[sdiy] Help with lm339

[Bob Weigel]

If I understand the question right, you are trying to keep it from 
jittering around when it's AT the point where the inv and non-inv inputs 
are equal roughly?  I think some others said the answer but you may have 
misunderstood.  The resistor that goes from V+ to the output is a pullup 
resistor which is needed due to the open collector output of most 
comparators.

However you also need a 'positive feedback' resistor that goes from the 
output also to the NON-INV input.  It's value will depend on the load of 
the line that is driving that input.  You want to work kirchoff's law 
for both states to set it up so that there is an offset on that point of 
enough to give yout he required hysteresis.  The word hysteresis is 
related to 'history'.  It basically means that the states invoke some 
kind of 'presence' to the matter of changing back the other way.  For 
instance in a piece of magnetic material, a hysteresis results from the 
bonding of atoms in the lattice.  A sufficient field has to be given to 
'flip flop' the state of electron spins in the lattice in order to cause 
the permeability property to change.  And this creates a 'jag' in the 
response curve of whatever that permeability is affecting..

Likewise, in a circuit like this, putting a resistance to create some 
positive feedback creates a 'jag' in the response curve also and keeps 
it from easily flip flopping back and forth due to tiny amounts of noise 
at the crossover point. -Bob

J C wrote:

>I think I am missing something easy.  I have put a rough sketch here:
>http://home.cfl.rr.com/entropy/comp.jpg
>What I have is a sustain voltage selected by the pot, buffered through
>the TL072 and then presented to the comparitor to be compared to the
>envelope being generated.  What I want to do is to add a slight
>voltage to the + input so that the comparitor will trip as the
>envelope decays and approaches the sustain level.  I do not need
>hysteresis per se because I am only looking at this when the envelope
>is decaying to the sustain level.  The sustain voltage is also being
>used by another part of the circuit and so the injected voltage can
>not change that - this circuit is to only add state LEDs to the EG. 
>Does this make any better sense?
>On 4/1/06, harrybissell <harrybissell at prodigy.net> wrote:
>  
>
>>without seeing your schematic, its tough to give exact advice...
>>but the usual solution to your problem is to add hysteresis
>>to the circuit... you need a series resistor in the positive input
>>and a large value resistor from the output to the positive input.
>>
>>Hysteresis moves the positive voltage a little bit... so that when the
>>trip point is reached the signal is suddently moved WELL into the
>>region and there is no oscillation.  Use a small amount, or you will
>>create a latch situation.
>>
>>I did an envelop gen years ago where I chose your proposed solution...
>>it will work but it is a little complex.
>>
>>H^) harry
>>
>>J C wrote:
>>    
>>
>>>I am trying to use a lm339 comparitor to determine when an ADSR signal
>>>reaches the sustain level.  What I want to do is add a small voltage
>>>to the sustain level voltage so that the comparator fires just above
>>>when the envelope reaches it (if I don't do this, it just waffles
>>>around when the two voltages are equal ie:when the envelope is
>>>sustaining).  Is there a simple way to do this (without adding a
>>>summing opamp)?
>>>      
>>>
>
>
>  
>