Signals & Noise: June 6, 1996

I read with amazement "Prototyping with surface-mount technology" by Jack Ganssle (EDN, March 28, 1996, pg 197). My experience is different from Ganssle's in a number of areas.

It's true that a pc board is the most reliable way to hook up and test a high-speed digital system. PC-board-layout tools and prototype boards are in- expensive—a few hundred dollars buys a fairly complex pc board. Design and simulation tools under Windows are pretty cheap. Therefore, you should be able to make a workable design with a reasonable number of yellow wires. And, if you can't buy the parts at Radio Shack, you can probably buy them through Digi-Key and Mouser.

A few years ago, when my company entered (also kicking and screaming) the surface-mount-technology (SMT) arena, it spent a few thousand dollars on SMT assembly and rework equipment. The only piece of equipment that the designers use, however, is a $500 hot-air gun with a heat shield and a suction cup for removing defective parts.

For soldering, designers use the Edsyn 930 Loner soldering irons with a fine tip. We don't use "lots of heat"; we use the right amount of heat. We have built many SMT pc-board prototypes this way, and they have all worked. The main problems occur when a new person does the soldering. To combat this, we have a designer knowledgeable in SMT teach the new designer how to put the parts together quickly and accurately. All you need are a good soldering iron; a correct-tip, correct-temperature, small-diameter solder; plenty of flux; a good eye; a steady hand; and plenty of patience.

Gannsle's "soldering guru" in Columbia, MD, sounds like a wannabe. You can easily hand assemble SMT components onto commercial-quality, prototype pc boards with good tools and have the systems work reliably. SMT should not be preventing any designer from building the design of his/her dreams: It should be helping.

Lyle Johnson
Vice President
Product Engineering
Modular Mining Systems
Tucson, AZ

The equation in "Circuit monitors multiple contact closures" (EDN, Feb 15, 1996, pg 116) should read:

FORMULA

Add StereoGraphics to the list of vendors for 3-D graphics products in "PC graphics struggle to incorporate 3-D" (EDN, March 14, 1996, pg 61). The company makes a shutter-glass product, SimulEyes, which works as a stand-alone product supported by numerous software packages.

StereoGraphics
San Rafael, CA
(415) 459-4500
www.sterographics.com

The diode bridge in Figure 1 of "LAN power supply generates isolated 9V" (EDN, April 11, 1996, pg 104) is a shorting bridge.

The FDC37C93XFR from Standard Microsystems, featured in "IRDA-protocol IR links make 35-fold leap in data-transfer speed" (EDN, April 11, 1996, pg 63), comes in a 160-pin package. "Math software adds features" (EDN, May 9, 1996, pg 30) lists an incorrect price for Mathematica 3.0. The software costs $1295. The electrical-engineering application pack costs $195.

Resistor trick to drive capacitive loads

The circuit proposed by Barry Harvey and Chris Siu in Figure 8 of "Simple techniques help high-frequency op amps drive reactive loads" (EDN, April 25, 1996, pg 133) falls short by one resistor. If you split the R_SERIES into two and if you drive the feedback register from the junction of the two register, the signal bandwidth at the load improves by a factor of two or three, and the group delay is leveled.

For example, driving a 470-pF load from the AD846 op amp might require a 100 Ohm series resistor. If the feedback resistor is 1000 Ohm, the resulting bandwidth is about 3.5 MHz. Splitting the 100 Ohm resistor into two 50 Ohm resistors and connecting the feedback to their common point increases the bandwidth to more than 9 MHz and gives a constant group delay. The only disadvantage is that the amplifier output peaks by about 7 dB at frequencies near the -3-dB point. This situation limits the maximum output swing.

Tom Napier
Principal Engineer
Aydin Corp
Horsham, PA