EEVblog #316 - More PFANG, More 13GHz Scope, & More Pulser

[EEVblog]

Dave.

[KTP]

There are some comparitors with 50ps rise/fall times

This paper talks about making a low cost fast rise time pulse gen:

http://www.ise.pw.edu.pl/~tomi/papers/wilga2006_50ps.pdf

The ADCMP580 used in that discussion data sheet is here:

http://www.analog.com/static/imported-files/data_sheets/ADCMP580_581_582.pdf

Sells for $18 at digikey...ouch.

From the data sheet:

"The ADCMP580/ADCMP581/ADCMP582 are ultrafast voltage comparators fabricated on the Analog Devices, Inc. proprietary XFCB3 Silicon Germanium (SiGe) bipolar process. The ADCMP580 features CML output drivers, the ADCMP581 features reduced swing ECL (negative ECL) output drivers, and the ADCMP582 features reduced swing PECL (positive ECL) output drivers.
All three comparators offer 180 ps propagation delay and 100 ps minimum pulse width for 10 Gbps operation with 200 fs random jitter (RJ). Overdrive and slew rate dispersion are typically less than 15 ps.
The ±5 V power supplies enable a wide ?2 V to +3 V input range with logic levels referenced to the CML/NECL/PECL outputs. The inputs have 50 ? on-chip termination resistors with the optional capability to be left open (on an individual pin basis) for applications requiring high impedance input."

[veryevil]

Hey, The letter says that you should have used a 20Mhz SINE WAVE not square wave input! Guessing that would have cut down on the ringing.

Now you have to build one and try it again!

[SeanB]

I did something similar dead bug style to get a better clock out from a PC interface card. 54S14 dead bugged 4 high, with a 0.1uF across the top, and a 10uF 25V wet tantalum as well over that.  Just used 1 inverter from the bottom to drive the 3 above it, and they in turn drove the rest of the inverters in parallel. Had no problem driving a 24MHz clock direct to 110R twinax with only end termination. Did run a little warm though, but the cerdip package was happy with that. If they had toasted I would have made another and would have added copper strip fins between the packages.

[Nermash]

One thing intrigues me about that 13 GHz scope, I kind of expected to see SMA or more advanced sexless connectors on it (can't remember the name).
From all the sources I have read, BNC is good only up to 3 GHz?

[EEVblog]

One thing intrigues me about that 13 GHz scope, I kind of expected to see SMA or more advanced sexless connectors on it (can't remember the name).
From all the sources I have read, BNC is good only up to 3 GHz?

I was surprised too.
But Agilent know what they are doing, so they clearly they have special ones that work to 13GHz.

Dave.

[LEECH666]

Dave, have you won the lottery recently? I wonder where all that super high end gear comes from.

Cheers,
Florian

[Monkeh]

Dave, have you won the lottery recently? I wonder where all that super high end gear comes from.

Cheers,
Florian

Loaned for review, I believe.

Do we get to see inside the obscenely expensive oscilloscope?

[T4P]

One thing intrigues me about that 13 GHz scope, I kind of expected to see SMA or more advanced sexless connectors on it (can't remember the name).
From all the sources I have read, BNC is good only up to 3 GHz?

F connector
N connector

[Monkeh]

One thing intrigues me about that 13 GHz scope, I kind of expected to see SMA or more advanced sexless connectors on it (can't remember the name).
From all the sources I have read, BNC is good only up to 3 GHz?

F connector
N connector

F? F?!

[Lukas]

The NBSG series by OnSemi offer ~40 ps rise/fall times. They are horribly expensive around 50$ on digikey, but onsemi offers free samples The most difficult part in designing a pulse generator using a NBSG16 or so would be the PCB...

[EEVblog]

Loaned for review, I believe.

Agilent heard I needed a high speed scope to measure the pulser, and they kindly loaned me one next day.
Wasn't sent for "review".

Do we get to see inside the obscenely expensive oscilloscope?

Sorry, no, was asked not to take it apart. You can see some insides in the data sheet.

I have some video of the "unboxing" in the garage, but not sure if anyone wants to see that. Probably just more complaints about Agilent owning the blog 

Dave.

[Monkeh]

Loaned for review, I believe.

Agilent heard I needed a high speed scope to measure the pulser, and they kindly loaned me one next day.
Wasn't sent for "review".

Ah. Not that they're uninterested in your comments (like touchscreens being annoying as hell).

Do we get to see inside the obscenely expensive oscilloscope?

Sorry, no, was asked not to take it apart. You can see some insides in the data sheet.

I have some video of the "unboxing" in the garage, but not sure if anyone wants to see that. Probably just more complaints about Agilent owning the blog 

Dave.

Fair enough. And unboxing is a fairly boring process unless it's something which needs putting together.

