AW: dustbins, differential amps, cables

Haible Juergen Juergen.Haible at
Mon Dec 8 13:28:55 CET 1997

Lot of questions !!!

	>OK :)  That makes perfect sense.  I am new to the world of
diode ladders
	>and OTA's.  My personal response to the CA3080 data sheet was
	>>Does this frighten you?
	>Nothing frightens me since I've seen the insides of a DAT
machine :)  

Well, when I look into my DAT (mechanical parts), I am frightened
still. (;->)
I fear we have to make peace with 3080 data sheets and the like
before we start building discrete amps.

	>So what are your three grounds again??

Earth / Enclosure GND, reference GND and "Dustbin" GND.

	[about the Minimoog and JH-4 output amp]
	>So what happens to this VCA when one of the collectors goes to
	>This is what will happen if you stick a TS plug into a TRS

It doesn't have any effect on the amp at all. That's the idea:
As it works in class A, current's are just steered to the cable
or directly to GND - it doesn't make any difference for the

	>>One word: Headroom.
	>>This isn't much of an issue for internal synth signals, but
	>>a problem as soon as you want to process external signals.
	>Now, I'm confused.  If it were designed to process hotter
internal signals
	>without clipping, wouldn't it also be able to handle hotter
	>signals??  I can forsee the opposite problem that external
signals would
	>not be hot enough, and the necessary gain would also amplify
the input

If you design your system for typical line levels (say, 1V),
short peaks in your external signal would slightly distort
a 3080 input for example, but this would be quite "musical"
and not much problem.
If your system is designed for 10V, however, your external
(line level) signal is too weak, thus you'd get noise problems.
So you have to amplify it befor it goes to the synth, and 
attenuate it again at the output. This alone will make no
disadvantage to SNR. What's amplified first is scaled
down again later. BUT if a short peak goes over your
typical level NOW (remember yout typical 1V level was raised
to 10V), your opamps will bump into the supply rails
easily. That's where the nasty stuff begins. Waveforms are
clipped the hard way, transistors in your opamp saturate
and collect charge (i.e. they will stay at the rails some time
after the input peak is gone) - all very unpleasant.

	>Isn't 10Vpp and +/-15V PS typical for synth??  Why don't they
run closer to
	>the rails??  Wouldn't using hotter signals greatly reduce these

See above. Just a few numbers to add: Most opamps guarantee
an output  swing of +/-10V for +/-15V supplies. In reality, most
of them are better (13, even 14V), but even 15V is just 6dB
more than 7.5V. Now what is the headroom of a good mixing
console, just for comparison ? - Yes, and they run line level 
internally, and also tend to have a +/-15V supply.

	>However, wouldn't it be better to balance the outputs and leave
the shield
	>unconnected at the input??

This is not as easy, and you named the reason yourself:

	>Also, you should seriously consider having all of your balanced
in and outs
	>built in such a way that they will accept unbalanced

	>>It's not RF what I want to fight in the first place, but hum
	>Will the CMRR remove the hum, and is crosstalk less than RF
	>I do not think I would want to lose the radiation immunity of a
	>system -- I think that is the primary function of a balanced

With long cables, yes. With short cables: no.
A distance of 1cm for the point of a certain GND connection
can decide on hum or no hum.


More information about the Synth-diy mailing list