Relay-Version: B 2.11 6/12/87; site scorn
Path: uunet!tut.cis.ohio-state.edu!zaphod.mps.ohio-state.edu!sdd.hp.com!hplabs!hpfcso!hpfcdj!hale
From: hale@hpfcdj.HP.COM (Bill Hale)
Newsgroups: rec.aviation
Subject: ANR Headset Theory (LONG)
Message-ID: <18070101@hpfcdj.HP.COM>
Date: Mon, 20 Aug 90 11:48:46 PDT
Organization: Hewlett Packard -- Fort Collins, CO
Lines: 185

As a Bose headset owner, and as one who built and used a so
called ANR headset before they became commercially available, I
became famailar with some of the design limitations and problems
that vendors like Bose and Telex have faced in order to complete
their designs.  I thought the net might be interested.
The biggies are--

1.  Creation of high enough sound pressure level.  In order to
acheive active cancellation of low frequency noise, it is
necessary to generate a very loud, low distortion, low frequency
signal.  You are competing against 300HP very nearby!

In my prototype, I used a KOSS PRO  over-the-ear headset having a
DC resistance of about 40 ohms that had a very compliant driver
measuring about 1.25" in diameter.  While 1v RMS created ear-
shattering music audio, I found that I had to drive the units
with nearly 20v peak-to-peak to get enough audio to acheive noise
cancellation, even in a relatively quiet airplane. This level is
well beyond the headset spec, and I damaged several units while
applying them in this manner.

The issue of driver power is influenced by other factors, too.
One discovers that the aircraft environment contains all sorts of
sub-audible very low frequency sound pressure components.
Examples: opening a vent, raising the gear, and adjusting the
postion of the headset.  Each of these shows up as a near-DC term
that causes the mechanical bias point of the speaker-driver to
shift around.  If there is inadequate voice coil throw, the cone,
which is already subjected to a huge AC component (at prop
frequency) will crash into the pole pieces.  The effect will be a
noisy, low frequency, buzzing sound.  An open-cockpit airplane
must be nearly an impossible environment for these headsets
because of the wind pressure components.

Since the servo feedback look is attempting to create 0 pressure
in the earcup, any leaks around the earcup will force even more
speaker cone excursion to generate a given sound *pressure*.
That is one of the main reasons the Bose gel earpieces are a good
solution (in addition to their superb high-frequency isolation).

One does not need to be rocket scientist to surmise that
considerable power is continuously required to generate these
sound pressures.  That is why the preference for A+14 to power
these units.  I am sure that the battery life in the Telex
solution will depend on the noise level in the airplane where
they are used.

Bose is currently updating older headsets to include a small vent
in the ear cavity that allegedly improves the drive capability of
their speaker-drivers by 6 db. I suspect that it is a tuned
affair, peaking up in the 100 hz range, not unlike a tunnel-
ported speaker enclosure.  I have found that their headsets
drivers generally don't crash in even very noisy airplanes and in
jet helicopters; though you can make them buzz as you adjust them
on your head.  They will also buzz while taxiing around with the
door cracked--it's so annoying that I usually turn off the noise
cancelling feature while on the ground.  The other brands? --I'd
critically listen to them for this effect when the opportunity
arises.  Compromising drive capability is a bad place to cheat.


2.  Gain Phase Stability:  As is true in any feedback system, the
feedback must not become positive at any frequency where the
overall loop gain is >1.

The little microphones (electret) that are in the ear cup
responding to pressure have very flat frequency response.  It is
much more difficult to control the gain-phase characteristics of
the speaker-phone, though.  The physical volume of the earcup is
important, as when it is very large, the time required for the
signal to travel from the speaker to the mic will introduce
further phase shift at higher frequencies, which eats into
available phase margin.  As a transport lag, this delay is not
compensatable.

Worse, when the headset is removed, the cavity opens and the
transfer function from speaker to mic dramatically changes
because the *pressure* goes away.  (The driver also crashes,
trying to make *pressure* against an open load) Some intermediate
cases are also troublesome:  when removing the headsets, as they
are moved from the ears, reflections can occur that drive the
phase characteristics nuts.

