Mail*Link(r) SMTP FWD>DES: bulk decoupling caps Jerry, I hope the following helps >Date: 8/2/96 11:22 AM >From: Jerry Cupples >Dear TN subscribers, >In looking at some new designs for high speed comm boards, we are seeing >competitors use 22 uF aluminum electrolytic caps in radial leaded packages >for decoupling. >Most of our designs have used solid tantalum parts in "D" case the past few >years. >Part of the reason for this may be cost, as it seems the Ta parts are about >40 cents vs. 10 cents for the AL parts. >Of course, it is much simpler to automatically place the SMT devices, but >what I'm wondering is... >Is there some electronic design superiority related to the package or >typical circuit "physics" (i.e. trace inductance, etc.) to choose the AL >radial leaded part? In previous life, there were some major electrical considerations is selecting Ta electrolytic capacitors over Al. One was tolerance, most Al capacitors had tolerances of -20 or -50% to +100 or more %. Due to the way of manufacture, Al capacitors had high inductances and therefore were less effective at high frequencies --- meaning the 10's or 100's of kHz not even MHz's. The "equivalent series resistance" as a function of frequency (which is a measure of the alternating current resistance) was like "lousy" for the Al capacitors. In addition the maximum storage/use temperature was about 20 degrees C less than the Ta's. Other than that, no problem. (;-) Things have somewhat improved over the years, and for most applications there is not much major differences between Ta and Al electrolytic capacitors for "bulk" brute force filter capacitors. As was mentioned in a previous posting on "technet" about bypass capacitors (I can dig it out and send it to you if you want/need) the key to bypass capacitors is to use "good" (relatively low valued about 1-100 nF) RF ceramic capacitors VERY close to the power pins of the IC. Give preference to keeping the grounded lead/connection of both the component and bypass capacitor as short as possible (except for ECL logic families). The purpose of the ceramic bypass capacitor is to take care of the very high frequency spiking currents inherent with the fast pulse rise/fall times and prevent the rounding of the edge or corners of the signal pulse/clock. The purpose of the "bulk" Ta or Al bypass capacitors is to stabilize the voltage of the power distribution system for the longer "flat or table" portion of the signal pulses. Therefore, the goal is to have the Ta or Al bypass capacitor work like a great big storage tank of energy (voltage) to maintain a constant voltage on the power distribution system. In theory, the "bulk" capacitor should not see any really high frequency stuff (it does see the clock frequency) if the power and ground system "is a good one". Jerry, the following sketch is a gross illustration of the verbal description of the "wave forms" for rises/fall times and "flat" portion of a pulse. In particular square and sometimes pulsed waveforms are kind of unique because if you know the basics on how to read them they can idicate what the problem is. xxxxxxx xxx xx x x x x x xx x x x x x xxxx x x x x x x x x x x x x xxxxxx xxxxxxxxxxx xxxxxxxxxxxxx xxxxxx The first is a This second This third pulse shows square pulse, pulse has the "flat" portion has just like what rounded corners "drooped", indicating you want. that indicate the supply system doesn't the bypass cap. have sufficient capacity isn't doing the Ta or Al to sustain the job. "constant voltage" to the load. >Any guess, speculation, or real intelligence would be appreciated! This let's me out--- >regards, >Jerry Cupples >Interphase Corporation >Dallas, TX USA Hope this helps, Ralph Hersey [log in to unmask] *************************************************************************** * TechNet mail list is provided as a service by IPC using SmartList v3.05 * *************************************************************************** * To unsubscribe from this list at any time, send a message to: * * [log in to unmask] with <subject: unsubscribe> and no text. * *************************************************************************** ------------------ RFC822 Header Follows ------------------ Received: by quickmail.llnl.gov with SMTP;2 Aug 1996 11:05:15 -0700 Received: from ipc.org by simon.ipc.org via SMTP (940816.SGI.8.6.9/940406.SGI) id MAA22288; Fri, 2 Aug 1996 12:53:14 -0700 Resent-Date: Fri, 2 Aug 1996 12:53:14 -0700 Received: by ipc.org (Smail3.1.28.1 #2) id m0umNJh-0000QqC; Fri, 2 Aug 96 11:43 CDT Resent-Sender: [log in to unmask] Old-Return-Path: <[log in to unmask]> X-Sender: [log in to unmask] Message-Id: <v02130501ae27e022afb2@[157.175.110.24]> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 2 Aug 1996 11:50:33 -0500 To: [log in to unmask] From: [log in to unmask] (Jerry Cupples) Subject: DES: bulk decoupling caps Cc: [log in to unmask] Resent-Message-ID: <"a-65K.0.G8K.I0Z0o"@ipc> Resent-From: [log in to unmask] X-Mailing-List: <[log in to unmask]> archive/latest/5539 X-Loop: [log in to unmask] Precedence: list Resent-Sender: [log in to unmask] *************************************************************************** * TechNet mail list is provided as a service by IPC using SmartList v3.05 * *************************************************************************** * To unsubscribe from this list at any time, send a message to: * * [log in to unmask] with <subject: unsubscribe> and no text. * ***************************************************************************