[free_electron]

I was surprised too.
But Agilent know what they are doing, so they clearly they have special ones that work to 13GHz.

Look very closely at that connector. It is NOT a traditional bnc.

- For starters there is no isolation sleeve around the center contact.
- the center contact is not 2 pronged like in a classic bnc ( a tube sawed in two halves)  but has a 6 prongs much like what you see on N connectors of their network analyzers.
- If you look deep in the connectors you will not see the classic white plasitc or teflon isolation material but a yellowish material, almost like bare FR-4 material. That is Torlon. Works up to 45 GHz without problems.

It looks like a bnc, and it mates with bnc, but it is a notably different construction... if you look a their HF mating probes that go up to those bandwidhts you will see a different construction there as well. These thing are more N-connector than bnc when it comes to the center pin.

the sexless connector real name is APC-7. those go beyond 15Ghz... but need alcohol cleaning and a torque wrench to set... Amphenol Precision Connector 7 mmillimeter.

the really high performance scopes like the 90000Q series use 3.5 mm connectors looks like a reverse SMA but is not quite the same ...
Those scopes are really extreme.. up to 63GHz analog bandwidth... 160 Gigsamples per second and 2 gigapoints, yep GIGA not mega ! deep memory per channel....

Read and cry your eye's out ... http://cp.literature.agilent.com/litweb/pdf/5990-9712EN.pdf
- Indium phosphide chips ( silicon germanium is for grampa's and gallium arsenide for losers .... )
- Can measure rise times as fast as 5pS ... you really need that to see wha tthe jim williams pulser can do ...
- the probes have a 200GHz fT ...
- has a built in 'calibrator' pulse rise time : 15pS .. eat that JW-pulser ...

ah well, at half a million US$ for a base setup ...

[Nermash]

Thanks for the explanation, that is some very interesting info!

[T4P]

Half a million! Wow!
Sadly there's something more frightening in this world, than the scope agilent loaned to dave jones
A basic car for the UK people like the toyota's cost maybe 10k pounds or less?
Over here it got so scary it costs at least 120,000 SGD at least! That's for an 13GHz 40GSa scope!

[T4P]

Half a million! Wow!
Sadly there's something more frightening in this world, than the scope agilent loaned to dave jones
A basic car for the UK people like the toyota's cost maybe 10k pounds or less?
Over here it got so scary it costs at least 120,000 SGD at least! That's for an 13GHz 40GSa scope!
Well a house way smaller than dave's house (or for reference, a 5rm flat that i have no idea how many sqft big it is)
goes for 200,000 SGD at least more than the 20GHz 80GSa scope!
(Read: A 5rm flat that is small but easily starts off at 700k! I want to live elsewhere ...)

[blackdog]

So mutch mony en still 8 bit resolution 

[free_electron]

this gives you an idea of what this kind of beast is used for :

http://www.home.agilent.com/agilent/redirector.jspx?action=ref&cname=AGILENT_EDITORIAL&ckey=2158864&lc=eng&cc=US&nfr=-11143.0.00&pselect=SR.GENERAL

[SeanB]

Dave, you are welcome to come live in ZA, no problem at all.

[SeanB]

Yup, ordered parts at previous work and saw the invoice. $100k each for power supplies........... They fit in the palm of your hand, and are about 150W each. Manufactured by Thompson CSF, now a part of Airbus Industrie.

[Alexei.Polkhanov]

There are some comparitors with 50ps rise/fall times
This paper talks about making a low cost fast rise time pulse gen:

If I see what I see - they give a screenshot of 300mV signal with sharp rise time. What this kind of pulse is useful for? It won't even reach a threshold voltage of any laser diode in my possession :-) Many applications of pulse signals also require significant power to be delivered such as with 60-100V pulsers. That is the problem with all this pulsers built with fast logic devices for many practical applications such as lasers or time domain reflectometry kind of stuff you need more power.

[T4P]

Dave, you are welcome to come live in ZA, no problem at all.

I know i know 
Nice place and such but generally i am a little afraid of ZA

Yup, ordered parts at previous work and saw the invoice. $100k each for power supplies........... They fit in the palm of your hand, and are about 150W each. Manufactured by Thompson CSF, now a part of Airbus Industrie.

100k in SAR? That's a lot 
But seriously, 500k here equates to 250 UK pounds

[SeanB]

Dave, you can literally stay for free......... 1 million refugees can't be wrong.

I ordered and got 10 of these......... The price was in US dollars BTW,not ZAR. These did not even have a case, and were a component.

[T4P]

Dave, you can literally stay for free......... 1 million refugees can't be wrong.

I ordered and got 10 of these......... The price was in US dollars BTW,not ZAR. These did not even have a case, and were a component.

jaw SMASHING