Contrary to the exotic, hi-tech idea the vendors are trying to
sell, compensation of the feedback path in these headsets is very
simple.  It's done like an op-amp: reduce the gain at 6 db/octave
until reaching unity gain.  The highest frequency where unity
gain is allowable depends on how well the phase of the overall
feedback path is controlled.  The Bose headsets acheive unity
unity gain at ~1.5KHz.

The crossover frequency (loop gain = 1) is very important,
because it uniquely determines how much noise cancellation is
available at lower frequencies.  6 db/octave = 20 db/decade => 20
db noise rejection at 150 Hz.  Prop frequencies are about 2400
rpm /60 s/min * 2 blades/rev = 80 hz.

So the trick is to figure out how to get phase control at high
frequencies so that higher gain-bandwidth can be acheived with
correspondingly increased LF rejection.  The headset must be
never oscillate, or it might deafen the user--good only for
lawyers!  If it is marginally stable, it will ring near the
crossever frequency (~1.5khz) every time some audio comes by.
That will give it an annoying, hollow sound.

What the manufacturers have done: A. Bose has installed a foam
plug inside the ear cup to reduce the destabilizing effects from
taking the headset off and to generally isolate the
earcup/feedback path from the outside world at frequencies near
crossover (one of their patents?).  They warn that the unit might
oscillate if the foam plug is removed; B. Telex has some sort of
detector which shuts the unit down if oscillation occurs--plus
they settled for slightly lower loop gain judging by their less
agressive LF rejection specs.  I can hear their unit ringing just
a bit near crossover. C. since the temperature coefficient of the
electret mic output is gross, Bose has compensated it with a
thermistor mounted directly on the element--necessary because the
loop gain isn't very high to start, and to have to waste
additional for margin over temperature would be painful.

My opinion is that more than 20 db ANR at prop frequencies
probably isn't worthwhile, as bone conduction and other
mechanisms would probably become predominat.  The ear tends to
servo its gain back up and the noise is still very much there.
(Just the pain is eliminated)


Some operational observations in no order:

1.  The fatigue on long trips is really reduced.  Moreover, my
wife--who is tolerant of flying--no longer complains about 4 hour
legs.  I teach for the ABS clinics in Bonanzas, which involves a
long day (7+ hours of dual given).  If I use the Bose headsets, I
am in good shape.  If not, = seriously beat at the end of the
day.

2.  The foam ear cushions exacerbate sweating, but aren't as bad
as I expected.

3.  The mechanical configuration of the Bose leaves something to
be desired: A. mic is loose and floppy; B. ear cushions tend to
pop out of holders; C. plug is weak and breaks easily; D.
headband is too thick, and tall person hits head on ceiling.

4.  I like the mics. Good noise cancelling, very close talking.
Cabin noise resolves as white noise when squelch open.

5.  Service from Bose has been very outstanding--fast and free.

6.  Clamp pressure is nominal; much lower than Clarks, which make
me cry when I put them on again in the middle of the 7 hour day.
Passengers not used to headsets don't mind them.

7.  Stereo into these is really nice for long trips.  The A/C
noise level is low enough that the result isn't just bedlam like
one might expect.

8.  Cable management is a nightmare. Let the accident report show
that the crew died of apparent strangulation!  Telex has the only good
solution.  The new mount it in A/C interfaces are great for
owners, useless for others.  I built a 2 hole intercom with Bose
connectors and NICADS for instructional use.  Bose is now
offering such a unit.

9.  Ability to carry radio volumes much lower than normal greatly
reduces bedlam.

10.  Sample of 10 units among friends, etc indicates fair reliability,
but not outstanding

11.  Mfgrs spend a lot to make the units failsafe if something in
the ANR quits.  Would prefer to have the headsets for less $
without this feature, as there are all sorts of other backups.

I looked at the other headsets at OSH, thinking that Bose could
use some competition to moderate prices.  I thought the Telex was
interesting, especially if they could be supplies with gel
earpieces.  Bose thinks the competition is weak, as they plan to
RAISE (!!) their prices soon.  They sure aren't deficient on
corporate hubris!  Opinion:  They're outrageous, and probably
worth it.

Bill Hale  hale@hpfclm.hp.com               
Thru ether K0QA; In ether BE33 N33HP CFIAIME A&P; Twisted pair 303-229-3037
Snail: c/o HP GTD, 3404 E Harmony, Ft Collins, CO 80525