Copyright © 2006 Penton Media, Inc., All rights reserved.
Printing of this document is for personal use only.
Reprints

Hello to Mr. Bob Pease!
In the datasheet for the LM135/LM235/ LM335, there is no mention of capacitive bypass (minimum acceptable, maximum acceptable). This publication also did not say anything: www.national.com/appinfo/tempsensors/files/temphb2.pdf.

(I apologize for this omission. You are right. This info should have been included. However, the LM135 came out about 1975, when you were probably in the sixth grade. Are you smarter now than when you were in the sixth grade? Don’t answer that question... /rap)

If I mount the sensor at the very end of a long coaxial cable, I have 30 pF per foot, and my gut-feel is high risk of parasitic oscillation. The complex impedance at the end of a coaxial line is a Smith Chart. If I include only the last 20 feet, I have 600 pF, which is exactly the type of value that often causes instability.

(There are four main solutions to your problem. First, try it, as is, with just the 600-pF load (A). Second, try it with 0.1 µF added (B). Third, try it with 0.1 µF in series added with a 100-Ω pot and see which setting of the pot gives the best results (C). Of course, when I say “try it,” I mean don’t just see if it oscillates.

If you are feeding about 1 mA into the LM135, take a 10k resistor in series with 0.1 µF and put an additional 0.1 mA + and – of square wave into the terminal of the LM135. See how the output rings or bounces. If it damps out with a Q of 1 or 2 or less, you are probably okay. It is entirely possible that cases A or B will be just fine. But it is likely that case C will be better. Let me know which R works best.

Now take your best result and cool off the LM135 to a slightly colder temp than you expect it to see. Also, heat it a little warmer than you expect it to see. An ice cube and a solder iron may work just fine (D). /rap)

I would prefer to include 0.1-µF capacitive bypass so that the impedance is welldefined. Is it okay? Isn’t it a big deficiency to not have any discussion about acceptable capacitor bypass in the datasheet?

(I suspect not really, as the LM135 is probably safe. I haven’t gotten a question on this in the last 16 years, so it may be okay. But you are wise to inquire.

I ’m not going to add this to the LM135 datasheet, but I will post this info nearby in my BEST 1982 data book, which is my master correction file. That way, if somebody asks me in 16 years when I am 84 years old, I’ll have the data at my fingertips.

Why won’t I put this into the datasheet? Because I would have to look at all temperatures, and at all bias currents between 1 and 10 mA, and at all capacitors between 22 pF and 22 µF. I cannot possibly justify looking at all these places, whereas you can easily reassure yourself for the places you care about in less than four minutes! /rap)

No discussion of acceptable impedance values seen by the two-terminal device LM335? It should be in the datasheet so I don’t have to ask the factory.

Norm Hill

Hi Norm,
Like I said above, you are safe; 0.1 µF + 22 Ω is fine.

RAP

Sir,
There is so much talk about analog engineering and all the demand for analog engineers. Why is it such a difficult area, or is it all just hype around it?

Ganesh

Hello Ganesh,
Of course there is a lot of talk about analog engineering, and for good reason. There are many things about analog circuits, techniques, and measurements that are not taught in schools. See my recent article on all this analog stuff (Oct. 2, 2008, p. 18; www.electronicdesign.com, ED Online 19754).

Or read one of Jim Williams’ books about the art of analog design. If your school doesn’t have it, the library should buy it. The art of understanding someone else’s circuit is one problem. The art of inventing one is something else again. His first book is a bit better... And read my book on troubleshooting analog circuits. Good stuff. If your school doesn’t have it, the library should buy it.

In addition, there are more problems with thermals. Spice almost never handles thermals well. Most analog circuits can be laid out well—or badly. Thermal interactions are one thing. Matching is another. Cross-talk of capacitive strays is another. And that’s not all.

After you have read a lot, ask me some more questions. Those of us who have been designing linear circuits for more than 50 years know it isn’t just hype. It is an art. What kinds of schools teach art? World-class art?

Best wishes.

RAP

Comments invited! rap@galaxy.nsc.com
—or:
Mail stop D2597a, National semiconductor
P.O. Box 58090,
Santa Clara, CA 95052-8090

Bob Pease obtained a BSEE from MIT in 1961 and is staff scientist at National semiconductor Corp., Santa Clara, Calif.