Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!zaphod.mps.ohio-state.edu!swrinde!elroy.jpl.nasa.gov!lll-winken!uwm.edu!csd4.csd.uwm.edu!info-high-audio-request
From: henry@ginger.sri.com (Henry Pasternack)
Newsgroups: rec.audio.high-end
Subject: Re: More comments on the Chater amp.
Message-ID: <8757@uwm.edu>
Date: 8 Jan 91 14:51:01 GMT
Sender: news@uwm.edu
Lines: 97
Approved: tjk@csd4.csd.uwm.edu
Originator: tjk@csd4.csd.uwm.edu


   Jeff Chan says:

>How is a monolithic different from a plain ceramic?  I always thought 
>they were one in the same.  Are you going to use a certain grade, like
>NPO?

   I don't know the difference, except that they have incredibly high
capacitance density for a non-electrolytic cap.  A 1uF 50V monolithic is
much smaller than comparable standard ceramic disks.  I have also heard
that the monolithics are very low impedance, as they are used primarily
for digital bypass applications.  I don't know about the audio quality
of these components.  I have heard it aid that ceramic caps should be
avoided at *all* places in audio circuits, even non-signal applications.
I'm not that concerned about it.  I'm using monolithics for all of the
1uF and 0.1uF locations.  Some of these are bypass capacitors and some
are integrators for the servo control loops.  In these applications,
temperature drift shouldn't be a problem.  Could it be that NPO caps
are more linear because they are unaffected by signal-induced micro
temperature variations?  Don't ask me.

>As a related question to all, is a film cap a reasonable substitute for 
>a ceramic for RF bypassing?

   You'll have a hard time finding film caps small enough to fit the
circuit board.  Incidentally, as I was soldering on my Chater boards
last week, a thought occurred to me.  Some people insist on using
OFHC wire even for short jumper runs on circuit boards.  But who knows
what the quality of the copper on the board is?

   Personally, I don't believe in most of the OFHC stuff, at least for
short runs.  Speaker cables, maybe.

>Using NTE458s was a mod he approved of in the same letter response as
>above, and he did indicate that it would mean different biasing.

   I should mention that the slew rate of the amplifier is affected by
the bias current and transconductance of the input stage.  It is probably
no big deal to set up the NTE458's correctly, but, again, I don't believe
it's worth the effort

>My only thoughts so far are to delete as many beads as is reasonable.
>The author did not indicate it, but when he found that a true source
>of some damaged output devices was rf noise from the AC power switch,
>which was cured through wire routing and the snubber cap, perhaps he 
>would have been more amenable to getting rid of the beads on the 
>output devices.  He really likes the beads as a way of killing RF 
>oscillation, though.

   I don't fully understand the rationale of using the beads.  I think
they are most effective when placed on component leads to kill stray
capacitive coupling over distances of a half inch or less.  I guess
that in certain stages, the addition of a little inductance will roll
off the stage response and minimize the likelihood of local oscillation,
but only where a parasitic feedback path exists.  In some locations, a
bead seems like a bad idea.  For instance, the output of an emitter
follower already looks inductive.  Putting on a bead might make matters
worse by increasing the inductance at that point.  I guess it's a matter
of controlled roll-off at very high frequencies.  If the oscillation
occurs at a frequency much higher than the open-loop bandwidth of the
amplifier, the bead is probably a good idea, provided it isn't isn't
excessively inductive.  There are also different grades of beads.  I
used a Ferroxcube "4A" material, but there is also a "3A" and "3B"
bead in our lab stock.  I'm not sure what's going to happen.

   I still don't understand why it is permissible to put a bead on each
lead of the output devices, but not on the speaker leads.

>I can't find a reference to it either except as you note in the schematic.
>Surprisingly, this does not appear to have come out in any letters.
>Perhaps people are leaving it out and still not getting oscillation...

   The only reason the choke is there is to keep the amplifier from
shorting out if it does go into oscillation.  Most designs parallel the
choke with a low-value resistor to damp out its ringing tendency.  There
is probably room for experimentation.  I don't have the schematic here
in front of me, but if you're set up to do phase margin measurements,
it's probably a good idea to spend time varying the values of the
compensation and feedback components to get best overshoot performance.
On the other hand, the amplifier should work fine if built as designed.

>A question from me about the power supply: R1, the "3mA current
>diode."  What kind of device is this?  Any ideas about a part
>or manufacturer?

   I haven't looked carefully at the power supply.  National makes an
LM334 which is a programmable current source requiring an external
resistor to set the current.  I haven't seen fixed current diodes, though.

>I definitely pays to check the letters for updates and bug fixes,
>even if the puzzle is still incomplete for hose of us trying to
>build from bare boards.

   I think I ought to call Old Colony and see if they can send me an
update sheet.

-Henry