Re: Ground Bounce?

Greg Edlund ([email protected])
Mon, 04 Mar 1996 10:04:47 -0600

>Any scope experts out there?
>
>I am getting some wierd scope readings and would like some insight into what
>the scope is telling me. I have an ungrounded eqipment shelf with a scope
>probe connected to a ground pin. The local probe ground wire is connected to
>the same ground pin. The scope is isolated from AC ground and is grounded to
>the shelf's logic ground. The shelf has three solid ground planes in the
>backplane. When I plug a board into the shelf, I get lots of noise and
>oscillations showing up on the scope, reading up to +/-0.5 volts. Most
>oscillations are in the range of 20 to 100 kHz.
>
>Can anybody tell me what I am measuring here? Is this ground bounce? What
>can I do to accurately look at the ground plane while plugging in a board?
>
>
>Thanks for any help,
>
>Brad
>
>------------------------------------------------------------------------------
>Brad Mendenhall [email protected]
>BroadBand Technologies, Inc. Phone: (919)405-4682
>4024 Stirrup Creek Drive FAX: (919)544-5356
>Research Triangle Park, NC 27709-3737
>
>

Brad,

Ground bounce from high-frequency transients (such as lots of CMOS
outputs switching at 100 MHz) has a frequency content much higher than
100 kHz. It usually looks like a damped harmonic oscillation, i.e.
a sine wave with exponentially decaying amplitude. The fundamental
frequency is going to depend on the parastic Ls and Cs in your power
distribution.

You said, "when I plug a board into the shelf..." Is this a hot-
swap board? If so, you certainly get transients from this!

I'm not sure what you get when you tie the signal and ground leads
of your probe to the same point. Sometimes when you do this, the
probe acts like an antenna (induction loop) and you can pick up some
pretty interesting radiated noise. The amplitude of these signals
will depend on the proximity their source.

I believe the way you want to measure "ground bounce" is by probing
between power and ground with a good high-frequency probe. Get rid
of the long ground lead. Use a spring-type ground lead. Of course,
you're really measuring the difference between power and ground
voltages now. If you want to measure what you can simulate, i.e.
the waveform on the ground itself, I think you have to jump through
some hoops. You may have to make a test board, define a "quiet
ground plane," filter it from the rest of the system, and reference
all your probe points to this plane. I've never tried to do this
myself!

It's worth noting that you can more easily measure ground bounce
on a chip by examining one of its non-switching outputs. Measuring
it on the PC board is it bit trickier.

Also, if you get rid of the ground loop and still see 500 mV
at 100 kHz between power and ground, your low-frequency decoupling
capacitors might be crapping out on you.

Here are some good references on the topic:

"Ground-Bounce Tests Revisited," David Shear, EDN, April 15, 1993.

"EDN's Advanced CMOS Logic Ground-Bounce Tests," David Shear, EDN,
March 2, 1989.

"High-Speed Digital Design: A Handbook of Black Magic," Johnson
and Graham, PTR Prentice Hall, 1993. (Chapter 3 has a good
discussion on how to roll your own high-frequency probes.)

Enjoy,
Greg

--
Greg Edlund                o---/\/\/\---+-----+---o
Circuits and Modeling                   |     |
Chen Systems Corp.                      |      )
1414 W. Hamilton Ave.                 -----    )
Eau Claire, WI 54701                  -----    )
voice (715) 833-7067                    |      )
FAX   (715) 833-7096                    |     |
email [email protected]    o------------+-----+---o