flip-flop

They've got a sample button. What more do you want???

Rick C.

- Get 6 months of free supercharging - Tesla referral code -

That part was $317 before ADI bought Hittite.

Here's a 50 Gbps part:

Looks like they multiplied by 2. Soak the people who are committed to it.

Those may be Russian imports.

The Hittite part has a lot more output swing, and variable swing.

Unfortunately, we need a reset input, so we can't use either.

everyone here should request a sample and say we all work for "VANDELAY INDUSTRIES"

On a sunny day (Wed, 30 Jan 2019 08:33:18 -0800) it happened John Larkin wrote in :

Toggle supply?

Does it have a defined power on state?

two ECL 2:1 multiplexers are probably $20 in singles, an ECL inverter around $5, probably get you a D flip flop within an order of magnitude of that speed on the "cheap"

or discourage those who doesn't _really_ need it from buy what is left

And on LTB, so stock up. ;)

Beasty clock rate, but surprisingly crappy deterministic jitter--2 ps out of a 23 ps cycle time.

Cheers

Phil Hobbs

The 68 GHz to 76 GHz parts used for 5G cellular phones show it can be made very cheap. It would be nice if someone made simple logic devices for these frequencies.

It is on "last time buy", so this is a "don't touch me" price.

B4 ADI bought Hittite you could barely get a data sheet. Now the stuff is at DK and Mouser.

But the rise time is nearly as fast as the 9 ps of my fastest scope plug-in. I always wanted something that could stress that spec.

:-) Gerhard

I thought flip-flops were those cheap sandals.

There aren't may parts, ECL or anything, with edge rates in the 10s of PS. Even the EclipsPlus SiGe stuff is too slow for what we have in mind.

We don't care much about parts cost. Performance sells.

Some of those Hittite parts would make nice sampling scope testers.

It can't be that bad in real life.

I once couldn't get delivery on a Maxim part. I had a friend who taught at an EE school, so we had his entire class each request the max of 10 samples. I paid them for the parts, nice beer/book money.

Not sure. Not every node in that thing will respond at the same speed, so I could easily believe pattern-dependent effects at that level.

They spec the random jitter at 200 fs, which is quite believable too.

Cheers

Phil Hobbs

OK, deterministic jitter could be that bad.

We've measured the slower NB7V52 random jitter at around 18 fs RMS, most of which could have been in the test setup.

Yes, it may be they don't want to dedicate $100K test gear to test this part on the line for that spec, so they are relaxing the datasheet test limits. I agree, anything that can clock THAT fast couldn't have 10% jitter. But, you'd need some REAL fancy gear to actually test it.

Jon

As Phil notes, that's deterministic jitter, namely data pattern sensitivity. That would be measured with a pseudo-random NRZ data pattern out. Random jitter, measured with a nice periodic square wave output, is much less.

We have a LeCroy scope with a 1 ps RMS jitter floor; and it only cost $50K or so. It has an "enhanced" mode with 100 fs jitter.

We've done dirty tricks to measure jitter below 20 fs RMS.