Hello Mr. Pease: I've read somewhere that it can be a
problem for the op amp to lose one of the supplies or
use the wrong startup sequence. (You are correct. A
startup sequence can cause great trouble on the positive and negative rails. /rap) Is it always a particular rail? Can you
shed some light on the cause of this? Does it apply for bipolar
and CMOS families?
• David Smith
• Pease: You have proposed a tough problem, and I don't
think there are any simple answers because there are only
a couple thousand kinds of op amps that people use and a
few dozen configurations for each. What a mess! Let's see if
I can provide a general solution. (A) Please try to avoid nailing the positive input of an op amp
to ground, even if it should be "grounded." Bob Widlar has
argued that even 1k is a good idea to prevent gross input
currents from flowing.
(B) Here is a worst case. Set up a nice, healthy analog system with nice, healthy power supplies—say, ±15 V at 1 A.
Connect an op amp as load with its positive input grounded
and with 30 Ω of load from the +15-V bus to the –15-V bus.
Now to get in trouble, connect the ground and the positive
rail. Refuse to connect the –15-V supply to the –15-V load.
The 30-Ω load would try to pull the negative power-supply
bus toward +13 V and feed 480 mA through the 30-Ω resistor to pull the op amp's negative rail above its positive
input—not a good idea. Many op amps will blow up quickly if
you try this. LF356 biFETs and LM324s will probably blow or
be badly damaged. Many bipolar and CMOS op amps, too.
(C) A partial solution is to take the load (the operating circuit
and system) and connect anti-reversal diodes across the
power-supply busses of the load (the system) so the negative supply can go barely 0.8 V above the ground bus and
the positive supply can go barely 0.8 V below the ground.
1N4004s would be good in most cases. I don't think you will
need 1N5819s.
(D) Avoid connecting a wire short from the input of the op
amp to ground. If you put at least 1k there, it will not add
much noise, but it will limit the currents and prevent
much harm.
Robert: Thanks so much for speaking at the Audio Engineering
Society conference in San Francisco in October. I was wondering
how PPS film capacitors such as Panasonic's ECH-U series compare with the rest in your testing. (Polypropylenes have had superior soakage (dielectric absorption) and superior leakage (Tau =
several years) /rap) I'm in an application where I have to use SMT,
and I've heard that PPS does even better than NP0 for audio signal paths. (If you can use a big enough poly, it can sound better. If you use NP0 and you chintz because it's not big enough in C value, you can't blame it on the inherent quality of the dielectric.
Many people use poly and other films where SMT is "required"
because they are superior and worth it in terms of performance.
Solder them in by hand. Have fun! /rap)
• Tyler Gleghorn
• Pease: Why are the PPS capacitors so good? Because they
are polypropylene, which is a very superior dielectric. Some
people think it's an octave worse than polystyrene. I think it
is an octave better, for all the applications I have seen.
There are many manufacturers of polypropylene and very
few now of polystyrene. Polypropylene is comparable to NP0
for audio work, but you can get 1 µF or more in a reasonable
package, whereas NP0/C0G is only available up to 0.1 µF.
However, many film capacitors such as poly aren't very
available in surface-mount (SMT) format. NP0/C0G is much
better than poly for tempco, though for audio circuits, who's
going to complain about –120 ppm/°C versus (0 ±30)? If
you think you can hear any difference, I suggest you try
some A-B-X testing. I don't think anybody can hear any difference in an audio circuit.
Bob: I finally got to read your Nov. 6, 2006 article ("Bob's Mailbox,"). I thought John Cook's comment on the two-chassis
RCA studio amplifier was interesting. I'm sure that separating it
into two chassis, one with the power supply, reduced hum due to
magnetic coupling from the power transformers. (Not to mention
the transients when the rectifiers carry big surges of current for
short times. /rap) Years ago, my company made audio mixers for
the television broadcast industry. We had mu metal shields on
our microphone input transformers and a steel shield around our
power transformer. The noise was great, until you brought up several channels. (Ha! /rap) The fix, developed by one of my partners, was to reverse both primary and secondary windings on
every other channel. Phase was maintained through the transformers, since both were reversed. But as more channels were
brought up, the magnetically coupled hum went down instead of
up. (And this was the input transformers (not the power transformers)? What a brilliant solution! I don't expect things like that
to cancel out. /rap) I always thought it was a clever cure!
• Harold Hallikainen
• Pease: Yeah, fiendishly... Thanks for the comments.
Comments invited! rap@galaxy.nsc.com —or:
Mail Stop D2597A, National Semiconductor
P.O. Box 58090, Santa Clara, CA 95052-8090