't have a zillion extra features to crash?

IMHO at those frequencies a glitch would exceed the driver's frequency response. I bet those signals look (much) like sine waves.