ECL input's impedance at high frequencies

[rfmw]

senad lomigora

At low frequency (few hundred MHz range) an ECL input has high-z impedance. But what impedance has ECL input at a few GHz range? Surely it has not high-z impedance

To be more specific, what is the impedance of ON Semi's ECLinPS Lite/Plus (or similar of other manufacturers) input at 1-3 GHz range?

Is series termination possible at all with these inputs, since absolute value of the reflection coefficient of the input is not 1...

thanks,
rfmw

 

[moda]

IN What configuration do you want to use ECL gate? If you use 50 Ohm termination (near the gate input) you are OK till more then 1GHz, If not then you cannot guarantee what will be the impedance.

 

[djalli]

Check datasheet what it says because in GigaHertz range things are not as in MegaHertz range.

It will become microwave domain and things are treated a little different in that domain.

 

[flatulent]

ECL inputs are a bipolar differential pair. The manufacturer may have a SPICE model you can use to do the simulation/measurement.

Try https://www.onsemi.com/site/support/literature/list/0,4456,1162_273,00.html for starters.

 

[rfmw]

Well, thanks guys for your good intentions....

Currently I'm running these ECLinPS Lite chips of ONSEMI at 2.5 Gbps, so needless to say that I'm working in the microwave range already and RF is not something I'm not familiar with.

What I need is an advice. When you approach ECL flip flop (i.e. mc100el31d) upper frequency toggle limit (2-3 GHz), many bad things happen: clock crosstalk to the outputs Q,/Q and inputs S,R,Data rises to approx 10-30% of ECL logic level, input impedances of flip flop drops from high-z to low-z and stuff like this...

If you're familiar with ECL techology and termination schemes and considering the before mentioned problems, what termination would be best? Parallel termination (50 ohms to Vtt) for obvious reasons doesn't cut it any more...

bye,
rfmw

edit:

@flatulent yes, i'm investigating spice modeling of the ON SEMI's chips right now and simulations roughly correspond to measurements. Thanks anyway

 

[Teddy]

AC coupling?
call Onsemi?

 

[krashkealoha]

rfmw,

I do know that the low impedance is due to the input esd diodes creating a low impedance shunt path to ground.

I've had experience with some of their custom eclplus flip-flops that had a toggle frequency of 6 GHz using standard parallel termination without any problems.

Since you are experiencing the problems at high frequency and not at lower clock rates, it may either be the part's IC design or the PCB design of the board you are using, which I assume you've already concluded.

Here are the application engineers you should contact at ONSEMI. Other customers may have had the same problem and they may already know the solution to your problem.

Paul Shockman (602) 244-6360
Senad Lomigora (602) 244-7761

They are pretty knowledgeable. The phone numbers are in the U.S. and their time zone is U.S. Mountain Standard time.

Good Luck!