The Cyber-Spy.Com Usenet Archive Feeds Directly
From The Open And Publicly Available Newsgroup
This Group And Thousands Of Others Are Available
On Most IS NNTP News Servers On Port 119.
Cyber-Spy.Com Is NOT Responsible For Any Topic,
Opinions Or Content Posted To This Or Any Other
Newsgroup. This Web Archive Of The Newsgroup And
Posts Are For Informational Purposes Only.
From: "Bill Sloman"
Subject: Re: Sorting out bitscope schematics
Date: Wed, 9 Oct 2002 00:49:42 +0200
Organization: Planet Internet
NNTP-Posting-Date: 8 Oct 2002 22:49:41 GMT
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
"Ryan Gammon" wrote in message
> Hello folks,
> I've been trying to sort out the schematics of the bitscope
> (http://www.bitscope.com/design) and I have a few questions that I was
> hoping someone would be able to help me with. I've tried to answer my
> own questions as best I can, but some of these answers are probably
> pretty wrong.
> A. Power supply
> My question here relates to the inductor separating the digital ground
> from the analog ground (see the analog power circut schematic).
> I understand the role of the inductor in keeping current spikes from
> digital switching out of the analog side of things -- for a large di/dt,
> the inductor will effectively separate the two grounds with a large
> I've also seen this done a different way at:
> http://www.pjrc.com/tech/mp3/sta013.html figure 6 cs4334 chip
> ... where the inductor is placed on the 5V rail instead of on the
> ground, and a 10uF cap is used instead of the additional set of regulators
> Q1. Why don't ceramic bypass capacitors take care of this? Don't they
> "fix" noise? Don't inductors suck in terms of parasitics, as compared
> with capacitors?
> A1. The ceramic bypass cap does a good job of keeping voltages constant:
> i = C dv/dt => It's good at instantaneously supplying current (if the
> voltage sags because of, say, inductive wires and an ic switching). It
> cannot instanteously change its voltage (without an infinite current,
> So what happens when digital inputs switch from, say, a 1 to a 0? A
> bunch of current gets dumped into the digital ground line. Some of this
> comes from the brief period when both the nmos and pmos transistors of
> the driver are simultaneously on, and some comes from the discharging
> the capacitance of whatever is being driven. This current, combined with
> the natural inductance of the ground wire, creates a voltage drop along
> the ground line, called ground bounce. This ground bounce is the noise
> that we're trying to block with the inductor/rf choke.
> In a perfect world, we want capacitors to remove the resulting bounce.
> To a capacitor, ground bounce looks a lot like a voltage sag (it now
> wants to have a smaller drop across it), which makes the capacitor want
> to discharge. It can do this in one of two ways:
> -Remove charge from the top plate
> -Add charge to the bottom plate.
This is where you start to go wrong - and quite spectacularly. Kirchoff's
Laws say that the currents flowing into any node always equal the current
flowing out, and the voltage drop around any closed loop is always zero.
In this instance this means that the charge you remove from the top plate of
the capacitor has to be balanced by an equal charge flowing into the bottom
plate of the capacitor.
Going back a bit in your discussion, a capacitor can only sink or source
current if the voltage across it changes - the voltage across a 100nF
capacitor has to change one hundred times faster than the voltage across a
10uF capacitor to supply the same current.
The great virtue of series inductors is that they allow the rail voltage to
sag at the digital devices, so that most of the current flowing through the
digital chips circulates through the capacitors, rather than travelling
acorss the whole board.
It is true that the major weakness of inductors is that they are very
imperfect - wound inductors only look like inductors below their resonant
frequency, and more like capacitors above it. Capacitors have their
limitations too - 100nF multilayer ceramics have resonant peaks in the
region of 100MHz (check the manufacturers data sheets) and look inductive or
resisitve about this frequency - while electrolytic capacitors have
equivalnet series resistance, and look like resistors from a few hundred
kiloherz on up (and some look resistive at much lower frequencies).
If you need serious decoupling over a really wide band of frequencies you
end up putting different sorts of capacitors in parallel, and different
sorts of inductors in series.
When decoupling GaAs lgic from Gigabit, we used big (1810 ?) 1nF microwave
capacitors with a porcelain dielectric which we bought from American
Techinical Ceramics at a relatively high price - about $2 IIRR. These still
look capacitative at 800MHz.
When running a switching driver next to a millidegree temperature sensor, we
mounted a ferrite bead in series with each filter inductor. A ferrite bead
(actually a ferrite chip in this case ) doesn't have any winding at all, and
degenerates from looking like a small inductor to looking like a couple of
hundred ohms of resistance at a few tens of MHz, and still looks resistive
at well over 100MHz.
Hope this helps.
Bill Sloman, Nijmegen
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup