MOTM

The reason a mixer is taking so long is the quest for the perfect op-amp.
This is an op amp
this is suitable for both audio mixing AND DC (control voltages). The DC is
the issue.

When you design a DC op amp (like the MXL1013s used as DC mixers in all the
modules) you
trade off speed (called 'Slew Rate' on the data sheet) with DC drift. One of
the main contributing
factors to past modular drift (Moog being the worst, just because they were
first) are the mixers. They
were OK for audio, but *horrible* for DC.

You would think with today's technology a good audio op amp with low drift
is available. NOT SO! In fact,
the trend is the opposite: there are insanely DC accurate op amps that have
a max frequency of 15HZ
(yep, 15 HERTZ). Then you have insane audio parts with 0.0001% THD and 95dB
SNR that drift like a mother
(because they assume AC coupling of the output).

So, I have toyed with at least 4 different ways to go, including a 100%
discrete cascode JFET design. But
what I'll have to do is just 'bite the bullet' and pick the best compromise
part.

Paul S.

> -----Original Message-----
> From: Paul Schreiber [mailto:synth1@...]
> the trend is the opposite: there are insanely DC accurate op amps
> that have
> a max frequency of 15HZ
> (yep, 15 HERTZ). Then you have insane audio parts with 0.0001%
> THD and 95dB
> SNR that drift like a mother
> (because they assume AC coupling of the output).
>

This is probably wildly impractical, but would it be possible to split out
the dc component of the signal, amplify the ac component with the killer
audio opamp, kill off any induced dc component at the audio opamp's output,
then add the original dc component back in?

Dave Bradley
Principal Software Engineer
Engineering Animation, Inc.
daveb@...

Paul S wrote:
So, I have toyed with at least 4 different ways to go, including a 100%
discrete cascode JFET design. But what I'll have to do is just 'bite the
bullet' and pick the best compromise part.
----------
Well, since the MOTM module design already centers around AC mixers on the
inputs, why not make the killer DC mixer for now and the killer AC mixer
later?  Would it be such a shame to have two?

Just another thought to add to the "different ways to go"  list.

Larry H

Nah, gimme one that does both.  That would, if I understand correctly, be
able to handle DC-offsets to AC signals without cause for alarm.  Or, would
you be able to handle DC offset with a normal AC mixer...?  At the very
least, you could send a sharp spike (like the output of a fast-attack EG)
into a DC offset signal.  I think...?  (Brain hurts, time to get off
work...)  

--PBr

> -----Original Message-----
> From:	J. Larry Hendry [SMTP:jlarryh@...]
> Sent:	Tuesday, July 18, 2000 5:23 PM
> To:	motm@egroups.com
> Subject:	Re: [motm] Mixer thoughts
> 
> Well, since the MOTM module design already centers around AC mixers on the
> inputs, why not make the killer DC mixer for now and the killer AC mixer
> later?  Would it be such a shame to have two?
> 
> Just another thought to add to the "different ways to go"  list.
>

how about an ac mixer with 1 seperate dc offset channel?  that should do for
almost
any application, because most modules already have a dc input..allowing you
to mix dc offsets by daisychaining...

-paulw

-----Original Message-----
From: Brousseau, Paul E (Paul) [mailto:noise@...]
Sent: Tuesday, July 18, 2000 5:31 PM
To: 'motm@egroups.com'
Subject: RE: [motm] Mixer thoughts


Nah, gimme one that does both.  That would, if I understand correctly, be
able to handle DC-offsets to AC signals without cause for alarm.  Or, would
you be able to handle DC offset with a normal AC mixer...?  At the very
least, you could send a sharp spike (like the output of a fast-attack EG)
into a DC offset signal.  I think...?  (Brain hurts, time to get off
work...)

--PBr

> -----Original Message-----
> From:	J. Larry Hendry [SMTP:jlarryh@...]
> Sent:	Tuesday, July 18, 2000 5:23 PM
> To:	motm@egroups.com
> Subject:	Re: [motm] Mixer thoughts
>
> Well, since the MOTM module design already centers around AC mixers on the
> inputs, why not make the killer DC mixer for now and the killer AC mixer
> later?  Would it be such a shame to have two?
>
> Just another thought to add to the "different ways to go"  list.
>

------------------------------------------------------------------------
Life's too short to send boring email. Let SuperSig come to the rescue.
http://click.egroups.com/1/6809/6/_/529958/_/963966612/
------------------------------------------------------------------------

>how about an ac mixer with 1 seperate dc offset channel?  that
>should do for almost any application, because most modules already
>have a dc input..allowing you to mix dc offsets by daisychaining...

People need to think about the possible applications of mixers
before saying they don't need both AC and DC in the same mixer.
Say you want to mix an EG's output with an LFO or OSC before sending
it to a filter or another oscillator.  Can't do that unless your
mixer does both.  I don't want just a DC offset, I want a mixer
that handles both AC and DC inputs and outputs the combined sum.

-Cary

In a message dated 7/18/2000 6:21:05 PM, cary.roberts@... writes:

>>how about an ac mixer with 1 seperate dc offset channel?  that
>>should do for almost any application, because most modules already
>>have a dc input..allowing you to mix dc offsets by daisychaining...
>
>People need to think about the possible applications of mixers
>before saying they don't need both AC and DC in the same mixer.
>Say you want to mix an EG's output with an LFO or OSC before sending
>it to a filter or another oscillator.  Can't do that unless your
>mixer does both.  I don't want just a DC offset, I want a mixer
>that handles both AC and DC inputs and outputs the combined sum.

You have a great point Cary! But I wonder if a compromise could be made such 
that a mixer designed for DC (or more precisely CV and modulation) would 
handle audio with some minor distortions (since I'd think a slow slew would 
more round an audio square or saw wave, and I typically use sines or 
triangles for audio modulation due to reduced sidebands). This would be 
reasonable tradeoff. An audio (AC) mixer should be maximized for audio 
obviously.

JB

cary.roberts@... writes:

>>People need to think about the possible applications of mixers
 before saying they don't need both AC and DC in the same mixer.
 Say you want to mix an EG's output with an LFO or OSC before sending
 it to a filter or another oscillator.  Can't do that unless your
 mixer does both.<<

Now you've got me wondering what is happening on MOTM modules that provide 
multiple CV inputs for modulation such as the Oscillator.  You can go into 
them at with slow EGs/LFOs and at audio rates simultaneously.

-Elhardt

In a message dated 00-07-18 11:21:37 EDT, you write:

<< what I'll have to do is just 'bite the bullet' and pick the best compromise
 part. >>


paul,
any way to use the same board layout but offer 2 versions of the mixer, one 
dc and one audio ? that way you can focus on optimizing each for its task.  
may be too much of a hassle ( and this i would certainly understand ) but 
given the motm`ers quest for module nirvana, the folks on the list might 
prefer that option - ?
a tricky problem for such a "simple" module, indeed !  good luck !
best,
dave v.

Or an AC/DC switch as on the router?

 -----Original Message-----
From: 	J. Larry Hendry [mailto:jlarryh@...] 
Sent:	Tuesday, 18 July, 2000 8:23 PM
To:	motm@egroups.com
Subject:	Re: [motm] Mixer thoughts


Well, since the MOTM module design already centers around AC mixers on the
inputs, why not make the killer DC mixer for now and the killer AC mixer
later?  Would it be such a shame to have two?

>
> Now you've got me wondering what is happening on MOTM modules that provide
> multiple CV inputs for modulation such as the Oscillator.  You can go into
> them at with slow EGs/LFOs and at audio rates simultaneously.
>

The VCO input mixer is a "slow" LT1013. We really want low drift here! The
FM-ing using audio
is inherently band-limited because we are using mainly sine waves for the
modulation (gradual
slopes) and things like THD are not important. Also, we aren't wailing at
10V pk-pk, especially
for linear FM.

Paul S.

My two cents:

Make it switchable.

DC (and "precision audio") path is around a low offset voltage opamp
that's fairly good for audio range as well (i.e. an LF411, but no OP-07).

Plus a discrete, no multistage feedback, "vintage" mixing amp.

You need a 2p2t switch to select between both options. One half of the
switch chooses the output, and the other half handles the virual summung
node to prevent the vintage path from compromising the precision path.

Precision path might be ordinary virtual GND summing node.

A similar thing for the vintage path might be the emitter of a transistor
with its base grounded.

From each input pot there would be two summing resistors, one to the
precision summing node, one (AC-coupled) to the almost-virtual-GND
emitter summing node. When the output is switched to select the precision
path, the emitter of the vintage path would be connected to GND to
avoid any talkback over the finite emitter impedance.

Hey, think of it: There are many compromise designs that try to cover
both worlds, and there are some dedicated designs which are good for
one but bad for the other application - so why not make the module
switchable. This would also allow the user to demonstrate the difference
in sound with the flick of one switch ...

JH.




-----Urspr\ufffdngliche Nachricht-----
Von: J. Larry Hendry <jlarryh@...>
An: <motm@egroups.com>
Gesendet: Mittwoch, 19. Juli 2000 02:23
Betreff: Re: [motm] Mixer thoughts

> Paul S wrote:
> So, I have toyed with at least 4 different ways to go, including a 100%
> discrete cascode JFET design. But what I'll have to do is just 'bite the
> bullet' and pick the best compromise part.
> ----------
> Well, since the MOTM module design already centers around AC mixers on the
> inputs, why not make the killer DC mixer for now and the killer AC mixer
> later?  Would it be such a shame to have two?
>
> Just another thought to add to the "different ways to go"  list.
>
> Larry H
>
>
>
> ------------------------------------------------------------------------
> Make new friends, find the old at Classmates.com:
> http://click.egroups.com/1/7075/6/_/529958/_/963966240/
> ------------------------------------------------------------------------
>
>
>
>