-=> Roy J. Tellason said to Greg Mayman
 -=> about "BAG OF CHIPS" on 01-17-04  12:06.....

 RJT> Close.  It was an 8038!  :-)

LOL! Another typical case of my mind remembering it, but getting
it mangled up 

 RJT> The original target was the audio range,  so I'd picked values to go
 RJT> down to 20 Hz.  There's a number of trimpots in there,  and one of them
 RJT> will make this effect go away -- but it also raises the minimum output
 RJT> frequency up toward 50-60 Hz.

Hmmm... it's a long time since I looked at the 8038 data, but I
seem to remember it was all direct coupled around the waveshaping
section...
 
 GM> Where did you connect the extra filter caps? Were they before or
 GM> after the zener regulator?

 RJT> Before,  across the ones that were there.

Aha! Now, I guessing at this, but I'll say there were some bypass
caps across the zeners.

Because of those bypass caps, at higher frequencies the resistors
between the power supply and the zeners would only have to supply
the AVERAGE current to the system, plus a little that flowed
through the zeners.

But at the lower frequencies, the current drain would get higher
than the resistor current for so long that the bypass caps
couldn't hold it up.

You could avoid this effect by making the power supply resistors
pass an amount of current equal to the maximum _peak_ current
required by the load, plus say another 10% for the zeners.

But when the load current drops to its minimum, the current
through the resistor DOES NOT CHANGE so the difference has to be
taken up by the zener. This means you'd probably need a much
higher rated zener.

Just for example, let us take a case where a 15v supply is needed
at an average current drain of 100mA. If the rectifier/filter
combination is suplying 20v, we need to drop 5 volts in the
resistor. Let us set the resistor current to 110mA with 10mA in
the zener. The resistance needed is 5/0.11 = 45 ohms, the zener
is disipating 15 x 10 = 150mW.

But if the current flow in the load swings between 50mA and 150mA
-- remember the 100mA is only the average -- and it does so at
such a slow rate that the bypass caps can't smooth it out, then
we get the following:
(a) when the load current drops to 50mA, the zener current rises
    to 110-50 = 60mA and the dissipation rises to 15 x 60 = 900mW.
(b) when the load current rises to 150mA, the zener current falls
    to zero, the current through the 45 ohm resistor is 150mA so
    the voltage drop across it is 6.75v, and the output voltage
    drops to 20 - 6.75 = 13.25v, possibly even lower as the
    transformer/rectifier/filter combination may not hold up to
    the full 20v under the increased current.

The best way to overcome this is to use a three terminal
regulator instead of the emitter/zener combination.

An alternative is to use an emitter follower transistor, with the
resistor/zener combination feeding its base, and the output taken
from the emitter. The collector connects directly to the
unreegulated supply. This can accomodate changes in the load
current a lot better than the simple resistor/zener combination.

 RJT> I just don't recall offhand if the transformer I have in there gives a
 RJT> high enough output voltage that will allow for them.  And if I have to
 RJT> go with a different transformer that could get awkward.  It's a pretty
 RJT> full case... 

Usually the 3-term regs can work down to LOWER in/out voltages
than are needed for a resistor/zener regulator.

From Greg Mayman, in beautiful Adelaide, South Australia
   "Queen City of The South"    34:55 S  138:36 E

... Impropriety is the soul of wit. - Somerset Maugham
___ Blue Wave/386 v2.30

--- FLAME v2.0/b