SD-24 guts

John Larkin a écrit :

The funny thing is I just grabbed one on feebay for a good price and was about to ask your opinion about how easy it is to repair...

Now I really hope it's working OK :-)

Oh my. I thought my S-6 was hardcore.

About all you can fix is stuff outside the hybrids, like a dead eeprom or something mechanical.

I can't imagine how much engineering went into this.

John

Oh man, that doesn't look like it's going to be fun. I hope you've got at least a schematic and a wire bonder. But in the Bay Area it should be possible to find a shop that can do the bonding if needed.

I wonder if, with today's materials, it would be possible to achieve similar performance with regular SMT parts on a material such as Rogers.

We're just replacing dead eeproms. I wouldn't dream of touching the wire-bonded stuff. Tek never revealed schematics on the 11801-series scopes anyhow.

I met Agoston Agoston at a trade show once. He designed most of this stuff. He has his own little company now, and did a 20 GHz sampler with surface-mount parts on FR4. Check out his Tek patents. AA had a deal with Tek that they wouldn't sue him if he didn't go past 20. He was also involved in the PSPL shock-line 100 GHz super-sampler project, which may or might not still be a real product.

There's a paper on this page about the 100G thing:

I got a half-bridge sampler to work nicely at 7 GHz without a huge effort.

ftp://jjlarkin.lmi.net/Sampler1.JPG

I'd think that 20 GHz on FR4 should be reasonable with available surface-mount parts. Aeroflex just announced some 0.08 pF packaged schottkies.

I just got an SD-30 (50 GHz) sampling head for $1500. There's tons of cheap Tek gear on the market now.

John

What do you suppose that one long wire bond is for (right in the middle near the 224BF marking)?

Bob

Beats me. The whole thing is a mystery.

The black spec in the very center is an SRD sitting on two parallel-plate caps. It appears to drive slotlines in both directions, towards the sampling diodes. The wirebond likely tunes some behavior of the slotlines; it appears to connect "ground" to "ground." The other channel doesn't have it.

Even stranger are the trim tabs above the top of the arch of that wirebond.

This looks like a 6-diode traveling-wave sampler using beam-lead diodes. The TDR step generators are in separate cavities.

John

Not only that, but that's 70's technology. They were way ahead of the curve at the time, but what did it cost ?.

If you have stuff like this now you need to take good care of it. Replacements are getting scarce and there's little affordable modern kit to replace it with...

Regards,

Chris

Hi, John,

I can ping your server, but when I try to get the file, the connection is reset. (Firefox or Filezilla) This has happened for weeks. Does anybody else have this problem?

regards, Gerhard

It's supposed to be an anonymous FTP site. I've never had trouble pointing other people to the files, using various browsers. Maybe somebody who knows more about this sort of thing can offer suggestions.

John

If the mainframe doesn't recognize it, it's a dead eeprom, which can be fixed. Anything else is usually un-fixable. Often one channel will be zapped on cheap ebay heads, so you at least get to use the other one.

John

I suspect that sampling phase detector (SPD) modules might provide the easiest way to roll your own sampling head beyond about 5GHz. These modules include an SRD coupled by small caps to a series pair of Schottky diodes, and seem to be stocked by some distributors. The integration of SRD, caps, and Schottkys ought to avoid the killer parasitic inductances of common surface mount packages without resorting to beam leads or wire bonding.

I don't know if the unspecified(?) series caps in e.g. the Skyworks SPD would give an ideal sampling pulse shape for sampling oscilloscope work. Another interesting question is whether the SRD in a second SPD (these are, after all, fast SRD's stocked by some distributors) could be pressed into service in a pulse generator usable for TDR.

James

No. ...Jim Thompson

-- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

In Memoriam... Duane Lee Thompson October 31, 1972 - April 20,2006 4th Child, 2nd Son, of Jim & Naomi Thompson Victim of Colon Cancer Would Have been 37 Today

I've looked at the sampling phase detectors, but I suspect they wouldn't make good oscilloscope-type samplers. Really tiny surface-mount parts, 0402 and SC79 type stuff, can go pretty fast these days.

There are some CML gates and comparators that would make nice TDR step generators, in the 15-40 ps risetime range.

One thing you can do nowadays is build a sampler or TDR that has yukky step response - ringing, overshoot, whatever - and run it through an equilization algorithm to make it beautiful. That's a lot easier than spending man-years tweaking all the uglies out. Deriving the FIR equalizer is an interesting problem. This is "the deconvolution problem" a member of the class of "ill-posed problems."

Fun to play with.

John

I'm using Firefox. I see it fine.

-- Cheers, James Arthur

Requires registration but was painless. Nice paper, wish he had at least put a photo of the open enclosure in there.

Highest I've gone on FR-4 was 2.5GHz. Would have done 5GHz as well but so far no clients needed that. 20GHz is truly brazen but sometimes we have to be brazen.

True. It would be nice though if someone offered a budget sampler head for regular DSO's or maybe even laptops so you can take it on the road.

WHAT! You didn't negotiate, or they wouldn't budge? ;)

Late at night, by candle light, Gerhard Hoffmann penned this immortal opus:

Works fine for me in Firefox.

- YD.

Gerhard, I can view the pictures from a fast connection here at work, But when I use dial up from home I can't view them... I assume the connection times out. Don't know if that is your problem or not.

George H.